KR20240114943A - Mehod and apparatus for estimating temperature of moter coil based on digital twin - Google Patents

Abstract

The motor coil temperature estimation method according to the present invention includes the steps of constructing a heat transfer model to predict indirect heat sources based on temperature data when temperature data is input, and constructing a flow model to predict indirect heat sources based on temperature data. It includes steps of building a digital twin model by coupling the results of the heat transfer model and the flow model, and estimating the temperature of the motor coil using the digital twin model. According to this, it is possible to estimate the temperature of a motor coil operating in a sealed internal space without installing a feedthrough. In addition, a reduction in verification costs during experiments can be expected.

Description

Digital twin-based motor coil temperature estimation method and device {MEHOD AND APPARATUS FOR ESTIMATING TEMPERATURE OF MOTER COIL BASED ON DIGITAL TWIN}

The present invention relates to a digital twin-based motor coil temperature estimation method and device. In particular, a digital twin-based motor coil temperature estimation method and device that can estimate the temperature of the motor coil only with temperature data measured outside the motor. will be.

A motor generates driving torque by flowing current through the motor coil, but if a large current continues to flow through the motor coil, the motor coil may generate heat and cause problems such as damage to the motor coil or demagnetization of the motor magnet. there is. Accordingly, protection such as measuring the temperature of the motor coil and limiting the current flowing through the motor coil is necessary.

However, it is physically difficult to install a temperature sensor directly on the motor coil. In particular, it is difficult to directly measure the temperature of the motor coil inside the module on a system-by-system basis rather than a single product. Moreover, creating a feedthrough through which a sensor can be inserted into the motor housing incurs additional processing costs and creates a discrepancy from the actual system.

To overcome this, digital twin technologies are being developed to measure the temperature of the motor coil without attaching a temperature sensor directly to the motor coil.

Digital twins begin with building a Functional Mock-up Unit (FMU) that closely reflects physical phenomena in reality, and FMUs are largely developed using direct problems and inverse problems.

Existing digital twin technologies are mainly based on the direct approach, and it is difficult to find technology that estimates the temperature of the motor coil using the inverse approach.

Currently, the motor coil temperature estimation method using the direct approach collects electrical data, calculates the power loss of the motor, inputs this as a heat source, and configures an FMU that outputs the temperature. At this time, the power loss of the motor can be largely divided into copper losses, core losses, and mechanical losses, but it is difficult to collect data to measure iron losses and mechanical losses. Therefore, only copper losses are often considered, which can lead to large estimation errors.

In addition, copper loss can be calculated as 3I^2R, and since resistance (R) increases as temperature increases, even though it is a nonlinear model, nonlinearity is often considered linearized, which causes errors.

Republic of Korea Patent No. 10-0797336

The present invention is intended to solve the above problems and provides a digital twin-based motor coil temperature estimation method and device that can estimate the temperature of the motor coil only with temperature data measured outside the motor.

In order to achieve the above object, a motor coil temperature estimation method according to an embodiment of the present invention includes the steps of: when temperature data is input, constructing a heat transfer model for predicting an indirect heat source based on the temperature data; Building a flow model to predict an indirect heat source based on the temperature data; Constructing a digital twin model by coupling the results of the heat transfer model and the flow model; And, estimating the temperature of the motor coil using the digital twin model.

Additionally, the step of constructing the heat transfer model includes constructing a heat transfer numerical model; Setting a stabilizer parameter to control noise of the temperature data; generating a sensitivity coefficient matrix; And, it may include predicting the indirect heat source using a heat transfer model built based on the heat transfer numerical model, the ballast parameter, and the sensitivity coefficient matrix.

In addition, the step of building the flow model includes nonlinear system identification, sparse identification of nonlinear dynamics (SINDy), dynamic mode decomposition with Koopman operator, and least squares method (Normal equation). The indirect heat source can be predicted using any one of the following methods: solution of least-squares problem.

In addition, the step of building the digital twin model includes an explicit coupling method, an implicit coupling method, a surface-2-Surface interface coupling method, The temperature can be estimated using either the Volume-2-Volume coupling method or the Point-2-Point coupling method.

More preferably, in the step of building the digital twin model, the temperature can be estimated using a staggered algorithm.

Meanwhile, a motor coil temperature estimation device according to an embodiment of the present invention includes a heat calculation unit that, when temperature data is input, builds a heat transfer model and uses the heat transfer model to predict an indirect heat source based on the temperature data; a flow calculation unit that builds a flow model and predicts an indirect heat source based on the temperature data using the flow model; And, it may include a heat-flow calculation unit that couples the results of the heat transfer model and the flow model to build a digital twin model, and estimates the temperature of the motor coil using the digital twin model.

In addition, the thermal calculation unit includes a numerical model construction unit for constructing a heat transfer numerical model; a parameter setting unit that sets stabilizer parameters for controlling noise of the temperature data; a matrix generator that generates a sensitivity coefficient matrix; And, it may include a prediction unit that predicts the indirect heat source using the heat transfer model built based on the heat transfer numerical model, the ballast parameter, and the sensitivity coefficient matrix.

In addition, the flow calculation unit includes nonlinear system identification, sparse identification of nonlinear dynamics (SINDy), dynamic mode decomposition with Koopman operator, and normal equation solution of least- The indirect heat source can be predicted using any one of the squares problem.

In addition, the heat-flow calculation unit includes an explicit coupling method, an implicit coupling method, a surface-2-Surface interface coupling method, and a volume-to-surface interface coupling method. The temperature can be estimated using either the Volume-2-Volume coupling method or the Point-2-Point coupling method.

More preferably, the heat-flow calculator can estimate the temperature using a staggered algorithm.

The digital twin-based motor coil temperature estimation method and device according to the present invention as described above can estimate the temperature of a motor coil operating in a closed internal space without installing a feedthrough.

In addition, a reduction in verification costs during experiments can be expected.

Furthermore, it can be used in systems that monitor the temperature of the motor coil and detect abnormalities even in structures where it is difficult to insert a temperature sensor directly into the motor coil, such as nuclear reactors and electric vehicles.

1 is a schematic diagram briefly showing a digital twin-based motor coil temperature estimation device according to an embodiment of the present invention.
Figure 2 is a block diagram schematically showing the configuration of a digital twin-based motor coil temperature estimation device according to an embodiment of the present invention.
Figure 3 is a flowchart illustrating a digital twin-based motor coil temperature estimation method according to an embodiment of the present invention.
Figure 4 is a diagram for explaining a coupling method according to an embodiment of the present invention.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the attached drawings to clarify the technical idea of the present invention. In describing the present invention, if it is determined that a detailed description of related known functions or components may unnecessarily obscure the gist of the present invention, the detailed description will be omitted. Components having substantially the same functional configuration among the drawings are given the same reference numbers and symbols as much as possible, even if they are shown in different drawings. For convenience of explanation, if necessary, the device and method should be described together.

1 is a schematic diagram briefly showing a digital twin-based motor coil temperature estimation device according to an embodiment of the present invention.

Referring to FIG. 1, when temperature data measured by the sensor 20 attached to the outside of the motor 10 is input, the motor coil temperature estimation device 100 can estimate the temperature of the motor coil based on the digital twin. there is.

In one embodiment, the motor coil temperature estimation device 100 may indirectly predict a heat source of temperature data using an inverse approach and estimate the temperature of the motor coil using the predicted heat source. To this end, the motor coil temperature estimation device 100 builds a heat conduction model and a flow model, couples the results of the heat conduction model and the flow model to build a digital twin model, and uses the digital twin model to determine the motor The temperature of the coil can be estimated.

Figure 2 is a block diagram schematically showing the configuration of a digital twin-based motor coil temperature estimation device according to an embodiment of the present invention.

Referring to FIGS. 1 and 2 , the motor coil temperature estimation device 100 may include a thermal calculation unit 110, a flow calculation unit 120, and a heat-flow calculation unit 130.

The thermal calculation unit 110 may construct a linear heat transfer model that represents the thermal behavior of the motor.

When temperature data is input, the thermal calculation unit 110 can build a heat transfer model and predict an indirect heat source based on the temperature data using the heat transfer model.

The thermal calculation unit 110 can calculate the indirect heat source and total temperature using Equation 1 below.

(Formula 1)

,

,

,

here, is a time in the past, is the current time, is the measurement time interval, is the predicted heat source vector, is the measured temperature vector, means the predicted total temperature vector (including temperatures other than those equipped with temperature sensors). also, silver cast It is a virtual temperature vector at the point equipped with a temperature sensor when heated uniformly, and becomes a constant when there is only one temperature sensor. and, is the sensitivity coefficient matrix of the sensor location and is defined as (Equation 2) below.

(Formula 2)

Here, N corresponds to the number of sensors used, and M corresponds to the number of spatially resolved heat sources. is defined in the same way, the number of columns is the same as M, but the number of rows is equal to the total degrees of freedom of the numerical model, which will be described later. For example, if the degree of freedom of the numerical model of the motor built in the numerical model construction stage is 100, becomes a 100×100 matrix.

In an embodiment, the thermal calculation unit 110 may include a numerical model building unit, a parameter setting unit, a matrix generating unit, and a prediction unit.

The numerical model construction unit can build a heat transfer numerical model. In other words, the numerical model construction unit can use various technologies such as the finite element method, finite difference method, and boundary element method to build the heat transfer numerical model.

Numerical analysis techniques mathematically model the laws and conditions that govern a phenomenon, express the answer as a combination of interpolation functions to obtain an approximate solution, and then calculate the size of each basis function. At this time, the mathematical expression is basically converted into a matrix equation for calculating the coefficients of the basis function, and the type of numerical analysis technique is determined depending on the definition method of the basis function.

For example, the finite element method performs numerical analysis by dividing the spatial region of the target object, that is, the motor, into small regions called finite elements in order to systematically define the basis function.

The finite difference method derives matrix equations by numerical integration of differential equations within a geometric domain that is the subject of natural phenomena. A finite number of points created within the geometric domain are called a grid and the density of the grid is Accordingly, the accuracy of the approximate solution improves.

The boundary element method integrates the product of a mathematical expression in the form of a differential equation and a kernel function called the green function over the entire area of the object. Afterwards, numerical analysis is performed by converting the integral over the entire region into a boundary integral format according to the boundary of the object according to Green's theorem and applying boundary conditions.

The parameter setting unit can set stabilizer parameters that control noise in temperature data. The stabilizer parameter is necessary to ensure stability against noise of the temperature sensor 20 used in the system. That is, the stabilizer parameter is The value is determined by the number, location, thermal diffusivity, and performance of the sensors. Since it plays a role in suppressing the instability of (Equation 1) due to measurement error by giving a penalty, it is important to enter an appropriate value. It is important.

The matrix generator may generate a sensitivity coefficient matrix.

The matrix generator predicts the indirect heat source of the motor. In other words, the matrix generator of (Equation 2) class You can generate a sensitivity coefficient matrix by receiving input. here, is used to calculate the indirect heat source, and the calculated heat source and Predict the overall temperature with

In order for the predicted overall temperature to be unique, The number of columns and rows must be the same or there must be more rows than columns, which means that the number of sensors must be greater than the number of heat sources. In general, since the number of spatially resolved heat sources is greater than the number of sensors, a method that can guarantee the above conditions is necessary. Otherwise, it becomes a mathematically ill-posed problem and has multiple non-unique solutions. cause consequences

The prediction unit can predict indirect heat sources using a heat transfer model built based on a heat transfer numerical model, ballast parameters, and sensitivity coefficient matrix.

In an embodiment, when predetermined temperature data is input, the prediction unit may estimate the temperature of the motor coil based on a numerical model, a ballast parameter, and a sensitivity coefficient matrix.

When temperature data is input, the flow calculation unit 120 can build a flow model and predict an indirect heat source based on the temperature data using the flow model. At this time, the equivalent heat source can be estimated.

The flow calculation unit 120 can build a nonlinear flow model to represent flow. That is, the flow calculation unit 120 can analyze the 3D flow, extract data, and build a flow model based on the data.

The flow calculation unit 120 performs nonlinear system identification, sparse identification of nonlinear dynamics (SINDy), dynamic mode decomposition with Koopman operator, and normal equations of the least squares method. You can predict the indirect heat source and estimate the temperature using any one of the equation solution of least-squares problem.

As an example, the model-free method can be represented as shown in Table 1.

approach architecture note System identification
Structural identification (M,D,K) : state
:output
:input Discovering governing equations of nonlinear dynamical systems Indirectly measuring the inputs

The heat-flow calculation unit 130 can build a digital twin model by coupling the results of the heat transfer model and the flow model, and estimate the temperature of the motor coil using the digital twin model. That is, the heat-flow calculation unit 130 separates the non-linear heat transfer model into a linear heat conduction model and a non-linear flow model and builds a digital twin model that couples them using a partitioned method, so that the interaction between flow and structure (fluid-structure interaction) can be modeled separately, and the results can be coupled to calculate supplemented results.

For this purpose, the heat-flow calculation unit 130 uses an explicit coupling method, an implicit coupling method, a surface-2-Surface interface coupling method, The temperature can be estimated using either the Volume-2-Volume coupling method or the Point-2-Point coupling method.

More preferably, the heat-flow calculation unit 130 can estimate the temperature using a staggered algorithm.

FIG. 3 is a flowchart for explaining a digital twin-based motor coil temperature estimation method according to an embodiment of the present invention, and FIG. 4 is a diagram for explaining a coupling method according to an embodiment of the present invention.

With reference to FIGS. 2 and 3 , a method for estimating the motor coil temperature of the motor coil temperature estimating device 100 will be described.

When the external temperature data of the motor 10 measured by the temperature sensor 20 is input, the heat calculation unit 110 and the flow calculation unit 120 build a heat transfer model and a flow model (S110, S115).

In the heat transfer model building step (S110), an indirect heat source is predicted by building a heat transfer model through the heat calculation unit 110. Specifically, the heat transfer model building step (S110) performs a numerical model building step, a stabilizer parameter setting step, a sensitivity coefficient matrix generation step for heat source prediction, and an overall temperature prediction step by inputting temperature data. .

In an embodiment, the numerical model building step mathematically models the laws and conditions governing the driving phenomenon of the motor using the finite element method through the numerical model building unit. At this time, to obtain an approximate solution, the answer can be expressed as a combination of interpolation functions and then the size of each basis function can be calculated. In addition, the mathematical expression is basically converted into a matrix equation for calculating the coefficients of the basis function, and for systematic definition of the basis function, the spatial region of the motor is divided into small elements called finite elements. Numerical analysis is performed by dividing into areas.

The stabilizer parameter setting step sets the stabilizer parameters that control noise in temperature data. In the above-mentioned (Equation 1) is a variable that determines the performance of the stabilizer.

The method of setting the value is ① collecting temperature information according to the driving time of the motor, ② filtering the collected temperature through post-processing and converting it into filtered data with noise removed, ③ This consists of the step of determining the value that minimizes the filtering data and least square error while changing the value.

Since the value is determined by the performance of the temperature sensor, temperature data (information) is collected at a certain period and the increase in noise is observed. At this time, the increase in noise can be observed using techniques such as generalized cross validation, and depending on the increase in noise, the amount of noise can be observed using the method described above. Calibration of values can be performed.

The automatic calibration method of the ballast is ① setting the initial ballast parameters based on the temperature data at the time of the initial test drive, ② calculating the correction values of the ballast parameters at a predetermined period based on the temperature data stored in the memory while heat treatment is performed. , ③ It can be done by setting the correction value as a stabilizer parameter.

The sensitivity coefficient matrix generation step predicts the indirect heat source received by the motor through the matrix generator.

In the overall temperature prediction step by inputting temperature data, indirect heat sources can be predicted using a heat transfer model built based on a heat transfer numerical model, stabilizer parameters, and sensitivity coefficient matrix through the prediction unit. That is, when predetermined temperature data is input, the temperature of the motor coil can be estimated using the constructed heat transfer model.

Meanwhile, in the flow model building step (S115), the flow model is built through the flow calculation unit 120 to predict the indirect heat source. At this time, a nonlinear flow model to express the flow is constructed. In other words, you can analyze 3D flow, extract data, and build a flow model based on the data.

In an embodiment, the fluid calculation unit 120 uses a model-free method to model by applying a gray box such as data-driven modeling without using an artificial intelligence network such as DNN. can do. At this time, the equivalent heat source can be predicted.

Thereafter, in the step of coupling the heat-flow model (S130), a digital twin model is built by coupling the results of the heat transfer model and the flow model through the heat-flow calculation unit 130. In other words, the non-linear heat transfer model is separated into a linear heat conduction model and a non-linear flow model through the heat-flow calculation unit 130, and a digital twin model is constructed by coupling them using a partitioned method, so that the interaction between flow and structure is established. By modeling each action (fluid-structure interaction) and coupling the results, a supplemented result can be calculated.

Referring to FIG. 4, a method of coupling will be described.

In the embodiment, the fluid-structure interaction of fluid (Fluid, X program) and structure (Structure, Y program) is modeled and coupled using a staggered algorithm. Here, the zigzag algorithm may also be referred to as a prediction-correction algorithm.

Y program calculated at time t _n is calculated as a predictor, and in the X program, x _(n+1) is calculated with the predicted value, and then y _(n+1) is calculated again for correction. Since the present invention is solved sequentially in the time direction, it can also be called an explicit coupling method.

Meanwhile, the implicit coupling method has a monolithic algorithm, and this method converges through an iterative calculation process at a specific point in time and then moves on to the next point in time. However, monolithic algorithms have very high computational costs and are inefficient. Therefore, it is desirable to use a zigzag algorithm.

Programs such as MpCCI or preCICE provide interfaces for zigzag or monolithic algorithms.

Then, when predetermined temperature data is input, the temperature of the motor coil is estimated using an inverse approach-based digital twin model (S140).

Meanwhile, the heat-flow calculation unit 130 uses an explicit coupling method, an implicit coupling method, a surface-2-Surface interface coupling method, and a volumetric coupling method. The temperature can be estimated using either the Volume-2-Volume coupling method or the Point-2-Point coupling method.

According to the motor coil temperature estimation method and device according to the present invention, temperature estimation is possible by processing point measured temperature data using digital twin technology, so measurement cost savings can be expected by not inserting a temperature sensor inside the motor. there is.

Meanwhile, the digital twin-based motor coil temperature estimation method according to the present invention can be implemented on a computer system, and a system composed of one or more computers uses software, firmware, hardware, or a combination thereof to enable the system to perform tasks while operating. By installing it on your system, it can be configured to perform each step of the method described above. One or more computer programs may be configured to perform specific operations or tasks by including instructions that, when executed by a data processing device, cause the device to perform the operations.

Meanwhile, the above-described digital twin-based motor coil temperature estimation method can be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. A computer-readable recording medium may include program instructions, data files, data structures, etc., singly or in combination. Program instructions recorded on the medium may be specially designed and constructed for the present invention or may be known and usable by those skilled in the art of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic media such as floptical disks. -Includes optical media (magneto-optical media) and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, flash memory, etc. Examples of program instructions include machine language code, such as that produced by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

So far, the present invention has been examined in detail, focusing on the preferred embodiments shown in the drawings. These embodiments are not intended to limit the invention but are merely illustrative and should be considered from an illustrative rather than a limiting perspective. The true scope of technical protection of the present invention should be determined by the technical spirit of the appended claims rather than the foregoing description. Although specific terms are used in this specification, they are used only for the purpose of explaining the concept of the present invention and are not used to limit the meaning or scope of the present invention described in the claims. Each step of the present invention does not necessarily have to be performed in the order described, but may be performed in parallel, selectively, or individually. Those skilled in the art to which the present invention pertains will understand that various modifications and other equivalent embodiments are possible without departing from the essential technical spirit of the present invention as claimed in the patent claims. Equivalents should be understood to include not only currently known equivalents but also equivalents developed in the future, that is, all components invented to perform the same function regardless of structure.

100: Motor coil temperature estimation device
110: thermal calculation unit
120: Flow calculation unit
130: Heat-fluid calculation unit

Claims (10)

When temperature data is input, constructing a heat transfer model to predict an indirect heat source based on the temperature data;
Building a flow model to predict an indirect heat source based on the temperature data;
Constructing a digital twin model by coupling the results of the heat transfer model and the flow model; and,
A motor coil temperature estimation method comprising: estimating the temperature of the motor coil using the digital twin model. According to paragraph 1,
The step of building the heat transfer model is,
Building a heat transfer numerical model;
Setting a stabilizer parameter to control noise in the temperature data;
generating a sensitivity coefficient matrix; and,
Predicting the indirect heat source using a heat transfer model built based on the heat transfer numerical model, the ballast parameter, and the sensitivity coefficient matrix. A motor coil temperature estimation method comprising a. According to paragraph 1,
The step of building the flow model is,
Nonlinear system identification, sparse identification of nonlinear dynamics (SINDy), dynamic mode decomposition with Koopman operator, and normal equation solution of least-squares problem. A motor coil temperature estimation method for predicting the indirect heat source using a method. According to paragraph 1,
The step of building the digital twin model is,
Explicit coupling method, Implicit coupling method, Surface-2-Surface interface coupling, Volume-2- A motor coil temperature estimation method that estimates the temperature using either volume coupling or point-2-point coupling. According to clause 4,
The step of building the digital twin model is,
A motor coil temperature estimation method that estimates temperature using a staggered algorithm. When temperature data is input, a heat calculation unit that builds a heat transfer model and predicts an indirect heat source based on the temperature data using the heat transfer model;
a flow calculation unit that builds a flow model and predicts an indirect heat source based on the temperature data using the flow model; and,
A motor coil temperature estimation device comprising a heat-flow calculation unit that couples the results of the heat transfer model and the flow model to build a digital twin model, and estimates the temperature of the motor coil using the digital twin model. According to clause 6,
The thermal calculation unit,
A numerical model construction unit that builds a heat transfer numerical model;
a parameter setting unit that sets stabilizer parameters for controlling noise of the temperature data;
a matrix generator that generates a sensitivity coefficient matrix; and,
A prediction unit that predicts the indirect heat source using the heat transfer model built based on the heat transfer numerical model, the ballast parameter, and the sensitivity coefficient matrix. A motor coil temperature estimation device comprising a. According to clause 6,
The flow calculation unit,
Nonlinear system identification, sparse identification of nonlinear dynamics (SINDy), dynamic mode decomposition with Koopman operator, and normal equation solution of least-squares problem. A motor coil temperature estimation device for predicting the indirect heat source using a method. According to clause 6,
The heat-fluid calculation unit,
Explicit coupling method, Implicit coupling method, Surface-2-Surface interface coupling, Volume-2- A motor coil temperature estimation device that estimates temperature using either volume coupling or point-2-point coupling. According to clause 9,
The heat-fluid calculation unit,
A motor coil temperature estimation device that estimates temperature using a staggered algorithm.

Images