A method for online synchronous identification of multiple parameters of permanent magnet motor

Abstract

The present invention discloses a method for online synchronous identification of multiple parameters of a permanent magnet motor. By adopting two groups of discrete voltage equations at different times to construct a full-rank online multi-parameter identification model, the method realizes synchronous identification of multiple parameters. The method not only solves the problem of insufficient rank of the parameter identification model, but also avoids the parameter coupling problem in the step-by-step identification process. In addition, a forgetting factor function based on the inverse cotangent function is designed, and the size of the forgetting factor can be adaptively adjusted according to the error during the parameter identification process. In the case where a single parameter of the motor suddenly changes or multiple parameters suddenly change or change slowly at the same time, the method can realize synchronous identification of multiple parameters, with fast convergence speed and high identification accuracy, and has high anti-interference and robustness.

Description

Technical Field

[0001] The invention relates to a parameter identification method of a permanent magnet motor. Background Art

[0002] Permanent magnet motors have the advantages of small size, high power density, and high efficiency, and are therefore widely used in new energy electric vehicles, marine vessels, aerospace, and other fields. The efficient operation of permanent magnet motors can improve the endurance of electric vehicles, marine vessels, and aerospace, and the optimal control of motor drive efficiency based on the loss model is largely affected by motor parameters, especially in the harsh environment of aerospace. When the motor is affected by temperature, magnetic saturation, and other factors, its parameters will change accordingly. In order to achieve efficient drive control of permanent magnet motors, it is particularly important to obtain motor parameters in real time and accurately.

[0003] In the current parameter identification algorithms, most of them use the d-axis or q-axis voltage equation of the permanent magnet motor to establish the relevant parameter identification model. The biggest problem of using only one voltage equation for multi-parameter online identification is that the identification model lacks rank, which will directly lead to inaccurate parameter identification results. In the multi-parameter identification algorithm, the step-by-step identification method is more commonly used, but the disadvantage of the step-by-step method is that the coupling parameters of the motor are considered to be constants, which will also lead to large deviations in the identification results. In the fixed forgetting factor recursive least squares parameter identification algorithm, the size of the forgetting factor is a fixed value, and the size of the forgetting factor cannot be adaptively adjusted according to the system error. If the forgetting factor is too large or too small, it will affect the convergence speed and results of the parameter identification algorithm. Therefore, it is necessary to comprehensively solve the problems of the lack of rank of the identification model and the fixed forgetting factor to realize the accurate online identification of multiple parameters of the permanent magnet motor. Summary of the invention

[0004] Technical problem: To address the problems of insufficient rank of the identification model, inability to simultaneously identify multiple parameters and determination of the forgetting factor, a method for online synchronous identification of multiple parameters of a permanent magnet motor is proposed.

[0005] Technical solution: A method for online synchronous identification of multiple parameters of a permanent magnet motor, comprising the following steps:

[0006] Step 1: Determine whether it is the first parameter identification, if so, go to step 2A, otherwise go to step 2B;

[0007] Step 2A: Initialize the parameters in multi-parameter identification;

[0008] Step 2B: Read the forgetting factor λ from the previous multi-parameter identification past , the parameter matrix to be identified and the covariance matrix P past ;

[0009] Step 3: Read the three-phase voltage, current and mechanical angle of the permanent magnet motor, and obtain the voltage u in the dq coordinate system through Clark and Park coordinate transformation d 、u q , current i d 、i q , and calculate the electrical angular velocity ω e ;

[0010] Step 4: Call the multi-parameter identification model with forgetting factor λ past The recursive least squares method is used to calculate the gain matrix K and the parameter matrix to be identified in this parameter identification. Get the resistance value R of the permanent magnet motor s , direct axis inductance value L d , quadrature axis inductance L q and the permanent magnet flux value ψ f , and simultaneously update the covariance matrix P in the multi-parameter identification model;

[0011] Step S5: Calculate the model identification error e, and update the forgetting factor λ based on the inverse cotangent function in the multi-parameter identification model according to the maximum identification error e_max.

[0012] Furthermore, in step 4, the band forgetting factor λ past The recursive least squares algorithm uses the last forgetting factor λ past , the parameter matrix to be identified and the covariance matrix P past , according to the input and output of the current system, the last parameter matrix to be identified Make corrections to obtain the current parameter matrix to be identified Thus, the relevant parameters of the motor can be identified; the band forgetting factor λ past The recursive least squares algorithm is expressed as:

[0013]

[0014] In the formula, the forgetting factor λ past is the dynamic forgetting factor based on the inverse cotangent function, 0<λ past ≤1, φ and y represent the system input and output matrices respectively, φ T is the transpose of the system input matrix, and I is the identity matrix.

[0015] Furthermore, in step 4, the multi-parameter identification model establishment step is:

[0016] First, the voltage equation of the permanent magnet motor in the dq coordinate system is used to perform Euler forward discretization on the voltage equation at time k and time (k-1), and it is divided into two groups. The first group is the voltage equation after discretization at time k, which is expressed as:

[0017]

[0018] Where u d(k) 、u q(k) and i d(k) 、i q(k) They are the voltage and current of the motor at time k in the dq coordinate system, i d(k+1) and i q(k+1) is the current at time (k+1), ω e(k) is the electrical angular velocity at time k, R s , L d , L q , f are the resistance, direct-axis inductance, quadrature-axis inductance and permanent magnet flux of the permanent magnet motor, and T is the sampling period;

[0019] The second group is the voltage equation after discretization at time (k-1), expressed as:

[0020]

[0021] Where u d(k-1) 、u q(k-1) and i d(k-1) 、i q(k-1) are the voltage and current of the motor at time (k-1) in the dq coordinate system, ω e(k-1) is the electrical angular velocity at time (k-1);

[0022] The two sets of discrete voltage equations are then combined as:

[0023]

[0024] Let the system output matrix Parameter matrix to be identified Transpose of the system input matrix Finally, a full-rank model for multi-parameter identification of permanent magnet motor is constructed:

[0025]

[0026] According to the current values ​​at (k-1), k, and (k+1), the voltage and electrical angular velocity values ​​at k and (k-1), the forgetting factor λ is called past The recursive least squares method is used to realize the online synchronous identification of the resistance, inductance and flux linkage of the permanent magnet motor.

[0027] Furthermore, step 5 includes the following steps:

[0028] First, the model identification error e is calculated. The model identification error e is expressed as:

[0029]

[0030] In the formula, φ and y represent the system input and output matrices respectively;

[0031] There are 4 errors in the multi-parameter identification model, namely: where e Rs is the resistance identification error, e ld is the direct axis inductance identification error, e lq is the quadrature-axis inductance identification error, e ψf is the flux identification error;

[0032] Then, according to the four errors identified by the model, the maximum identification error e_max is calculated. The maximum identification error e_max is expressed as:

[0033] e_max=max{e Rs 、e ld 、e lq 、e ψf} (7)

[0034] The forgetting factor function is expressed as:

[0035]

[0036] Finally, the forgetting factor function based on the inverse cotangent function is updated according to the maximum identification error e_max.

[0037] Beneficial effects: The method for online synchronous identification of multiple parameters of a permanent magnet motor of the present invention realizes synchronous identification of multiple parameters by using two sets of discrete voltage equations at different times to construct a full-rank online multi-parameter identification model. The present invention not only solves the problem of insufficient rank of the parameter identification model, but also avoids the parameter coupling problem in the step-by-step identification process. In addition, a forgetting factor function based on the inverse cotangent function is designed, and the size of the forgetting factor can be adaptively adjusted according to the error during the parameter identification process. In the case of a sudden change in a single parameter of the motor or a sudden change or slow change in multiple parameters at the same time, the method can realize synchronous identification of multiple parameters, with a fast convergence speed and high identification accuracy, and at the same time has high anti-interference and robustness. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 is a flow chart of the method of the present invention;

[0039] Figure 2 It is the block diagram of permanent magnet synchronous motor vector control;

[0040] Figure 3 YesR s Identification result simulation diagram;

[0041] Figure 4 YesL d Identification result simulation diagram;

[0042] Figure 5 YesL q Identification result simulation diagram;

[0043] Figure 6 is f Simulation diagram of identification results. DETAILED DESCRIPTION

[0044] The present invention will be further explained below in conjunction with the accompanying drawings.

[0045] like Figure 1 As shown, a method for online synchronous identification of multiple parameters of a permanent magnet motor comprises the following steps:

[0046] Step 1: Determine whether it is the first parameter identification. If so, execute step 2A; otherwise, execute step 2B.

[0047] Step 2A: Initialize the relevant parameters in multi-parameter identification.

[0048] Step 2B: Read the forgetting factor λ from the previous multi-parameter identification past , the parameter matrix to be identified and the covariance matrix P past .

[0049] Step 3: If Figure 3 As shown in the figure, the three-phase voltage, current and mechanical angle of the permanent magnet motor are read, and the voltage u in the dq coordinate system is obtained through Clark and Park coordinate transformation. d 、u q , current i d 、i q , and calculate the electrical angular velocity ω e , providing relevant physical information for parameter identification.

[0050] Step 4: Call the multi-parameter identification model with forgetting factor λ past The recursive least squares method is used to calculate the gain matrix K and the parameter matrix to be identified in this parameter identification. Get the resistance value R of the permanent magnet motor s , direct axis inductance value L d , quadrature axis inductance L q and the permanent magnet flux value ψ f , and update the covariance matrix P in the multi-parameter identification model at the same time.

[0051] Among them, with forgetting factor λ past The recursive least squares algorithm uses the last forgetting factor λ past , the parameter matrix to be identified and the covariance matrix P past , according to the input and output of the current system, the last parameter matrix to be identified Make corrections to obtain the current parameter matrix to be identified So as to identify the relevant parameters of the motor; with forgetting factor λ past The recursive least squares algorithm is expressed as:

[0052]

[0053] In the formula, the forgetting factor λ past is the dynamic forgetting factor based on the inverse cotangent function, 0<λ past ≤1, φ and y represent the system input and output matrices respectively, φ T is the transpose of the system input matrix, and I is the identity matrix.

[0054] The steps to establish the multi-parameter identification model are:

[0055] First, the voltage equation of the permanent magnet motor in the dq coordinate system is used to perform Euler forward discretization on the voltage equation at time k and time (k-1), and it is divided into two groups. The first group is the voltage equation after discretization at time k, which is expressed as:

[0056]

[0057] Where u d(k) 、u q(k) and i d(k) 、i q(k) They are the voltage and current of the motor at time k in the dq coordinate system, i d(k+1) and i q(k+1) is the current at time (k+1), ω e(k) is the electrical angular velocity at time k, R s , L d , L q , f are the resistance, direct-axis inductance, quadrature-axis inductance and permanent magnet flux of the permanent magnet motor, and T is the sampling period;

[0058] The second group is the voltage equation after discretization at time (k-1), expressed as:

[0059]

[0060] Where u d(k-1) 、u q(k-1) and id(k-1) 、i q(k-1) are the voltage and current of the motor at time (k-1) in the dq coordinate system, ω e(k-1) is the electrical angular velocity at time (k-1);

[0061] The two sets of discrete voltage equations are then combined as:

[0062]

[0063] Let the system output matrix Parameter matrix to be identified Transpose of the system input matrix Finally, a full-rank model for multi-parameter identification of permanent magnet motor is constructed:

[0064]

[0065] According to the current values ​​at (k-1), k, and (k+1), the voltage and electrical angular velocity values ​​at k and (k-1), the forgetting factor λ is called past The recursive least squares method is used to realize the online synchronous identification of the resistance, inductance and flux linkage of the permanent magnet motor.

[0066] Step S5: Calculate the model identification error e, and update the forgetting factor λ based on the inverse cotangent function in the multi-parameter identification model according to the maximum identification error e_max, including the following steps:

[0067] First, the model identification error e is calculated. The model identification error e is expressed as:

[0068]

[0069] In the formula, φ and y represent the system input and output matrices respectively;

[0070] There are 4 errors in the multi-parameter identification model, namely: where e Rs is the resistance identification error, e ld is the direct axis inductance identification error, e lq is the quadrature-axis inductance identification error, e ψf is the flux identification error;

[0071] Then, according to the four errors identified by the model, the maximum identification error e_max is calculated. The maximum identification error e_max is expressed as:

[0072] e_max=max{e Rs 、e ld 、e lq 、e ψf} (7)

[0073] The forgetting factor function is expressed as:

[0074]

[0075] Finally, the forgetting factor function based on the inverse cotangent function is updated according to the maximum identification error e_max.

[0076] In step 2A, the initial value of the forgetting factor is λ=1, the sampling period is T=0.00001s, and the voltage u of the dq axis is d(0) 、u d(-1) 、u q(0) 、u q(-1) , current i d(0) 、i d(-1) 、i q(0) 、i q(-1) , and the electrical angular velocity ω of the motor e(0) ,ω e(-1) The initial value of is 0. P past =10 4 I, I is the identity matrix.

[0077] Figure 3-Figure 6 The figure shows a parameter identification simulation example. The single parameter mutation process occurs within 0.3s-0.6s. At 0.3s, the d-axis inductance L d From 22.6mH to 21.6mH, q-axis inductance L at 0.4s q The resistance R changes from 23mH to 22mH in 0.5s. s From 2.32Ω to 3.32Ω, the magnetic flux ψ at 0.6s f From 0.0676Wb to 0.0576Wb; at 0.8s, multiple parameters mutated simultaneously, R s The mutation is 2.32Ω, L d Mutation to 22.6mH, L q Mutation to 23mH, ψ f The sudden change is 0.0676Wb; multiple parameters change slowly at the same time in 1s to 1.4s, R s From 2.32Ω to 3.32Ω, L d From 22.6mH to 21.6mH, L q From 23mH to 22mH, ψ f Slowly changes from 0.0675Wb to 0.0575Wb.

[0078] Figure 3 The resistor R is shown sThe identification results show that after a single parameter mutation of 0.5s and a multiple parameter (resistance, inductance and magnetic flux) mutation of 0.8s, the identification algorithm responds quickly, the convergence time is kept within 0.0002s, the steady-state identification error is less than 0.00005%, and during the slow change of the resistance parameter from 1s to 1.4s, the identification error is within 0.3%.

[0079] Orthogonal axis inductance L d , L q Online identification results are as follows Figure 4 and Figure 5 As shown in the figure, after single parameter mutation and multi-parameter mutation occur, the convergence time of the parameter identification algorithm is within 0.0002s, the steady-state identification error is within 0.00000005%, and the identification result is almost unaffected by the motor parameter coupling. During the parameter slow change process, the identification error can be kept within 0.001%.

[0080] Figure 6 is the magnetic flux ψ f The identification result of f A single parameter mutation occurs at 0.6s, and a multi-parameter mutation occurs at 0.8s. After the mutation occurs, the algorithm converges within 0.0002s, the steady-state identification error is within 0.000003%, and the error is within 1s~1.4sψ f During the slow change process, the identification error is within 0.02%. Therefore, this method can realize multi-parameter synchronous identification, with fast convergence speed and high identification accuracy. It can identify parameters in real time during the slow change process of parameters, and has high anti-interference and robustness.

[0081] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for online synchronous identification of multiple parameters of a permanent magnet motor, characterized in that: The steps include: Step 1: Determine whether it is the first parameter identification, if so, go to step 2A, otherwise go to step 2B; Step 2A: Initialize the parameters in multi-parameter identification; Step 2B: Read the forgetting factor λ from the previous multi-parameter identification past , the parameter matrix to be identified and the covariance matrix P past ; Step 3: Read the three-phase voltage, current and mechanical angle of the permanent magnet motor, and obtain the voltage u in the dq coordinate system through Clark and Park coordinate transformation d 、u q , current i d 、i q , and calculate the electrical angular velocity ω e ; Step 4: Call the multi-parameter identification model with forgetting factor λ past The recursive least squares method is used to calculate the gain matrix K and the parameter matrix to be identified in this parameter identification. Get the resistance value R of the permanent magnet motor s , direct axis inductance value L d , quadrature axis inductance L q and the permanent magnet flux value ψ f , and simultaneously update the covariance matrix P in the multi-parameter identification model; Step S5: Calculate the model identification error e, and update the forgetting factor λ based on the inverse cotangent function in the multi-parameter identification model according to the maximum identification error e_max; In step 4, the band forgetting factor λ past The recursive least squares algorithm uses the last forgetting factor λ past , the parameter matrix to be identified and the covariance matrix P past , according to the input and output of the current system, the last parameter matrix to be identified Make corrections to obtain the current parameter matrix to be identified Thus, the relevant parameters of the motor can be identified; the band forgetting factor λ past The recursive least squares algorithm is expressed as: In the formula, the forgetting factor λ past is the dynamic forgetting factor based on the inverse cotangent function, 0<λ past ≤1, φ and y represent the system input and output matrices respectively, φ T is the transpose of the system input matrix, I is the identity matrix; In step 4, the multi-parameter identification model establishment steps are: First, the voltage equation of the permanent magnet motor in the dq coordinate system is used to perform Euler forward discretization on the voltage equation at time k and time (k-1), and it is divided into two groups. The first group is the voltage equation after discretization at time k, which is expressed as: Where u d(k) 、u q(k) and i d(k) 、i q(k) They are the voltage and current of the motor at time k in the dq coordinate system, i d(k+1) and i q(k+1) is the current at time (k+1), ω e(k) is the electrical angular velocity at time k, R s , L d , L q , f are the resistance, direct-axis inductance, quadrature-axis inductance and permanent magnet flux of the permanent magnet motor, and T is the sampling period; The second group is the voltage equation after discretization at time (k-1), expressed as: Where u d(k-1) 、u q(k-1) and i d(k-1) 、i q(k-1) are the voltage and current of the motor at time (k-1) in the dq coordinate system, ω e(k-1) is the electrical angular velocity at time (k-1); The two sets of discrete voltage equations are then combined as: Let the system output matrix Parameter matrix to be identified Transpose of the system input matrix Finally, a full-rank model for multi-parameter identification of permanent magnet motor is constructed: According to the current values ​​at (k-1), k, and (k+1), the voltage and electrical angular velocity values ​​at k and (k-1), the forgetting factor λ is called past The recursive least squares method is used to realize the online synchronous identification of the resistance, inductance and flux linkage of the permanent magnet motor; Step 5 includes the following steps: First, the model identification error e is calculated. The model identification error e is expressed as: In the formula, φ and y represent the system input and output matrices respectively; There are 4 errors in the multi-parameter identification model, namely: where e Rs is the resistance identification error, e ld is the direct axis inductance identification error, e lq is the quadrature-axis inductance identification error, e ψf is the flux identification error; Then, according to the four errors identified by the model, the maximum identification error e_max is calculated. The maximum identification error e_max is expressed as: e_max=max{e Rs ,and ld ,and lq ,and ψf } (7) The forgetting factor function is expressed as: Finally, the forgetting factor function based on the inverse cotangent function is updated according to the maximum identification error e_max.

Images