CN111797517B - A self-diagnosis method for on-orbit faults of magnetic torque devices based on linear regression - Google Patents

Abstract

A linear regression-based self-diagnosis method for on-orbit faults of the magnetic torquer. By establishing the input and output model of the magnetic torquer, the least square method is used to solve the configuration parameters of the input and output model of the magnetic torquer after linear fitting. The output characteristics of the rail, using the integral smoothing method to integrate the output of the magnetic torque device in the time domain, and solving the output change threshold of the magnetic torque device output within a given period by integrating the output of the magnetic torque device within a given period , so as to obtain the effective criterion of the on-orbit data of the magnetic torque device, improve the effectiveness of the on-orbit output of the magnetic torque device, an important actuator, and realize the effective judgment of the magnetic torque device's on-orbit output.

Description

A self-diagnosis method for on-orbit faults of magnetic torque devices based on linear regression

technical field

The invention relates to a method for self-diagnosing on-orbit faults of a magnetic torque device based on linear regression, and belongs to the field of spacecraft attitude control.

Background technique

As an important executive mechanism in the spacecraft control subsystem, the magnetic torque device is responsible for the tasks of spacecraft attitude control and spacecraft angular momentum unloading. It is directly related to the accuracy of spacecraft attitude control, and its output effectiveness is very important. To a certain extent, it determines the effect of the attitude control of the spacecraft and significantly improves the mission benefits of the spacecraft.

The pulse width control amount of the magnetic torquer changes in real time with the real-time attitude and orbit changes of the spacecraft. This change is nonlinear and cannot be absolutely predicted. Moreover, as the orbit changes, when the cosine value of the angle between the magnetic torque device and the geomagnetic field is greater than a certain value, the magnetic torque device cannot cut the magnetic induction line of the geomagnetic field, and the pulse width control value will instantly jump to zero. These changes, even the jump from a large effective output value of the pulse width control amount to zero, are the real working characteristics of the magnetic torque device, and this nonlinear change brings a belt for the magnetic torque device to implement on-orbit fault self-diagnosis. Great difficulty came.

Contents of the invention

The technical problem solved by the present invention is: Aiming at the problem that in the current prior art, the traditional magnetic torque device output judgment method is difficult to cope with the jump of the pulse width control quantity, a linear regression-based self-diagnosis of magnetic torque device on-orbit faults is proposed method.

The present invention solves the problems of the technologies described above and is achieved through the following technical solutions:

A method for self-diagnosing on-orbit faults of magnetic torque devices based on linear regression, the steps are as follows:

(1) Establish the input and output model of the magnetic torque device according to the input and output of the magnetic torque device, and define the configuration parameters of the input and output model of the magnetic torque device;

(2) carry out linear fitting to step (1) gained magnetic torque device input-output model, and calculate the configuration parameter of magnetic torque device input-output model according to least square method;

(3) Integrate the output of the magnetic torque device in a given period by means of integral smoothing, and determine the output change threshold of the magnetic torque device according to the integral operation result;

(4) Use the output change threshold of the magnetic torque device obtained in step (3) as the basis for judging the validity of the output of the magnetic torque device within a given period, and establish a self-diagnosis model for the on-orbit fault of the magnetic torque device, so as to realize the self-diagnosis of the on-orbit fault of the magnetic torque device .

In the described step (1), the input and output model of the magnetic torque device is specifically:

y=f(x,a)+b

In the formula, x is the excitation current of the magnetic torque device, y is the pulse width control value received by the magnetic torque device and calculated and output by the on-board computer, f(x, a) is the nonlinear continuous current related to the magnetic torque device excitation current function, a and b are configuration parameters describing the relationship between the magnetic torque device excitation current and the pulse width control quantity.

In described step (2), the specific method of calculating the configuration parameter of magnetic torque device input and output model is:

(2-1) Obtain n sets of sample data of excitation current and pulse width control amount, specifically:

[(x ₁ ,y ₁ ),(x ₂ ,y ₂ ),(x ₃ ,y ₃ ),...,(x _n ,y _n )],(n∈Z+);

In the formula, x ₁ ~ x _n are specific values of excitation current for n groups, and y ₁ ~ y _n are specific values for n groups of pulse width control quantities;

(2-2) Carry out linear fitting to the input and output of the magnetic torque device, and use the least square method to calculate the dispersion Q of the sample data, specifically:

y=ax+b

In the formula, y _i is any sample from y ₁ to y _n , and _xi is any sample from x ₁ to x _n ;

(2-3) Calculation When the dispersion Q is the minimum value, the configuration parameters a and b are specifically:

分别为x，y样本集的平均值。In the formula,

are the mean values of the x and y sample sets, respectively.

In described step (3), the concrete step of determining magnetic torque device output variation threshold is:

(3-1) Fit the pulse width control quantity and excitation current curve that do not satisfy the linear relationship, and integrate n sets of excitation current and pulse width control quantity sample data in the time domain of S control cycles to obtain the pulse The theoretical value Y _i of wide control quantity linear fitting and the real value y _i of telemetry sampling are specifically:

均为不满足线性关系的脉宽控制量及激磁电流拟合曲线的配置参数；In the formula,

Both are the configuration parameters of the pulse width control quantity and the excitation current fitting curve that do not satisfy the linear relationship;

(3-2) Make a difference after integrating the theoretical value Y _i of the linear fitting of the pulse width control quantity and the actual value Y _i of telemetry sampling, where:

In the formula, E _S is the integral difference;

次积分后确定遥测采样阈值b_Smin、b_Smax，具体为：(3-3) Take the maximum value E _nSmax and the minimum value E _nSmin of the integral difference of the pulse width control quantity linear fitting theoretical value Y _i and the real value y _i of telemetry sampling in each control period, respectively, and carry out

Determine the telemetry sampling thresholds b _Smin and b _Smax after the second integration, specifically:

E _nSmin ＝min(E _S ), E _nSmax ＝max(E _S ),

b _Smin = min(E _nSmin ), b _Smax = max(E _nSmax );

(3-4) According to the telemetry sampling threshold obtained in step (3-3), the magnetotorque output change threshold that needs to be satisfied by the pulse width control amount telemetry sampling true value is specifically:

In the step (4), if the output of the magnetic torque device in the selected S control cycles meets the judgment of the output change threshold of the magnetic torque device, the output data is valid, otherwise the output data is invalid.

The advantage of the present invention compared with prior art is:

The present invention provides a linear regression-based on-orbit fault self-diagnosis method for a magnetic torque device. Aiming at the nonlinear output characteristics of the magnetic torque device, a linear relationship between the excitation current of the magnetic torque device and the pulse width control output of the on-board computer is established. The data model establishes the on-orbit fault self-diagnosis model of the magnetic torque device by means of linear fitting, improves the effectiveness of the magnetic torque device, an important actuator, on-orbit output, and can realize the effective judgment of the magnetic torque device's on-orbit output, and improve the Effectiveness of spacecraft attitude control. In the case of adding a small amount of calculation, the validity judgment of the output information of the magnetic torque device is obtained according to the curve parameters fitted by the linear regression.

Description of drawings

Fig. 1 is a schematic flow chart of the on-orbit fault self-diagnosis method of the magnetic torque device provided by the invention;

Fig. 2 is a schematic diagram of the distribution of the linear fitting curve and the actual telemetry value provided by the invention;

Fig. 3 is a schematic diagram of the validity judgment threshold distribution after integral smoothing processing provided by the invention;

Detailed ways

A linear regression-based self-diagnosis method for on-orbit faults of the magnetic torquer, by establishing the input and output model of the magnetic torquer, defining the parameter configuration of the input and output model of the magnetic torquer, and using the least square method to solve the input of the magnetic torquer after linear fitting Output the configuration data of the model, and according to the output characteristics of the magnetic torque device on orbit, use the integral smoothing method to integrate the output of the magnetic torque device in the time domain, and solve the problem by integrating the output of the magnetic torque device within a given period The output change threshold of the magnetic torquer output within a given cycle is used as the basis for self-diagnosis of faults, as shown in Figure 1, and the specific steps are as follows:

(1) Establish the input and output model of the magnetic torquer according to the input and output of the magnetic torquer, define the configuration parameters of the input and output model of the magnetic torquer, and clarify the output characteristics of the magnetic torquer, specifically including:

Since the output of the magnetic torquer is related to its own physical device characteristics, and there is an inductance in the magnetic torquer drive circuit, plus the pulse width control amount of the magnetic torquer and the remote measurement sampling period of the magnetic torquer excitation current are different. Therefore, the excitation current generated by the magnetic torque device has a relatively hysteresis relationship with the pulse width control amount it receives. For the excitation current and pulse width output of the magnetic torquer, the input and output model of the magnetic torquer is established as follows:

y=f(x,a)+b

In the formula, x is the excitation current of the magnetic torque device, y is the pulse width control value received by the magnetic torque device and calculated and output by the on-board computer, f(x, a) is the nonlinear continuous current related to the magnetic torque device excitation current function, a and b are configuration parameters describing the relationship between the magnetic torque device excitation current and the pulse width control quantity;

(2) Carry out linear fitting to step (1) gained magnetic torque device input and output model, and calculate the configuration parameter of magnetic torque device input and output model according to least square method, wherein:

The values of a and b are related to the physical characteristics of the magnetic torque device itself. First, obtain enough sample data through ground experiments, and the specific method for calculating the configuration parameters of the input and output model of the magnetic torque device is as follows:

(2-1) Obtain n sets of sample data of excitation current and pulse width control amount, specifically:

[(x ₁ ,y ₁ ),(x ₂ ,y ₂ ),(x ₃ ,y ₃ ),...,(x _n ,y _n )],(n∈Z+);

In the formula, x ₁ ~ x _n are specific values of excitation current for n groups, and y ₁ ~ y _n are specific values for n groups of pulse width control quantities;

(2-2) Carry out linear fitting to the input and output of the magnetic torque device, and use the least square method to calculate the dispersion Q of the sample data, specifically:

y=ax+b

In the formula, y _i is any sample from y ₁ to y _n , and _xi is any sample from x ₁ to x _n ;

(2-3) Calculate and solve Q to get:

Solving the right side of the equation for the Q value can be obtained. When the dispersion Q is the minimum value, the configuration parameters a and b are specifically:

分别为x，y样本集的平均值；In the formula,

are the mean values of the x and y sample sets, respectively;

(3) Integrate the output of the magnetic torque device in a given period by means of integral smoothing, and determine the output change threshold of the magnetic torque device according to the integral operation result, wherein:

The curve satisfying y=ax+b approximately characterizes the linear relationship between the pulse width control amount of the magnetic torque device and the excitation current of the magnetic torque device. However, the pulse width control quantity is calculated in real time according to the attitude control algorithm, and there is no linear relationship between two adjacent pulse width control quantities in the time domain;

(3-1) Fit the pulse width control quantity and excitation current curve that do not satisfy the linear relationship, and integrate n sets of excitation current and pulse width control quantity sample data in the time domain of S control cycles to obtain the pulse The theoretical value Y _i of wide control quantity linear fitting and the real value y _i of telemetry sampling are specifically:

均为不满足线性关系的脉宽控制量及激磁电流拟合曲线的配置参数；S≤n,S∈Z+；In the formula,

Both are the configuration parameters of the pulse width control quantity and the excitation current fitting curve that do not satisfy the linear relationship; S≤n, S∈Z+;

(3-2) Make a difference after integrating the theoretical value Y _i of the linear fitting of the pulse width control quantity and the actual value Y _i of telemetry sampling, where:

In the formula, E _S is the integral difference;

次积分后确定遥测采样阈值b_Smin、b_Smax，具体为：(3-3) Take the maximum value E _nSmax and the minimum value E _nSmin of the integral difference of the pulse width control quantity linear fitting theoretical value Y _i and the real value y _i of telemetry sampling in each control period, respectively, and carry out

Determine the telemetry sampling thresholds b _Smin and b _Smax after the second integration, specifically:

E _nSmin ＝min(E _S ), E _nSmax ＝max(E _S ),

b _Smin = min(E _nSmin ), b _Smax = max(E _nSmax );

(3-4) According to the telemetry sampling threshold obtained in step (3-3), the magnetotorque output change threshold that needs to be satisfied by the pulse width control amount telemetry sampling true value is specifically:

(4) Use the output change threshold of the magnetic torque device obtained in step (3) as the output validity diagnosis basis of the magnetic torque device within a given period, establish a self-diagnosis model for the on-orbit fault of the magnetic torque device, and realize the self-diagnosis of the on-orbit fault of the magnetic torque device , as shown in Figure 3, for any magnetic torquer excitation current x _i , if the corresponding magnetic torquer pulse width control value y _i , the integral within S control cycles satisfies:

Then the output of the magnetic torque device in the S control cycles is valid; otherwise, it is invalid. Among them, [b _Smin ,b _Smax ] depends on the physical characteristics of the magnetic torque device itself, and is also related to the value of S.

Further explanation is carried out below in conjunction with specific embodiment:

(1) For the excitation current and pulse width output of the magnetic torque device, the input and output model of the magnetic torque device is established as follows:

y=f(x,a)+b

Among them, x is the excitation current of the magnetic torque device, y is the pulse width control quantity received by the magnetic torque device and calculated and output by the on-board computer, x and y are the observable remote measurements respectively; f(x,a) represents the same as A nonlinear continuous function related to the excitation current of a magnetic torquer. a and b are configuration parameters describing the relationship between the magnetic torque device excitation current and the pulse width control quantity;

(2) Three million sets of sample data are obtained through ground experiments, and the input and output sample sets of the magnetic torquer are obtained as follows:

[(x ₁ ,y ₁ ),(x ₂ ,y ₂ ),(x ₃ ,y ₃ ),...,(x _n ,y _n )],(n≤3000000,n∈Z+)

Assuming that the curve fitted by the input and output model of the magnetic torque device is linear, it satisfies:

y=ax+b

The least squares method is used to solve the discrete point combination in the sample set. Let the dispersion Q satisfy:

求解Q可得：When Q is the smallest, the coefficient value that satisfies the fitting curve equation is obtained, and the average values of x and y in the sample set are respectively

Solving Q can be obtained:

It can be seen from the right side of the equation for solving the Q value above that when:

The dispersion Q is the minimum value. At this time, the linear fitting model of the input and output of the magnetic torquer is:

y=70.8731x-170.2057

Among them, x is the excitation current of the magnetic torque device, and y is the pulse width control value received by the magnetic torque device and calculated and output by the on-board computer. At this time, the data distribution of the input and output model of the magnetic torque device after linear fitting is shown in Fig. 2 .

(3) The curve satisfying y=70.8731x-170.2057 approximately characterizes the linear relationship between the pulse width control amount of the magnetic torque device and the excitation current of the magnetic torque device. However, the pulse width control quantity is calculated in real time according to the attitude control algorithm, and there is no linear relationship between two adjacent pulse width control quantities in the time domain.

Let the fitted curve be:

y=70.8731x-170.2057

For the entire data sample [(x ₁ ,y ₁ ),(x ₂ ,y ₂ ),(x ₃ ,y ₃ ),...,(x _n ,y _n )], (n≤3000000,n∈Z+ ), a pair of data samples corresponds to a pair of control period telemetry, and the value of the data sample in the time domain is integrated for 40 control periods, and the linear fitting of the pulse width control quantity generated by the excitation current input within 40 control periods can be obtained The theoretical value Y _i , and the real value y _i of telemetry sampling are respectively:

Make a difference between the integral pulse width control quantity linear fitting theoretical value and the real value of telemetry:

使得样本中的脉宽控制量遥测采样真实值y_i满足：After 75,000 integral operations, there is

So that the actual value y _i of the pulse width control quantity telemetry sampling in the sample satisfies:

的范围为[-798.2160,851.4999]；configuration parameters

The range is [-798.2160,851.4999];

(4) Establish the on-orbit fault self-diagnosis model of the magnetic torque device:

For any magnetic torque device excitation current x _i , the corresponding magnetic torque device pulse width control value y _i , the integral within 40 control cycles, if it satisfies:

Then the output of the magnetic torque device within the 40 control cycles is valid;

的范围[-798.2160,851.4999]取决于磁力矩器本身的物理特性，也与积分周期的取值有关。If it is not satisfied, the output of the magnetic torque device in the 40 control cycles is invalid, and the configuration parameters

The range of [-798.2160,851.4999] depends on the physical characteristics of the magnetic torque device itself, and is also related to the value of the integration period.

The content that is not described in detail in the description of the present invention belongs to the well-known technology of those skilled in the art.

Claims (4)

1. A magnetic torque device on-orbit fault self-diagnosis method based on linear regression, characterized in that the steps are as follows: (1) Establish the input and output model of the magnetic torque device according to the input and output of the magnetic torque device, and define the configuration parameters of the input and output model of the magnetic torque device; (2) carry out linear fitting to step (1) gained magnetic torque device input-output model, and calculate the configuration parameter of magnetic torque device input-output model according to least square method; Wherein, in step (2), the specific method for calculating the configuration parameters of the input and output model of the magnetic torque device is: (2-1) Obtain n sets of sample data of excitation current and pulse width control amount, specifically: [(x ₁ ,y ₁ ),(x ₂ ,y ₂ ),(x ₃ ,y ₃ ),...,(x _n ,y _n )],(n∈Z+); In the formula, x ₁ ~ x _n are specific values of excitation current for n groups, and y ₁ ~ y _n are specific values for n groups of pulse width control quantities; (2-2) Carry out linear fitting to the input and output of the magnetic torque device, and use the least square method to calculate the dispersion Q of the sample data, specifically: y=ax+b

In the formula, y _i is any sample from y ₁ to y _n , and _xi is any sample from x ₁ to x _n ; (2-3) Calculation When the dispersion Q is the minimum value, the configuration parameters a and b are specifically:

In the formula,

are the mean values of the x and y sample sets, respectively; (3) Integrate the output of the magnetic torque device in a given period by means of integral smoothing, and determine the output change threshold of the magnetic torque device according to the integral operation result; (4) Use the output change threshold of the magnetic torque device obtained in step (3) as the basis for judging the validity of the output of the magnetic torque device within a given period, and establish a self-diagnosis model for the on-orbit fault of the magnetic torque device, so as to realize the self-diagnosis of the on-orbit fault of the magnetic torque device . 2. a kind of magnetic torque device on-orbit fault self-diagnosis method based on linear regression according to claim 1, is characterized in that: In the described step (1), the input and output model of the magnetic torque device is specifically: y=f(x,a)+b In the formula, x is the excitation current of the magnetic torque device, y is the pulse width control value received by the magnetic torque device and calculated and output by the on-board computer, f(x, a) is the nonlinear continuous current related to the magnetic torque device excitation current function, a and b are configuration parameters describing the relationship between the magnetic torque device excitation current and the pulse width control quantity. 3. a kind of magnetic torque device on-orbit fault self-diagnosis method based on linear regression according to claim 1, is characterized in that: In described step (3), the concrete step of determining magnetic torque device output variation threshold is: (3-1) Fit the pulse width control quantity and excitation current curve that do not satisfy the linear relationship, and integrate n sets of excitation current and pulse width control quantity sample data in the time domain of S control cycles to obtain the pulse The theoretical value Y _i of wide control quantity linear fitting and the real value y _i of telemetry sampling are specifically:

In the formula, a,

all It is the configuration parameter of the pulse width control quantity and the excitation current fitting curve that do not satisfy the linear relationship; (3-2) Make a difference after integrating the theoretical value Y _i of the linear fitting of the pulse width control quantity and the actual value Y _i of telemetry sampling, where:

In the formula, E _S is the integral difference; (3-3) Take the maximum value E _nSmax and the minimum value E _nSmin of the integral difference of the pulse width control quantity linear fitting theoretical value Y _i and the real value y _i of telemetry sampling in each control period, respectively, and carry out Determine the telemetry sampling thresholds b _Smin and b _Smax after the second integration, specifically: E _nSmin ＝min(E _S ), E _nSmax ＝max(E _S ), b _Smin = min(E _nSmin ), b _Smax = max(E _nSmax ); (3-4) According to the telemetry sampling threshold obtained in step (3-3), the magnetotorque output change threshold that needs to be satisfied by the pulse width control amount telemetry sampling true value is specifically:

4. a kind of magnetic torque device on-orbit fault self-diagnosis method based on linear regression according to claim 1, is characterized in that: In the step (4), if the output of the magnetic torque device in the selected S control cycles meets the judgment of the output change threshold of the magnetic torque device, the output data is valid, otherwise the output data is invalid.

Images