JP2011131643A - Electric power steering device - Google Patents

Abstract

When performing sensorless control, a decrease in steering feeling is suppressed.
A dead zone processing unit 112 receives a calculation induced voltage e ′ calculated by an induced voltage calculation unit 111, and sets a magnitude of a steering torque Tr according to a dead zone designation signal S output from a dead zone changing unit 113. When the torque Tr0 is larger than the torque Tr0, the dead zone processing of the calculation induced voltage e ′ is performed with reference to the dead zone processing map. When the magnitude of the steering torque Tr is equal to or less than the set torque Tr0, the dead zone of the calculation induced voltage e ′. Do not process. Accordingly, the steering handle does not vibrate during the steering operation. Further, when the steering wheel is cut out, the estimated electrical angle is not fixed, so that the steering operation is not caught.
[Selection] Figure 3

Description

  The present invention relates to an electric power steering apparatus that drives and controls an electric motor based on a steering operation of a driver to generate a steering assist torque.

  Conventionally, an electric power steering device detects a steering torque applied to a steering wheel by a driver, calculates a target steering assist torque according to the detected steering torque, and obtains the target steering assist torque. Controls the amount of power applied to the motor. An electric power steering apparatus using a brushless DC motor as an electric motor has also been generalized. The brushless DC motor is energized to the U phase, the V phase, and the W phase by inverter switching control. Therefore, when using a brushless DC motor, a rotation angle sensor for detecting the electrical angle of the rotor is provided to control the phase.

  When the rotation angle sensor fails, the motor cannot be controlled. Therefore, when the rotation angle sensor fails, the electrical angle is estimated based on the induced voltage (back electromotive force) generated in the motor, and the motor is driven and controlled using this estimated electrical angle (estimated electrical angle). Technology is known. In general, control of a brushless DC motor using an estimated electrical angle is called sensorless control.

  In sensorless control, the motor angular velocity is calculated from the induced voltage by utilizing the proportional relationship between the induced voltage generated by the motor and the motor angular velocity. Then, from the calculation cycle of the sensorless control and the motor angular velocity, the angle at which the motor rotates per cycle is obtained, and the current electrical angle, that is, by adding this rotational angle to the electrical angle one cycle before in the rotation direction, that is, Calculate the estimated electrical angle. An electric power steering apparatus using such sensorless control is proposed in Patent Document 1, for example.

JP 2008-87756 A

  When sensorless control is performed, an induced voltage generated in the motor is calculated. The calculated value includes measurement errors such as a current value and a voltage value necessary for calculating the induced voltage. For this reason, in the state where the steering wheel is being held, even if the rotation of the motor is stopped, the electrical angle is added due to the calculation error of the induced voltage, and the torque fluctuation of the motor occurs. This torque fluctuation appears as vibration on the steering wheel. In order to prevent this, a dead zone is provided in the calculated value of the induced voltage, so that the estimated electrical angle can be prevented from changing with respect to the error of the induced voltage. However, when a dead zone is provided in the calculated value of the induced voltage, the estimated electrical angle does not advance when the driver starts to turn the steering wheel, which gives a feeling of being caught in the steering operation.

  Further, in the electric power steering apparatus proposed in Patent Document 1, when the motor angular velocity calculated from the induced voltage is within the dead zone near zero, a positive offset is used to obtain electrical angle information. The offset value of the negative value and the negative value are alternately added to the angular velocity at a predetermined cycle to prevent the motor angular velocity from becoming zero. However, in this apparatus, since the offset value (fixed value) is simply switched between positive and negative and added to the estimated angular velocity, the motor always rotates forward and backward during steering. Therefore, the vibration of the motor is transmitted to the steering wheel, and the steering feeling is bad.

  An object of the present invention is to cope with the above problem, and is to suppress a decrease in steering feeling when performing sensorless control.

In order to achieve the above object, the present invention is characterized in that a permanent magnet synchronous motor that is provided in a steering mechanism and generates steering assist torque, a steering torque sensor that detects steering torque input from a steering handle to a steering shaft, and Motor control value calculating means for calculating a motor control value so that the steering assist torque increases as the steering torque detected by the steering torque sensor increases, and rotation for detecting the electrical angle of the permanent magnet synchronous motor An angle sensor, a sensor abnormality detecting means for detecting abnormality of the rotation angle sensor, and an induced voltage generated in the permanent magnet synchronous motor when the abnormality of the rotation angle sensor is detected by the sensor abnormality detection means The permanent magnet synchronous motor is estimated based on the calculated induced voltage, which is the calculated value. According to the electric angle estimating means for calculating the electric angle and the motor control value calculated by the motor control value calculating means, when the abnormality of the rotation angle sensor is not detected, the electric angle detected by the rotation angle sensor is calculated. Motor control for driving and controlling the permanent magnet synchronous motor using the estimated electrical angle calculated by the electrical angle estimating means when an abnormality of the rotation angle sensor is detected. An electric power steering apparatus comprising:
The electrical angle estimation means sets a range in which the calculated induced voltage is equal to or lower than a set voltage as a dead zone, and the estimated electrical angle does not change according to the calculated induced voltage when the calculated induced voltage enters the dead zone. And a dead zone changing unit that narrows the dead zone as the steering torque detected by the steering torque sensor decreases.

  In the electric power steering apparatus of the present invention, a permanent magnet synchronous motor is provided in the steering mechanism, and the motor control means generates steering assist torque by drivingly controlling the permanent magnet synchronous motor. The motor control means drives and controls the permanent magnet synchronous motor using the electrical angle detected by the rotation angle sensor in accordance with the motor control value calculated by the motor control value calculation means. The motor control value calculation means calculates a motor control value (for example, a current command value) such that the steering assist torque increases as the steering torque detected by the steering torque sensor increases.

  The motor control means, for example, has a d-axis that is a direction through which a permanent magnet magnetic field of a permanent magnet synchronous motor (hereinafter simply referred to as a motor) penetrates, and a direction orthogonal to the d-axis (an electrical angle of π / 2 from the d-axis). The motor is driven and controlled by current vector control using dq coordinates that define the q axis that is the forward direction).

  When the rotation angle sensor fails, the motor cannot be controlled. Therefore, an electric power steering apparatus according to the present invention includes a sensor abnormality detection unit that detects an abnormality of a rotation angle sensor, and an electric angle that estimates an electric angle of a motor when the abnormality of the rotation angle sensor is detected by the sensor abnormality detection unit. An estimation means is provided. The electrical angle estimation means calculates an induced voltage generated in the motor by calculation, and calculates an estimated electrical angle of the motor based on the calculated induced voltage that is the calculated value. The induced voltage generated in the motor is proportional to the motor angular velocity. Therefore, for example, the electrical angle estimation means obtains the estimated angular velocity of the motor from the calculation induced voltage every predetermined time, calculates the amount of rotation of the motor per predetermined time at this estimated angular velocity as the electrical angle addition amount, The estimated electrical angle can be calculated by adding the electrical angle addition amount to the estimated electrical angle in the motor rotation direction. The motor control means drives and controls the motor using the estimated electrical angle estimated by the electrical angle estimation means when an abnormality of the rotation angle sensor is detected. That is, sensorless control is performed.

  In order to obtain the calculation induced voltage, current measurement and voltage measurement are required. For this reason, the voltage induced by the current measurement error, voltage measurement error, etc. is included in the calculation induced voltage, and the calculation induced voltage is zero during motor energization even when the motor is stopped. do not become. Therefore, even during steering (a state where the steering wheel is stopped at the cut-in position), the estimated electrical angle is advanced and the motor torque fluctuates. This torque fluctuation appears as vibration of the steering wheel, and lowers the driver's steering feeling. Therefore, in the present invention, when the dead zone processing means sets the range where the calculated induced voltage is equal to or lower than the set voltage as the dead zone, and the calculated induced voltage enters the dead zone, the estimated electrical angle changes according to the calculated induced voltage. Do not. Note that a value derived from the operation induced voltage in the process of calculating the estimated electric angle of the motor, for example, a dead band provided in the estimated angular velocity of the motor corresponds to the dead band processing means of the present invention. In this case, the angular velocity corresponding to the set voltage of the calculation induced voltage sets the range of the dead zone of the estimated angular velocity.

  When a dead zone is set for the calculation induced voltage, the estimated electrical angle does not change in a state where the rotation of the motor is stopped. For this reason, the steering handle does not vibrate during steering. However, when the driver cuts out the steering wheel (when turning from the neutral position), the estimated electrical angle is difficult to advance, and the steering operation is caught. When the steering torque is detected by cutting the steering handle and the motor coil is energized, the permanent magnet provided on the rotor is attracted by the magnetic field generated by the motor coil. At this time, since the estimated electrical angle does not advance unless the calculation induced voltage can get over the dead zone, the magnetic field generated by the motor coil does not rotate. Therefore, the rotor is settled at a position where the permanent magnet is attracted to the magnetic field of the motor coil. For this reason, the steering assist torque cannot be generated and the steering operation is caught.

  Therefore, in the present invention, the dead zone changing means narrows the dead zone as the steering torque detected by the steering torque sensor decreases, that is, sets the setting voltage of the calculation induced voltage for setting the dead zone. At the moment when the steering wheel is cut out, the dead zone is set narrow because the steering torque is small. Accordingly, when the steering wheel is cut out, the calculation induced voltage easily exceeds the dead zone, so that the estimated electrical angle can be advanced. Thereby, the driver does not feel the steering operation when he / she cuts out the steering wheel.

  As a result, according to the present invention, it is possible to suppress a decrease in steering feeling when performing sensorless control.

  The “steering torque” in the present invention represents the magnitude of the steering torque, and does not distinguish the steering direction. In the present invention, the dead zone changing means reduces the set voltage and narrows the dead zone as the steering torque decreases. However, the dead zone setting (setting voltage setting) changes linearly according to the steering torque. Any of those to be changed and stepwise to be changed according to the steering torque may be used. For example, a wide dead zone (a dead zone with a large set voltage) may be provided when the steering torque is greater than the set torque, and a narrow dead zone (a dead zone with a small set voltage) may be provided when the steering torque is equal to or less than the set torque. Further, “narrowing the dead zone” includes those that do not provide a dead zone. Therefore, a dead zone may be provided when the steering torque is greater than the set torque, and no dead zone may be provided when the steering torque is equal to or less than the set torque.

  Another feature of the present invention is that the dead zone changing means does not provide the dead zone when the steering torque detected by the steering torque sensor is equal to or lower than a preset torque. .

  ADVANTAGE OF THE INVENTION According to this invention, when a driver | operator cuts out a steering wheel, it can prevent reliably giving a feeling of being caught in steering operation. In this case, the set torque may be set to a value smaller than the steering torque that the driver applies to the steering wheel during steering.

1 is a schematic configuration diagram of an electric power steering apparatus according to an embodiment of the present invention. It is a functional block diagram showing the process of the microcomputer of assist ECU. It is a functional block diagram showing the process of an electrical angle estimation part. It is explanatory drawing explaining a dq coordinate and a (gamma) -delta coordinate. It is a graph showing an assist map. It is a circuit diagram of the electric motor used for calculation of an induced voltage. It is a graph showing a dead zone processing map. It is a graph explaining the calculation mechanism of an electrical angle correction amount. It is a graph showing the setting angle area | region of (delta) axis in dq coordinate. It is explanatory drawing explaining angle (theta) 1. FIG. It is explanatory drawing explaining angle (theta) 3. FIG. It is a graph showing the relationship between detection value eγ / e and electrical angle error Δθe. It is a graph showing an electrical angle correction amount calculation map. It is a graph showing the assist torque characteristic with respect to the direction of an electric current vector. It is explanatory drawing showing the state to which the presumed electrical angle was fixed. It is a graph showing the assist torque characteristic with respect to the direction of an electric current vector. It is explanatory drawing showing the relationship of the force when an electric current vector is orient | assigned to the q-axis direction. It is a graph showing the relationship between an actual motor angular velocity and a calculation induced voltage. It is a graph showing a dead zone processing map as a modification.

  Hereinafter, an electric power steering apparatus according to an embodiment of the present invention will be described with reference to the drawings. FIG. 1 shows a schematic configuration of an electric power steering apparatus for a vehicle according to the embodiment.

  The electric power steering apparatus includes a steering mechanism 10 that steers steered wheels by a steering operation of a steering handle 11, an electric motor 20 that is assembled to the steering mechanism 10 and generates a steering assist torque, and an electric motor 20 for driving the electric power steering apparatus. The motor drive circuit 30 and the electronic control device 100 that controls the operation of the electric motor 20 are provided as main parts. Hereinafter, the electronic control device 100 is referred to as an assist ECU 100.

  The steering mechanism 10 is a mechanism for turning the left and right front wheels FWL and FWR by a rotation operation of the steering handle 11, and includes a steering shaft 12 connected to the steering handle 11 so as to rotate integrally with the upper end. A pinion gear 13 is connected to the lower end of the steering shaft 12 so as to rotate integrally. The pinion gear 13 meshes with rack teeth formed on the rack bar 14 and constitutes a rack and pinion mechanism together with the rack bar 14. Knuckles (not shown) of the left and right front wheels FWL and FWR are steerably connected to both ends of the rack bar 14 via tie rods 15L and 15R. The left and right front wheels FWL and FWR are steered left and right according to the axial displacement of the rack bar 14 accompanying the rotation of the steering shaft 12 around the axis.

  An electric motor 20 is assembled to the rack bar 14. The electric motor 20 corresponds to the permanent magnet synchronous motor of the present invention, and in this embodiment, a three-phase brushless DC motor, which is a representative example thereof, is used. The rotating shaft of the electric motor 20 (hereinafter simply referred to as the motor 20) is connected to the rack bar 14 via the ball screw mechanism 16 so as to be able to transmit power, and by this rotation, a steering force is applied to the left and right front wheels FWL and FWR. Grant and assist steering operation. The ball screw mechanism 16 functions as a speed reducer and a rotation-linear converter, and decelerates the rotation of the motor 20 and converts it into a linear motion and transmits it to the rack bar 14.

  A steering torque sensor 21 is provided on the steering shaft 12. The steering torque sensor 21 detects, for example, a twist angle of a torsion bar (not shown) interposed in an intermediate portion of the steering shaft 12 with a resolver or the like, and the steering torque Tr applied to the steering shaft 12 based on the twist angle. Is detected. As for the steering torque Tr, the operation direction of the steering wheel 11 is identified by positive and negative values. In the present embodiment, the steering torque Tr when the steering handle 11 is steered in the right direction is indicated by a positive value, and the steering torque Tr when the steering handle 11 is steered in the left direction is indicated by a negative value. Therefore, when discussing the magnitude of the steering torque Tr, the absolute value | Tr | is used. In this embodiment, the torsion angle of the torsion bar is detected by a resolver, but it can also be detected by another rotation angle sensor such as an encoder.

  The motor 20 is provided with a rotation angle sensor 22. The rotation angle sensor 22 is incorporated in the motor 20 and outputs a detection signal corresponding to the rotation angle position of the rotor of the motor 20, and is constituted by a resolver, for example. The rotation angle sensor 22 outputs a detection signal indicating the rotation angle θm of the motor 20 to the assist ECU 100. The assist ECU 100 detects the electrical angle θe of the motor 20 from the rotation angle θm. The electrical angle θe of the motor 20 has two types, that is, an electrical angle detected by the rotation angle sensor 22 and an electrical angle obtained by estimation which will be described later. The electrical angle detected by the sensor 22 is called an actual electrical angle θea, and the electrical angle obtained by estimation is called an estimated electrical angle θeb. In the present embodiment, a resolver is used as the rotation angle sensor 22, but other rotation angle sensors such as an encoder may be used.

  The motor drive circuit 30 comprises a three-phase inverter circuit composed of six switching elements 31 to 36 made of MOS-FET (Metal Oxide Semiconductor Field Effect Transistor). Specifically, a circuit in which a first switching element 31 and a second switching element 32 are connected in series, a circuit in which a third switching element 33 and a fourth switching element 34 are connected in series, a fifth switching element 35 and a first switching element 6 is connected in parallel with a circuit in which switching elements 36 are connected in series, and a power supply line 37 to the motor 20 is drawn from between two switching elements (31-32, 33-34, 35-36) in each series circuit. The configuration is adopted.

  The motor drive circuit 30 is provided with a current sensor 38 that detects a current flowing through the motor 20. The current sensor 38 detects the current flowing in each phase (U phase, V phase, W phase), and outputs detection signals corresponding to the detected current values Iu, Iv, Iw to the assist ECU 100. Hereinafter, the measured three-phase current values are collectively referred to as a motor current Iuvw. The motor drive circuit 30 is provided with a voltage sensor 39 that detects the terminal voltage of the motor 20. The voltage sensor 39 detects the terminal voltage of each phase (U phase, V phase, W phase), and outputs detection signals corresponding to the detected voltage values Vu, Vv, Vw to the assist ECU 100. Hereinafter, the measured three-phase terminal voltages are collectively referred to as a motor terminal voltage Vuvw.

  As for each switching element 31-36 of the motor drive circuit 30, a gate is connected to assist ECU100, respectively, and a duty ratio is controlled by the PWM control signal output from assist ECU100. Thereby, the drive voltage of the motor 20 is adjusted to the target voltage.

  The assist ECU 100 includes a microcomputer including a CPU, a ROM, a RAM, and the like as a main part. The assist ECU 100 is connected to a steering torque sensor 21, a rotation angle sensor 22, a current sensor 38, a voltage sensor 39, and a vehicle speed sensor 25 that detects a vehicle speed, and a steering torque Tr, a rotation angle θm, motor currents Iu, Iv, Iw. , Detection signals representing motor terminal voltages Vu, Vv, Vw and vehicle speed v are input. Based on the input detection signal, a command current to be supplied to the motor 20 is calculated so as to obtain an optimum steering assist torque (hereinafter simply referred to as assist torque) according to the driver's steering operation. The duty ratios of the switching elements 31 to 36 of the motor drive circuit 30 are controlled so as to flow.

  Next, a power supply system of the electric power steering apparatus will be described. The electric power steering device is supplied with power from an in-vehicle power supply device 80. The in-vehicle power supply device 80 is configured by connecting in parallel a main battery 81 that is a general in-vehicle battery having a rated output voltage of 12V and an alternator 82 having a rated output voltage of 14V that is generated by the rotation of the engine. A power supply source line 83 and a ground line 84 are connected to the in-vehicle power supply device 80. The power supply source line 83 branches into a control system power line 85 and a drive system power line 86. The control system power supply line 85 functions as a power supply line for supplying power to the assist ECU 100. The drive system power supply line 86 functions as a power supply line that supplies power to both the motor drive circuit 30 and the assist ECU 100.

  An ignition switch 87 is connected to the control system power supply line 85. A main power relay 88 is connected to the drive system power line 86. The main power supply relay 88 is turned on by a control signal from the assist ECU 100 to form a power supply circuit to the motor 20. The control system power supply line 85 is connected to the power supply + terminal of the assist ECU 100, and includes a diode 89 on the load side (assist ECU 100 side) from the ignition switch 87 in the middle. The diode 89 is a backflow prevention element that is provided with the cathode facing the assist ECU 100 and the anode facing the in-vehicle power supply device 80, and allows energization only in the power supply direction.

  The drive system power line 86 is provided with a connecting line 90 branched from the main power relay 88 and connected to the control system power line 85 on the load side. This connection line 90 is connected to the assist ECU 100 side from the connection position of the diode 89 in the control system power supply line 85. A diode 91 is connected to the connecting line 90. The diode 91 is provided with the cathode facing the control system power line 85 side and the anode facing the drive system power line 86 side. Therefore, the circuit configuration is such that power can be supplied from the drive system power supply line 86 to the control system power supply line 85 via the connection line 90, but power cannot be supplied from the control system power supply line 85 to the drive system power supply line 86. . The drive system power supply line 86 and the ground line 84 are connected to the power supply input section of the motor drive circuit 30. The ground line 84 is also connected to the ground terminal of the assist ECU 100.

  Next, control of the motor 20 performed by the assist ECU 100 will be described. As shown in FIG. 4, the assist ECU 100 has a d-axis in a direction through which a magnetic field of a permanent magnet provided in the rotor of the motor 20 passes, and a direction perpendicular to the d-axis (the electrical angle with respect to the d-axis is only π / 2). The rotation of the motor 20 is controlled by current vector control using dq coordinates with the q axis defined in the advance direction. The electrical angle θe is a rotation angle between the axis passing through the U-phase coil and the d-axis. The d-axis component of the current vector is called d-axis current, and the q-axis component is called q-axis current. The q-axis current acts so that a magnetic field is generated in the q-axis direction. Therefore, the q-axis current generates motor torque. On the other hand, since the d-axis current generates a magnetic field in the d-axis direction, it cannot generate motor torque and is used for field-weakening control. The assist ECU 100 controls the current phase so that the current vector moves on the q axis in order to obtain the maximum motor torque efficiency (d-axis current is zero).

  The assist ECU 100 determines the dq coordinate by detecting the electrical angle θe when performing such current vector control. The electrical angle θe is obtained from the rotation angle signal detected by the rotation angle sensor 22, but the electrical angle θe cannot be obtained if the rotation angle sensor 22 fails. Therefore, when the rotation angle sensor 22 fails, the assist ECU 100 calculates an estimated electrical angle θeb by a process described later, and performs current vector control using the estimated electrical angle θeb. In this case, the control axis that estimates the d axis is called the γ axis, and the control axis that estimates the q axis is called the δ axis.

  Next, functions of the assist ECU 100 will be described with reference to FIG. FIG. 2 is a functional block diagram showing functions processed by program control of the microcomputer of the assist ECU 100. The assist ECU 100 includes an assist current command unit 101. As shown in FIG. 5, the assist current command unit 101 stores an assist map for setting the target assist torque T * according to the steering torque Tr and the vehicle speed v. The assist current command unit 101 receives the steering torque Tr output from the steering torque sensor 21 and the vehicle speed v output from the vehicle speed sensor 25, and calculates the target assist torque T * by referring to this assist map. In this case, the target assist torque T * increases as the steering torque Tr increases and decreases as the vehicle speed v increases. FIG. 5 is an assist map for steering in the right direction, and the assist map for steering in the left direction has opposite signs to the steering torque Tr and the target assist torque T * with respect to those in the right direction ( In other words, it will be negative). The assist current command unit 101 calculates the q-axis command current Iq * in the dq coordinates by dividing the calculated target assist torque T * by the torque constant. Further, the assist current command unit 101 sets the d-axis command current Id * to zero (Id * = 0).

  The q-axis command current Iq * and the d-axis command current Id * calculated in this way are output to the feedback control unit 102. The feedback control unit 102 calculates a deviation ΔIq obtained by subtracting the q-axis actual current Iq from the q-axis command current Iq *, and the q-axis actual current Iq is changed to the q-axis command current Iq * by proportional-integral control using the deviation ΔIq. The q-axis command voltage Vq * is calculated so as to follow. Similarly, a deviation ΔId obtained by subtracting the d-axis actual current Id from the d-axis command current Id * is calculated, and the d-axis actual current Id follows the d-axis command current Id * by proportional-integral control using the deviation ΔId. D-axis command voltage Vd * is calculated.

  The q-axis actual current Iq and the d-axis actual current Id are obtained by converting the detected values Iu, Iv, and Iw of the three-phase current actually flowing in the coil of the motor 20 into a two-phase current in the dq coordinate. The conversion from the three-phase currents Iu, Iv, and Iw to the two-phase currents Id and Iq in the dq coordinate is performed by the three-phase / two-phase coordinate conversion unit 103. The three-phase / two-phase coordinate conversion unit 103 receives the electrical angle θe output from the electrical angle selection unit 122, and based on the electrical angle θe, the three-phase currents Iu, Iv, and Iw detected by the current sensor 38 are obtained. Conversion into d-q coordinate two-phase currents Id and Iq.

A conversion matrix C for converting from three-phase coordinates to dq coordinates is expressed by the following equation (1).

  As will be described later, the electrical angle selection unit 122 outputs the actual electrical angle θea of the motor 20 as the electrical angle θe when the abnormality of the rotation angle sensor 22 is not detected, and the abnormality of the rotation angle sensor 22 is detected. If it is, the estimated electrical angle θeb of the motor 20 is output as the electrical angle θe.

  The q-axis command voltage Vq * and the d-axis command voltage Vd * calculated by the feedback control unit 102 are output to the 2-phase / 3-phase coordinate conversion unit 105. The two-phase / three-phase coordinate conversion unit 105 converts the q-axis command voltage Vq * and the d-axis command voltage Vd * into the three-phase command voltages Vu * and Vv * based on the electrical angle θe output from the electrical angle selection unit 122. , Vw *, and the converted three-phase command voltages Vu *, Vv *, Vw * are output to the PWM signal generator 106. The PWM signal generator 106 outputs PWM control signals corresponding to the three-phase command voltages Vu *, Vv *, Vw * to the switching elements 31 to 36 of the motor drive circuit 30. As a result, the motor 20 is driven, and an assist torque that follows the target assist torque T * is applied to the steering mechanism 10.

  The rotation detection signal output from the rotation angle sensor 22 is output to the actual electrical angle conversion unit 120 and the sensor abnormality detection unit 121. The actual electrical angle conversion unit 120 calculates the actual electrical angle θea of the motor 20 from the rotation detection signal output by the rotation angle sensor 22, and outputs the calculated actual electrical angle θea to the electrical angle selection unit 122. The rotation angle sensor according to the present invention includes the rotation angle sensor 22 and the actual electrical angle conversion unit 120. The sensor abnormality detection unit 121 detects an abnormality of the rotation angle sensor 22 based on the rotation detection signal output from the rotation angle sensor 22. When a resolver is employed as the rotation angle sensor 22, it is conceivable that the detection coil or the excitation coil in the resolver may be disconnected or cause an insulation failure. Therefore, the sensor abnormality detection unit 121 monitors the amplitude of the output signal of the detection coil, and determines that it is a sensor abnormality when the amplitude is out of the preset allowable range. In addition, since a pair of detection coils are provided such that the phase of the output signal is shifted by π / 2, an abnormality can be detected by comparing the two output signals. For example, when a sine wave signal is output from one detection coil and a constant value signal is output from the other detection coil, it is abnormal even when the combination of two output signals contradicts each other. It can be determined that there is. The sensor abnormality detection unit 121 determines whether or not the rotation angle sensor 22 is abnormal in this way, and outputs a sensor abnormality determination signal F indicating the presence or absence of the abnormality. For example, the sensor abnormality detection unit 121 sets the sensor abnormality determination signal F to “1” when it is determined that there is an abnormality, and sets the sensor abnormality determination signal F to “0” when it is determined that there is no abnormality. To do.

  If an abnormality occurs in the rotation angle sensor 22, the electric angle cannot be detected, and the motor 20 cannot be driven by current vector control. Therefore, the assist ECU 100 includes an electrical angle estimation unit 110 that estimates the electrical angle so that the rotation control of the motor 20 can be continued even when the rotation angle sensor 22 is abnormal. The electrical angle estimator 110 starts to operate when a sensor abnormality determination signal (F = 1) is input. When attention is paid to the function implemented by program control, as shown in FIG. A dead zone processing unit 112, a dead zone changing unit 113, an estimated angular velocity calculation unit 114, an estimated electrical angle calculation unit 115, an estimated electrical angle correction unit 116, a rotation direction estimation unit 117, and an electrical angle error detection unit 118. And an electrical angle correction amount calculation unit 119.

  The induced voltage calculator 111 receives detection signals representing the motor terminal voltages Vu, Vv and Vw output from the voltage sensor 39 and detection signals representing the motor currents Iu, Iv and Iw output from the current sensor 38. The induced voltage e ′ generated in the motor 20 is calculated as follows.

As shown in FIG. 6, when the induced voltage of the U phase of the motor 20 is eu, the induced voltage of the V phase is ev, and the induced voltage of the W phase is ew, the induced voltages eu, ev, and ew are expressed by the following equations (2), It is obtained by (3) and (4).
eu = Vu-Iu.R-Vm (2)
ev = Vv−Iv · R−Vm (3)
ew = Vw−Iw · R−Vm (4)
Here, Vm is a midpoint voltage, and R is the winding resistance of each phase coil. The midpoint voltage Vm may be calculated as Vm = (Vu + Vv + Vw) / 3.

  In this case, to be exact, a voltage component (L · dI / dt) due to the inductance L of the coil of each phase should be added. However, in the calculation of the induced voltage, the influence of the inductance L is very small. Is considered to be zero. The calculation may be performed in consideration of the voltage (L · dI / dt) due to the inductance L.

誘起電圧演算部１１１は、誘起電圧ｅ’の演算結果を不感帯処理部１１２に出力する。以下、誘起電圧演算部１１１により演算された誘起電圧ｅ’を演算誘起電圧ｅ’と呼ぶ。尚、式（５）における電気角θｅは、現時点において推定されている電気角となる。つまり、後述するように所定の短い周期で繰り返し計算される推定電気角θｅｂの最新値が使用される。 The induced voltage e ′ of the motor 20 is obtained by converting the three-phase induced voltages eu, ev, ew into the induced voltages ed, eq in the two-phase dq coordinate system by the following formula (5), and then the following formula (6 ).

The induced voltage calculation unit 111 outputs the calculation result of the induced voltage e ′ to the dead band processing unit 112. Hereinafter, the induced voltage e ′ calculated by the induced voltage calculation unit 111 is referred to as a calculated induced voltage e ′. Note that the electrical angle θe in equation (5) is the electrical angle estimated at the present time. That is, as will be described later, the latest value of the estimated electrical angle θeb that is repeatedly calculated at a predetermined short period is used.

  The dead band processing unit 112 receives the calculation induced voltage e ′ output from the induced voltage calculation unit 111 and the dead band designation signal S output from the dead band changing unit 113. The dead zone processing unit 112 stores a dead zone processing map as shown in FIG. In this dead zone processing map, the horizontal axis represents the operation induced voltage e ', and the vertical axis represents the induced voltage e after the dead zone processing (referred to herein as the corrected induced voltage e). The dead zone processing unit 112 receives the dead zone designation signal S from the dead zone changing unit 113. When the dead zone designation signal S is “1”, the dead zone processing map is used to calculate the induced voltage e ′. A dead zone process is performed to calculate a corrected induced voltage e. When the dead zone designation signal S is “0”, no dead zone processing is performed on the calculation induced voltage e ′.

  The dead zone changing unit 113 inputs the steering torque Tr detected by the steering torque sensor 21, and determines whether or not the magnitude | Tr | of the steering torque Tr is larger than a preset torque Tr0. When the magnitude | Tr | of the steering torque Tr is larger than the preset set torque Tr0, the dead zone designation signal S is set to “1”. Further, when the magnitude | Tr | of the steering torque Tr is equal to or less than the set torque Tr0, the dead zone designation signal S is set to “0”. The dead zone changing unit 113 outputs the dead zone designation signal S to the dead zone processing unit 112.

  Accordingly, the dead zone processing unit 112 performs the dead zone processing on the calculated induced voltage e ′ using the dead zone processing map only when the magnitude | Tr | of the steering torque Tr is larger than the set torque Tr0. This set torque Tr0 is set in advance to a value smaller than the steering torque applied to the steering wheel 11 by the driver during steering.

  The dead zone processing map is obtained by setting the relationship between the calculated induced voltage e 'and the corrected induced voltage e in advance in order to obtain the corrected induced voltage e' from the calculated induced voltage e 'output from the induced voltage calculating unit 111. The dead zone processing unit 112 refers to the dead zone processing map, and when the calculation induced voltage e ′ is equal to or lower than the first set voltage e1 (0 ≦ e ′ ≦ e1), the corrected induced voltage e is set to 0 (zero). Set to. The range (0 ≦ e ′ ≦ e1) of the calculation induced voltage e ′ is the dead zone Y. When the calculation induced voltage e ′ is greater than the first set voltage e1 and equal to or less than the second set voltage e2 (e1 <e ′ ≦ e2), a value obtained by multiplying the calculation induced voltage e ′ by the gain K (K · e ′) is set to the corrected induced voltage e. The gain K is set to increase from 0 to 1 as the calculation induced voltage e ′ increases in the range of e1 to e2. Therefore, the range (e1 <e ′ ≦ e2) of the calculation induced voltage e ′ is a low sensitivity band. When the calculated induced voltage e ′ exceeds the second set voltage e2 (e ′> e2), the calculated induced voltage e ′ is set as the corrected induced voltage e as it is. Since the dead zone Y is provided to compensate for the calculation error of the calculation induced voltage e ', the first set voltage e1 is set to a magnitude (small value) corresponding to the degree of the calculation error.

  The dead zone processing unit 112 outputs a corrected induced voltage e obtained by correcting the calculation induced voltage e ′ by such dead zone processing. Further, when the magnitude | Tr | of the steering torque Tr is equal to or less than the set torque Tr0, the dead zone processing unit 112 outputs the calculation induced voltage e ′ as it is. Hereinafter, the output (corrected induced voltage e or calculated induced voltage e ′) of the dead band processing unit 112 is simply referred to as an induced voltage e.

  Here, the reason why such dead zone processing is performed will be described. Since the induced voltage generated in the motor 20 is proportional to the motor angular velocity, the motor rotation angle per unit time can be estimated by detecting this induced voltage, and the electrical angle is advanced in synchronization with the rotation of the rotor. be able to. However, since the calculated induced voltage e ′ calculated by the induced voltage calculation unit 111 includes voltage components due to errors such as current values, voltage value measurement errors, winding resistance values, etc., as shown in FIG. Even in the state where 20 (rotor) is not rotating, the operation induced voltage e ′ does not become zero while the motor is running. Therefore, if the electrical angle is estimated using the calculated induced voltage e ′ calculated by the induced voltage calculation unit 111 as it is, the electrical angle is advanced during the steering and the torque of the motor 20 varies. This torque fluctuation appears as vibration on the steering wheel 11. Therefore, in order to suppress the output of the calculation induced voltage e ′ due to an error, a dead zone is provided in the calculation induced voltage e ′.

  When a dead zone is set for the operation induced voltage e ′, the estimated electrical angle does not change in a state where the rotation of the motor 20 is stopped. For this reason, the steering handle 11 does not vibrate during steering. However, when the driver cuts out the steering handle 11 (when turning from the neutral position), the estimated electrical angle is difficult to advance, and the steering operation is caught.

  When the steering torque is detected by cutting out the steering handle 11 and the motor coil is energized, the permanent magnet provided on the rotor is attracted by the magnetic field generated by the motor coil. At this time, since the estimated electrical angle does not advance unless the operation-induced voltage e ′ can overcome the dead zone, the magnetic field generated by the motor coil does not rotate. Therefore, as shown in FIG. 15, the rotor is settled at a position where the permanent magnet is attracted to the magnetic field of the motor coil. That is, the current vector is held in a state in which it is directed in the d-axis direction. For this reason, the steering assist torque cannot be generated and the steering operation is caught.

  Therefore, when the magnitude | Tr | of the steering torque Tr is equal to or less than the set torque Tr0, the dead zone changing unit 113 instructs the dead zone processing unit 112 not to provide a dead zone in the calculated induced voltage e ′. Therefore, at the moment when the driver starts to turn the steering wheel 11 from the neutral position, the dead zone of the calculation induced voltage e 'is not set because the steering torque Tr is small, and the estimated electrical angle advances. Thereby, the rotor of the motor 20 rotates in synchronization with the movement of the current vector. As a result, the driver does not feel the steering operation. Further, when the steering handle 11 is not operated, the steering torque Tr becomes zero and the motor 20 is not energized, so that vibration of the steering handle 11 due to torque fluctuation does not occur.

  In addition, when the magnitude | Tr | of the steering torque Tr becomes larger than the set torque Tr0, the dead zone changing unit 113 instructs the dead zone processing unit 112 to provide a dead zone in the calculated induced voltage e ′. As a result, the estimated electrical angle does not change during steering, and the steering handle 11 does not vibrate.

The induced voltage e output from the dead zone processing unit 112 is input to the estimated angular velocity calculation unit 114. The estimated angular velocity calculation unit 114 estimates the motor angular velocity ω by the following equation (7) by using the proportional relationship between the induced voltage e generated in the motor 20 and the motor angular velocity.
ω = e / Ke (7)
Ke is a motor induced voltage constant [V / (rad / s)] representing the relationship between the angular velocity of the motor 20 and the induced voltage. Hereinafter, the estimated motor angular velocity ω is referred to as an estimated angular velocity ω.

  The estimated angular velocity calculation unit 114 outputs the estimated angular velocity ω that is the calculation result to the estimated electrical angle calculation unit 115. The assist ECU 100 performs various arithmetic processes by a microcomputer at a predetermined short arithmetic cycle. Therefore, the electrical angle that the rotor of the motor 20 has rotated during one calculation cycle can be obtained from the estimated angular velocity ω and the calculation cycle. Therefore, the estimated electrical angle calculation unit 115 calculates an electrical angle obtained by rotating the rotor of the motor 20 during one calculation cycle as an electrical angle addition amount Δθa, and calculates the estimated electrical angle calculated at the calculation timing one cycle before. The current estimated electrical angle is calculated by adding the electrical angle addition amount Δθa.

  In this case, since it is necessary to determine the direction in which the electrical angle is added, that is, the rotation direction of the motor 20, the estimated electrical angle calculation unit 115 inputs information representing the rotation direction of the motor 20 from the rotation direction estimation unit 117. . The rotation direction estimation unit 117 regards the direction of the steering torque Tr detected by the steering torque sensor 21 as the rotation direction of the motor 20, and outputs information indicating the direction (sign) of the steering torque Tr.

The electrical angle addition amount Δθa is calculated by the following equation (8).
Δθa = Kf · sign (Tr) · ω (8)
Here, Kf is a constant for obtaining the electrical angle (rad) at which the rotor of the motor 20 rotates during one calculation cycle from the motor angular velocity (rad / s), and corresponds to the calculation cycle (s). Sign (Tr) represents the sign of the steering torque Tr (the direction of the torque acting on the steering shaft 12). If the steering torque Tr is a positive value or zero, sign (Tr) = 1, and the steering torque Tr is negative. If the value is, sign (Tr) = − 1.

Since the estimated electrical angle calculation unit 115 needs to grasp the estimated electrical angle calculated at the calculation timing one cycle before, the latest value of the estimated electrical angle θeb output from the estimated electrical angle correction unit 116 described later, that is, The estimated electrical angle θeb (n−1) calculated at the calculation timing one cycle before is input and stored. As shown in the following equation (9), the estimated electrical angle θeb (n−1) The current estimated electrical angle θeb ′ is calculated by adding the addition amount Δθa.
θeb ′ = θeb (n−1) + Δθa (9)

  In this case, the initial value of the estimated electrical angle θeb (n−1) is a value immediately before the abnormality of the rotation angle sensor 22 is detected by the sensor abnormality detection unit 121. The estimated electrical angle calculation unit 115 inputs the actual electrical angle θea output from the actual electrical angle conversion unit 120 from when no abnormality of the rotation angle sensor 22 is detected, and always stores and updates the latest actual electrical angle θea. . When it is detected that the sensor abnormality determination signal F output from the sensor abnormality detection unit 121 is switched to “1” representing the abnormality of the rotation angle sensor 22, the actual electrical angle θea immediately before the abnormality detection is calculated as the estimated electrical angle θeb (n -1) and the calculation of the estimated electrical angle θeb ′ described above is started. Thereafter, the estimated electrical angle calculation unit 115 uses the estimated electrical angle θeb calculated by the estimated electrical angle correction unit 116 as the estimated electrical angle θeb (n−1) in Expression (9) in the next calculation cycle. The estimated electrical angle θeb is replaced with the estimated electrical angle θeb (n−1) to sequentially store and update.

The estimated electrical angle calculation unit 115 outputs the calculated estimated electrical angle θeb ′ to the estimated electrical angle correction unit 116. In addition to the estimated electrical angle θeb ′, the estimated electrical angle correction unit 116 receives the rotation direction information output from the rotation direction estimation unit 117 and the electrical angle correction amount Δθc output from the electrical angle correction amount calculation unit 119. To do. The estimated electrical angle correction unit 116 corrects the estimated electrical angle θeb ′ by adding the electrical angle correction amount Δθc to the estimated electrical angle θeb ′ in the rotational direction in order to improve robustness against the step-out of the motor 20 described later. To do. That is, the estimated electrical angle correction unit 116 calculates the estimated electrical angle θeb by the following equation (10). This estimated electrical angle θeb is the final estimated electrical angle.
θeb = θeb ′ + sign (Tr) · Δθc (10)

  The estimated electrical angle correction unit 116 outputs the calculated estimated electrical angle θeb to the electrical angle selection unit 122. The electrical angle selection unit 122 inputs the actual electrical angle θea and the estimated electrical angle θeb, reads the sensor abnormality determination signal F from the sensor abnormality detection unit 121, and the sensor abnormality determination signal F indicates that the rotation angle sensor 22 is abnormal. In the case of “1” representing this, the estimated electrical angle θeb is selected. When the sensor abnormality determination signal F is “0” indicating that the rotation angle sensor 22 is normal, the actual electrical angle θea is selected. The electrical angle selector 122 outputs the selected actual electrical angle θea or estimated electrical angle θeb as the electrical angle θe.

  The electrical angle θe is output to the three-phase / two-phase coordinate conversion unit 103 and the two-phase / three-phase coordinate conversion unit 105, and is used for the coordinate conversion calculation described above. Therefore, the assist ECU 100 performs current vector control using dq coordinates defined by the estimated electrical angle, that is, γ-δ coordinates, when an abnormality of the rotation angle sensor 22 is detected.

  Next, a configuration for calculating the electrical angle correction amount Δθc will be described. As described above, the assist current command unit 101 sets the current command value so that the current vector is directed in the q-axis direction in order to maximize the motor torque efficiency. In this case, as shown in FIG. 17, the energization amount of the motor coil is controlled so that the combined force of the assist torque Tm and the driver's steering force Ts is balanced with the axial force F. The axial force F represents a force necessary for displacing the rack bar 14.

  When the axial force increases rapidly and the assist torque generated by the motor 20 is insufficient, the rotor is rotated in the direction opposite to the steering direction. If the rotation angle sensor 22 is operating normally, the electrical angle can be detected with high accuracy, so that the direction of the current vector can follow the q-axis direction in accordance with the rotation of the rotor. Therefore, as shown at the position P1 in FIG. FIG. 16 shows the characteristic of assist torque (motor torque) with respect to the direction (electrical angle) of the current vector. However, when the rotation angle sensor 22 malfunctions and performs sensorless control using the estimated electrical angle, the direction in which the steering torque acts and the rotation direction of the rotor are opposite, and the estimated angular velocity ω is relatively large. Become. For this reason, as shown by the arrow in FIG. 16, the estimated electrical angle advances with respect to the rotor and the motor 20 steps out. As a result, the assist torque that can be generated by the motor 20 is reduced, and the current vector enters a reverse assist region where the direction of the current vector has advanced by 180 ° or more from the d-axis. The assist torque characteristic in the reverse assist region is opposite in sign to the assist torque characteristic in which the direction of the current vector is in the range of 0 ° to 180 °.

  In addition, when sensorless control is performed, the actual dq coordinates and the estimated γ-δ coordinates are shifted by an error of the estimated electrical angle. For this reason, if current vector control is performed on the basis of the γ-δ coordinates, the direction of the current vector deviates from the q-axis direction. In this case, if the current vector is directed in a direction delayed from the q-axis, the motor 20 can be driven even if the electrical angle is relatively advanced by rotating the rotor in the opposite direction to the steering direction due to the increase of the axial force. Since the assist torque that can be generated by the motor increases, the reverse rotation of the rotor is suppressed, and the motor 20 can be prevented from stepping out. However, when the current vector is directed in the q-axis direction or in a direction advanced from the q-axis, the assist torque that can be generated by the motor 20 is increased when the rotor rotates backward and the electrical angle relatively advances. descend. Therefore, the reverse rotation of the rotor cannot be suppressed and the motor 20 steps out.

  Therefore, in the present embodiment, an electrical angle correction amount Δθc for correcting the estimated electrical angle is calculated so that the δ axis is directed in a direction in which the electrical angle is delayed from the q axis, and this electrical angle correction amount Δθc is calculated as the estimated electrical angle. By supplying the correction unit 116, the step-out of the motor 20 is suppressed. The electrical angle correction amount Δθc is calculated by the electrical angle error detection unit 118 and the electrical angle correction amount calculation unit 119.

In calculating the electrical angle correction amount Δθc, it is necessary to grasp the difference in electrical angle between the q-axis and the δ-axis. The state of difference between the q axis and the δ axis can be detected based on the γ-axis induced voltage generated in the γ-axis direction (γ-axis direction component of the induced voltage e generated in the q-axis direction). The γ-axis induced voltage eγ can be expressed by the following equation (11).
eγ = Vγ−R · Iγ + ω · L · Iδ (11)
Where Vγ is the γ-axis direction component of the armature voltage, R is the coil winding resistance, Iγ is the γ-axis direction component of the armature current, L is the coil inductance, and Iδ is the δ-axis direction component of the armature current. To express.

  The induced voltage e always occurs on the q axis. Therefore, as shown in FIG. 8A, when the q-axis and the δ-axis coincide, the γ-axis induced voltage eγ becomes zero (eγ = 0). Further, as shown in FIG. 8B, when the δ axis is delayed with respect to the q axis, the value of the γ-axis induced voltage eγ becomes a negative value (eγ <0). Further, as shown in FIG. 8C, when the δ axis is advanced with respect to the q axis, the value of the γ-axis induced voltage eγ becomes a positive value (eγ> 0).

Further, an electrical angle error between the q axis and the δ axis can also be detected. When the electrical angle error between the q axis and the δ axis is Δθe, the relationship between the induced voltage (actually induced voltage) e generated in the q axis direction and the γ axis induced voltage eγ is expressed by the following equation (12). be able to.
eγ = e · sin Δθe (12)
Therefore,
sin Δθe = eγ / e = (Vγ−R · Iγ + ω · L · Iδ) / e (13)
The electrical angle error Δθe can be obtained from the relational expression (13).

  Here, Vγ, Iγ, and Iδ on the right side of Expression (13) are the motor terminal voltages Vu, Vv, Vw, and motor using the conversion matrix C of Expression (1) that converts from the three-phase coordinates to the dq coordinates. It can be calculated from the currents Iu, Iv, Iw.

  If the electrical angle error Δθe between the q-axis and the δ-axis can be detected, the δ-axis can be made to substantially coincide with the q-axis position by correcting the estimated electrical angle by the electrical angle error Δθe. However, as described above, if the δ-axis is corrected so as to coincide with the q-axis, the motor torque efficiency is maximized, but the motor 20 is likely to step out due to fluctuations in the axial force. Therefore, in the present embodiment, the estimated electrical angle θeb ′ is set so that the electrical angle of the δ axis is delayed from the electrical angle of the q axis, that is, the δ axis is directed in a direction in which the electrical angle is delayed from the q axis. to correct. In this case, in order to ensure a good balance between the robustness against the motor 20 step-out and the motor torque efficiency, as shown in FIG. 9, the estimated electrical angle θeb so that the δ-axis falls within the set angle region A in the dq coordinates. Correct '. In the set angle region A, the boundary angle at which the electrical angle delay from the q axis is minimum is called the minimum delay electrical angle θmin, and the boundary angle at which the electrical angle delay from the q axis is maximum is the maximum delay electrical angle θmax. Call.

The minimum delay electrical angle θmin is based on the angle θ1 set from the margin for the load increase, the angle θ2 considering various errors, and the angle θ3 that can be corrected by the estimated electrical angle correction feedback as follows. Is set.
θmin = θ1 + θ2-θ3
Note that the angles θ1, θ2, and θ3 are defined as positive angles [deg] in the counterclockwise direction (electrical angle delay direction) in FIG. 9 from the q axis.

As the current command value Iq * increases, the influence of vibration due to the step-out of the motor 20 increases. Therefore, the angle θ1 is set to a margin angle so that the motor 20 does not step out with respect to an increase in load when the current command value Iq * at which the influence of vibration becomes large becomes the predetermined value I0. As shown in FIG. 10, when the current command value Iq * is I0, the motor torque is T0, the axial force balanced with the motor torque T0 is F, the expected load increase is ΔF, and the force is directed toward the angle θ1. If the motor torque that generates the motor torque T0 is T1,
F = T0
F + ΔF = T1
T1 · cos θ1 = T0
Therefore, the angle θ1 can be expressed by the following equation.
θ1 = cos ⁻¹ {(T0 / (T0 + ΔF)}

  The angle θ2 is set in consideration of various calculation errors and current measurement errors. Further, as shown in FIG. 11, even if the direction of the current vector advances from the q-axis, the angle θ3 is stepped out by returning the direction of the current vector to the q-axis side by correction feedback of the estimated electrical angle described later. A limit angle that can be corrected so as not to be set is set, and is determined by experiment in advance.

  On the other hand, the maximum delay electrical angle θmax is set in consideration of thermal performance and assist performance. The motor coil and the motor drive circuit 30 generate heat as the direction of the current vector approaches the d-axis side. Accordingly, the angle θ4 is set as the angle θ4 of the delay angle from the q-axis of the current vector that is the limit that satisfies the thermal performance when performing sensorless control. Further, the assist torque decreases as the direction of the current vector approaches the d-axis side. Accordingly, the delay angle from the q-axis of the current vector that is the limit satisfying the assist performance when performing sensorless control is set as the angle θ5. The assist performance when performing sensorless control is the minimum necessary assist performance, and is set, for example, to the ability to output assist torque that allows the driver to steer to the maximum steering angle while the vehicle is traveling. . The maximum delayed electrical angle θmax is set to an angle with a smaller delay angle from the q-axis among the angles θ4 and θ5 so that both the thermal performance and the assist performance are satisfied.

  In this way, the current vector during the sensorless control is corrected by correcting the estimated electrical angle θeb ′ so that the δ axis enters the set angle region A set between the minimum delayed electrical angle θmin and the maximum delayed electrical angle θmax. Is directed toward the set angle region A. In the present embodiment, the range from the q axis to the minimum delay electrical angle θmin is set smaller than the range of the set angle region A (angle θx from the minimum delay electrical angle θmin to the maximum delay electrical angle θmax). .

  Returning to the description of the electrical angle estimation unit 110 in FIG. The electrical angle estimation unit 110 includes an electrical angle error detection unit 118 and an electrical angle correction amount calculation unit 119. The electrical angle error detection unit 118 includes detection signals indicating motor terminal voltages Vu, Vv, and Vw output from the voltage sensor 39, detection signals indicating motor currents Iu, Iv, and Iw output from the current sensor 38, and a dead zone. The induced voltage e output from the processing unit 112 and the estimated angular velocity ω output from the estimated angular velocity calculating unit 114 are input, and the above-described value of eγ / e (γ-axis induction with respect to the induced voltage e generated in the q-axis direction). The ratio of the voltage eγ is calculated. As described above, the relationship of sin Δθe = eγ / e is established. Therefore, eγ / e is a value corresponding to the electrical angle error Δθe between the q axis and the δ axis. Therefore, the electrical angle error detection unit 118 outputs the value of eγ / e as a detection value to the electrical angle correction amount calculation unit 119 without calculating the electrical angle error Δθe. FIG. 12 shows the relationship between the detected value e / eγ and the electrical angle error Δθe.

  The electrical angle correction amount calculation unit 119 receives the detection value eγ / e output from the electrical angle error detection unit 118, and calculates the electrical angle correction amount Δθc in one calculation cycle with reference to the electrical angle correction amount calculation map. . As shown in FIG. 13, the electrical angle correction amount calculation map sets a correspondence relationship between the detected value eγ / e and the electrical angle correction amount Δθc.

  The electrical angle correction amount Δθc is a value for correcting the estimated electrical angle θeb ′ so that the δ axis falls within the set angle region A in the dq coordinates in order to achieve both robustness against step-out and motor torque efficiency. . Therefore, in the electrical angle correction amount calculation map, the dead zone Z is set in the range of the detected value eγ / e corresponding to the minimum delay electrical angle θmin and the maximum delay electrical angle θmax which are the boundaries of the set angle region A. Here, the detected value eγ / e corresponding to the minimum delayed electrical angle θmin is set to x1, and the detected value eγ / e corresponding to the maximum delayed electrical angle θmax is set to x2. In this case, the dead zone Z of the detection value eγ / e is set between x1 and x2. These x1 and x2 are determined from the relationship between the electrical angle error Δθe and the detected value eγ / e (see equation (13), FIG. 12) and set.

  When the detected value eγ / e is a positive value, the δ axis is directed in a direction where the electrical angle is advanced with respect to the q axis. Therefore, in this case, the electrical angle correction amount Δθc is set to a negative value. When the detected value eγ / e is a negative value, the δ axis is oriented in a direction in which the electrical angle is delayed from the q axis. In this case, if the detected value eγ / e exceeds x1, the electrical angle correction amount Δθc is set to a negative value in order to improve the robustness against the step-out of the motor 20. When the detected value eγ / e is equal to or smaller than x1 and equal to or larger than x2, since the δ axis is in the set angle region A, the electrical angle correction amount Δθc is set to 0 (zero). When the detected value eγ / e is less than x2, the electrical angle correction amount Δθc is set to a positive value.

In the electrical angle correction amount calculation map, when the detected value eγ / e is greater than or equal to x0 that is greater than x1, and when the detected value eγ / e is less than or equal to x3 that is less than x2, the electrical angle correction amount Δθc is Limited to a certain upper limit. In this embodiment, as shown in FIG. 13, the upper limit value of the electrical angle correction amount Δθc when the electrical angle of the δ axis is behind the set angle region A is Δθc _maxlag, and the electrical angle of the δ axis is set. Assuming that the upper limit value of the electrical angle correction amount Δθc when it is ahead of the angle region A is Δθc _maxlead , the upper limit value Δθc _maxlead is set to a value larger than the upper limit value Δθc _maxlag (absolute value comparison).

  The electrical angle correction amount Δθc is such that the detected value eγ / e is in the dead zone Z when the detected value eγ / e is in the range of x0 to x1 and when the detected value eγ / e is in the range of x2 to x3. An electrical angle correction amount Δθc having a magnitude proportional to the amount deviating from is set. In this case, the electrical angle correction sensitivity is set to be different depending on whether the electrical angle of the δ axis is ahead or behind the set angle region A. The electrical angle correction sensitivity represents the ratio of the electrical angle correction amount Δθc to the amount by which the detected value eγ / e deviates from the dead zone Z. In the present embodiment, the correction sensitivity (inclination of the characteristic line G1 in FIG. 13) when the electrical angle of the δ axis is advanced from the set angle region A is delayed from the set angle region A by the electrical sensitivity of the δ axis. Is set to be larger than the correction sensitivity (inclination of the characteristic line G2 in FIG. 13).

  When the electrical angle correction amount calculation unit 119 calculates the electrical angle correction amount Δθc with reference to the electrical angle correction amount calculation map, the electrical angle correction amount calculation unit 119 outputs the calculation result to the estimated electrical angle correction unit 116. As described above, the estimated electrical angle correction unit 116 corrects the estimated electrical angle θeb ′ by adding the electrical angle correction amount Δθc to the estimated electrical angle θeb ′ output from the estimated electrical angle calculation unit 115 in the rotation direction. . Then, the estimated electrical angle θeb obtained by the correction is output to the electrical angle selection unit 122.

  The assist ECU 100 repeats the above-described process at a short calculation cycle. Accordingly, when no abnormality of the rotation angle sensor 22 is detected (sensor abnormality determination signal F = 0), current vector control is performed using the dq coordinates set based on the actual electrical angle θea. When an abnormality of the rotation angle sensor 22 is detected (sensor abnormality determination signal F = 1), the electrical angle estimation unit 110 calculates the estimated electrical angle θeb, and the dq coordinate is calculated based on the estimated electrical angle θeb. Current vector control is performed using the γ-δ coordinates that are estimated. At the time of sensorless control, the electrical angle correction amount Δθc is calculated according to the electrical angle correction amount calculation map and the estimated electrical angle θeb ′ is corrected. Therefore, as shown by the arrow in FIG. Feedback control to enter. Thereby, the direction of the current vector is directed in a direction delayed from the actual q-axis. Therefore, as shown in the position P2 in FIG. 14, the motor torque efficiency is not maximized, but the assist torque increases even if the direction of the current vector advances due to the reverse rotation of the rotor due to the rapid increase of the axial force. Can be suppressed.

According to the electric power steering apparatus of the present embodiment described above, the following operational effects are obtained.
1. When calculating the induced voltage e, if the magnitude of the steering torque Tr is small, no dead zone is set for the calculated induced voltage e ′. Thereby, since the estimated electrical angle is not fixed, when the steering wheel 11 is cut out, the driver does not feel the steering operation being caught. On the other hand, when the magnitude of the steering torque Tr is large, a dead zone is set for the calculation induced voltage e ′. As a result, the estimated electrical angle does not change, and the steering handle 11 does not vibrate during steering. As a result, it is possible to suppress a decrease in steering feeling during sensorless control.
2. In the dead zone processing of the calculation induced voltage e ′, the sensitivity of the calculation induced voltage e ′ is lowered when the calculation induced voltage e ′ falls within the range of the first set voltage e1 to the second set voltage e2 that is the boundary of the dead zone. The gain K (0 to 1) is multiplied, and the gain K is set to be large according to the increase of the calculation induced voltage e ′. Therefore, when the calculation induced voltage e ′ exceeds the dead zone Y, the corrected induced voltage e does not suddenly change to a large value, and appropriate dead zone processing can be performed.
3. When an abnormality of the rotation angle sensor 22 is detected, the estimated electrical angle is corrected so that the electrical angle of the δ axis is delayed from the electrical angle of the q axis. Accordingly, even if the axial force increases rapidly and the rotor of the motor 20 rotates in the reverse direction with respect to the steering direction, the assist torque generated by the motor 20 increases, so that the reverse rotation of the rotor can be suppressed. . As a result, the robustness against the step-out of the motor 20 is improved.
4). Since information on the motor rotation direction necessary for calculating the estimated electrical angle is acquired from the direction of the steering torque Tr detected by the steering torque sensor 21, a special sensor for detecting the motor rotation direction is not required. It can be implemented at a cost. When acquiring information on the rotational direction from the steering torque Tr, when the rotor of the motor 20 rotates in the opposite direction to the steering direction, the rotational direction of the rotor is opposite to the direction in which the estimated electrical angle is advanced. Thus, the relative estimated angular velocity ω increases, so that the motor 20 is likely to step out. However, in this embodiment, since the estimated electrical angle is corrected so that the electrical angle of the δ axis is delayed from the electrical angle of the q axis, reverse rotation of the rotor can be suppressed. As a result, it is possible to achieve both robustness against the step-out of the motor 20 and cost reduction.
5. The electrical angle correction amount Δθc is set so that the electrical angle of the δ axis enters the set angle region A in which the electrical angle is delayed from the q axis. This set angle region A is set in consideration of the margin for load increase, various errors, an angle that can be corrected by the estimated electrical angle correction feedback, thermal performance, and assist performance. Robustness and motor torque efficiency can be secured in a good balance. In addition, the minimum necessary assist performance can be ensured.
6). In calculating the electrical angle correction amount Δθc, it is necessary to detect the difference in electrical angle between the δ axis and the q axis, but in the present embodiment, since the detection value eγ / e is calculated, the δ axis Can be detected up to an angle difference (electrical angle error) between the q axis and the q axis. As a result, the δ axis can be properly maintained in the set angle region A. In addition, referring to the electrical angle correction amount calculation map, in order to calculate the electrical angle correction amount Δθc from the detected value eγ / e, the set angle region A can be easily set by providing the dead zone Z of the detected value eγ / e. can do.
7). In calculating the electrical angle correction amount Δθc, an upper limit value that can correct the estimated electrical angle at one time is set. In this case, the upper limit value Δθc _maxlead when the electrical angle of the δ axis is ahead of the set angle region A is higher than the upper limit value Δθc _maxlag when the electrical angle of the δ axis is delayed from the set angle region A. It is set to a large value. For this reason, when the electrical angle of the δ axis is in a region where the motor 20 is likely to step out, the electric angle correction amount Δθc can be set large, so that the motor 20 can be quickly prevented from stepping out. . On the other hand, when the electrical angle of the δ axis is in a region where the motor torque efficiency is low, the estimated electrical angle can be corrected at a stable control speed to increase the motor torque efficiency.
8). In calculating the electrical angle correction amount Δθc, the correction when the electrical angle of the δ axis is ahead of the set angle region A is higher than the correction sensitivity when the electrical angle of the δ axis is behind the set angle region A. The sensitivity is set larger. For this reason, when the electrical angle of the δ axis is in a region where the motor 20 is likely to step out, the electric angle correction amount Δθc can be set large, so that the motor step out can be quickly avoided. On the other hand, when the electrical angle of the δ axis is in a region where the motor torque efficiency is low, the estimated electrical angle can be corrected at a stable control speed to increase the motor torque efficiency.
9. The range from the q axis to the minimum delay electrical angle θmin is set smaller than the range of the set angle region A (the angle θx from the minimum delay electrical angle θmin to the maximum delay electrical angle θmax). Therefore, the motor torque efficiency can be maintained as high as possible while suppressing the step-out of the motor 20.

  The electric power steering apparatus according to the present embodiment has been described above. However, the present invention is not limited to the above-described embodiment, and various modifications can be made without departing from the object of the present invention.

  For example, in this embodiment, when the magnitude | Tr | of the steering torque Tr is larger than the set torque Tr0, dead zone processing of the calculation induced voltage e ′ is performed, and the magnitude | Tr | of the steering torque Tr is equal to or less than the set torque Tr0. In such a case, the control mode is selectively switched between two modes in which the dead zone processing is not performed, but the control mode can be set to three or more modes. For example, as shown in FIG. 19, when the magnitude | Tr | of the steering torque Tr is equal to or less than the first set torque Tr1, the minimum dead zone Y1 is set, and the magnitude | Tr | When it is greater than the first set torque Tr1 and less than or equal to the second set torque Tr2, a medium dead zone Y2 is set, and when the magnitude | Tr | of the steering torque Tr is greater than the second set torque Tr2, You may make it set the dead zone Y3 of the maximum width. In this case, the dead zone processing unit 112 stores three dead zone processing maps as shown in FIG. Then, the dead zone changing unit 113 outputs a dead zone designation signal S that designates one of the three dead zone processing maps to the dead zone processing unit 112 based on the magnitude | Tr | of the steering torque Tr. The dead zone changing unit 113 calculates a corrected induced voltage e by performing a dead zone process on the calculation induced voltage e ′ using the designated dead zone processing map.

  According to this modified example, since a more appropriate dead zone can be set, it is difficult for the dead zone to be excessive or insufficient, and the steering feeling during sensorless control can be further suppressed. In this modification as well, the dead zone may not be provided when the magnitude | Tr | of the steering torque Tr is equal to or smaller than the set torque smaller than the first set torque Tr1. Further, the width of the dead zone may be continuously changed according to the magnitude | Tr | of the steering torque Tr.

  In this embodiment, a dead zone processing map in which a low sensitivity zone is provided next to the dead zone is used. However, a dead zone processing map in which a low sensitivity zone is not provided may be used.

  In the present embodiment, the estimated electrical angle θeb ′ calculated by the estimated electrical angle calculation unit 115 is corrected using the electrical angle correction amount Δθc. However, it is not always necessary to correct the estimated electrical angle θeb ′. Instead, the estimated electrical angle θeb ′ calculated by the estimated electrical angle calculation unit 115 may be output to the electrical angle selection unit 122. In this case, the electrical angle error detection unit 118, the electrical angle correction amount calculation unit 119, and the estimated electrical angle correction unit 116 are omitted.

  In the present embodiment, the estimated electrical angle is calculated by regarding the direction of the steering torque Tr detected by the steering torque sensor 21 as the rotational direction of the motor 20, but the rotational direction of the motor 20 such as a steering angle sensor can be detected. When a sensor is provided, the rotation direction of the motor 20 may be determined using the detection value of the sensor.

  Further, in this embodiment, the value of the actual electrical angle θea immediately before the abnormality of the rotation angle sensor 22 is detected is used as the initial value θeb (n−1) of the estimated electrical angle, but it is fixed instead. Any value such as a value may be used. This is because even if the initial estimated electrical angle is different from the actual electrical angle, the permanent magnet is attracted and synchronized in the direction of the current vector while the motor 20 is rotating.

  In the present embodiment, the upper limit value of the electrical angle correction amount Δθc and the electrical angle correction sensitivity are different depending on whether the electrical angle of the δ axis is advanced from the set angle region A or delayed. However, it may be set to the same value.

  In the present embodiment, the set angle region A is provided and the estimated electrical angle is corrected so that the electrical angle of the δ axis falls within this set angle region A. The estimated electrical angle may be corrected so that the electrical angle of the axis is delayed by a set angle with respect to the electrical angle of the q axis.

  In the present embodiment, the rack assist type electric power steering apparatus that applies the torque generated by the motor 20 to the rack bar 14 has been described. However, the column assist type that applies the torque generated by the motor to the steering shaft 12 is described. An electric power steering device may be used.

  DESCRIPTION OF SYMBOLS 10 ... Steering mechanism, 11 ... Steering handle, 20 ... Electric motor, 21 ... Steering torque sensor, 22 ... Rotation angle sensor, 25 ... Vehicle speed sensor, 30 ... Motor drive circuit, 38 ... Current sensor, 39 ... Voltage sensor, 100 ... Electronic control device (assist ECU), 101 ... Assist current command unit, 102 ... Feedback control unit, 103 ... 3-phase / 2-phase coordinate conversion unit, 105 ... 2-phase / 3-phase coordinate conversion unit, 106 ... PWM control signal generation unit , 110 ... electrical angle estimation unit, 111 ... induced voltage calculation unit, 112 ... dead zone processing unit, 113 ... dead zone change unit, 114 ... estimated angular velocity calculation unit, 115 ... estimated electrical angle calculation unit, 116 ... estimated electrical angle correction unit, 117: Rotation direction estimation unit, 118 ... Electrical angle error detection unit, 119 ... Electrical angle correction amount calculation unit, 120 ... Actual electrical angle conversion unit, 121 ... Sensor abnormality detection unit, 22 ... the electrical angle selection unit.

Claims (2)

A permanent magnet synchronous motor that is provided in the steering mechanism and generates steering assist torque;
A steering torque sensor for detecting a steering torque input from the steering handle to the steering shaft;
Motor control value calculating means for calculating a motor control value so that the steering assist torque increases as the steering torque detected by the steering torque sensor increases;
A rotation angle sensor for detecting an electrical angle of the permanent magnet synchronous motor;
Sensor abnormality detecting means for detecting abnormality of the rotation angle sensor;
When an abnormality of the rotation angle sensor is detected by the sensor abnormality detecting means, an induced voltage generated in the permanent magnet synchronous motor is obtained by calculation, and the permanent magnet synchronous motor is calculated based on the calculated induced voltage that is the calculated value. An electrical angle estimating means for calculating an estimated electrical angle of
According to the motor control value calculated by the motor control value calculation means, when no abnormality of the rotation angle sensor is detected, the permanent magnet synchronous motor is driven and controlled using the electrical angle detected by the rotation angle sensor. An electric power steering apparatus comprising: motor control means for driving and controlling the permanent magnet synchronous motor using the estimated electrical angle calculated by the electrical angle estimation means when an abnormality of the rotation angle sensor is detected;
The electrical angle estimation means includes
A dead zone processing means for setting a range in which the calculated induced voltage is equal to or lower than a set voltage as a dead band and preventing the estimated electrical angle from changing according to the calculated induced voltage when the calculated induced voltage enters the dead zone; ,
An electric power steering apparatus comprising: dead zone changing means for narrowing the dead zone as the steering torque detected by the steering torque sensor decreases.   2. The electric power steering according to claim 1, wherein the dead zone changing means does not provide the dead zone when a steering torque detected by the steering torque sensor is equal to or lower than a preset set torque. apparatus.

Images