Permanent magnet synchronous motor inductance parameter identification method based on data pre-screening

Abstract

The application discloses a kind of permanent magnet synchronous motor inductance parameter identification method based on data pre-screening, including establishing the voltage-current discrete equation of permanent magnet synchronous motor, data screening and inductance parameter identification, wherein, according to voltage-current discrete equation in data screening, the change of current mainly depends on voltage vector under steady state, to construct voltage screening model, according to voltage screening model, select current voltage, current sampling current and next time sampling current and speed information at the moment, and input to parameter identification.This application realizes the fast, accurate and robust online identification of inductance parameter of permanent magnet synchronous motor.

Description

Technical Field

[0001] The present invention relates to the technical field of motors, and more particularly to a method for identifying inductance parameters of a permanent magnet synchronous motor based on data pre-screening. Background Art

[0002] Permanent magnet synchronous motors (PMSMs) are widely used in new energy vehicles, high-end CNC machine tools, industrial automation systems, robotics, rail transportation, and sharp power generation due to their advantages such as high power density, high efficiency, fast response speed, high control precision, and good reliability. They are gradually becoming the mainstream choice for high-performance electric drive systems.

[0003] In order to meet the above performance requirements, constructing a high-precision mathematical model of the permanent magnet synchronous motor has become the basis for the design and optimization of the electric drive system. The current modeling research of the permanent magnet synchronous motor has formed a relatively systematic structural framework, but its ultimate accuracy still depends on the accuracy of the identification of key electrical parameters. Specifically, the main electrical parameters of the permanent magnet synchronous motor include stator resistance, permanent magnet flux and d / q-axis inductance. Although the stator resistance is easily affected by temperature changes, its impact on control accuracy is relatively small. The permanent magnet flux is usually relatively stable under normal operating conditions and the variation range is limited. The d / q-axis inductance changes significantly with current, directly affecting the magnetic field distribution, electromagnetic coupling characteristics and current control response of the permanent magnet synchronous motor. It is one of the core parameters that determine the control performance and dynamic behavior of the permanent magnet synchronous motor.

[0004] However, in actual engineering applications, the accurate values ​​of the d / q-axis inductance at different operating points cannot usually be obtained through simple calibration. Currently, most offline and online identification methods are used to identify the d / q-axis inductance. However, traditional offline identification methods rely on specific excitation tests and data post-processing, which makes it difficult to cover all operating conditions and cannot meet the online modeling requirements of the real-time changes of inductance with the working state. The traditional online identification method of "new model + high-frequency excitation / multiple sampling" has the ability to update in real time, but is limited by problems such as insufficient signal excitation intensity and uneven data quality. Its stability and accuracy are difficult to guarantee, especially in scenarios with high-speed and high-dynamic response requirements. Estimation drift or mutation is prone to occur. In addition, online methods based on identification algorithms (such as observers, least squares, optimization search, etc.) have problems such as insufficient noise robustness, difficulty in balancing convergence speed and accuracy, and lack of data optimization mechanism.

[0005] In view of this, the applicant conducted in-depth research on this basis, which led to the emergence of this case. Summary of the Invention

[0006] The purpose of the present invention is to provide a permanent magnet synchronous motor inductance parameter identification method based on data pre-screening, which can improve the convergence and accuracy of inductance parameter identification, enhance noise robustness and algorithm stability, and realize full-operating-condition and adaptive parameter tracking.

[0007] To achieve the above object, the solution of the present invention is:

[0008] A method for identifying inductance parameters of a permanent magnet synchronous motor based on data pre-screening comprises the following steps:

[0009] Step 1: Establish the voltage-current discrete equation of the permanent magnet synchronous motor as follows:

[0010] (1),

[0011] ,

[0012] ,

[0013] ,

[0014] Where, are parameter matrices, for k Stator sampling current at time +1; for k The stator sampling current at time , , for k time d Stator sampling current on the shaft, for k time q Stator sampling current on the shaft; for k The input voltage at the moment, , for k time d Input voltage on the axis, for k time q Input voltage on the shaft; for k Electrical angular velocity at a moment; for k The sampling noise at time , , for k time d Sampling noise on the axis, for k time q Sampling noise on the axis; To control the cycle time, is the stator resistance; is a 2D rotation matrix and ; is the inductance matrix, ,in and They correspond to the inductance on the d / q axis respectively, is the permanent magnet flux; is the 2nd-order identity matrix and ;

[0015] Step 2: Data screening. According to the voltage-current discrete equation in step 1, the change in current in steady state mainly depends on the input voltage vector of the permanent magnet synchronous motor. Based on this, a voltage screening model is established. Screening data is obtained according to the voltage screening model. The screening data are the output voltage at the current moment, the stator sampling current at the current moment, the stator sampling current at the next moment, and the speed information at the current moment, and are input into the parameter identification.

[0016] Step 3, parameter identification, establish an objective function based on the voltage-current discrete equation, substitute the filtered data into the objective function and the equality constraints in the equality constraints respectively, update the stator sampling current and the stator estimated current in the objective function according to the filtered data and the equality constraints, and substitute the equality constraints substituted into the objective function to obtain an unconstrained minimization problem model, then use the Gauss-Newton method to obtain the solution equation, and then convert it into the parameter matrix correction formula of the permanent magnet synchronous motor, and then obtain the inductance parameter estimation value according to the calculation formula of the inductance parameter.

[0017] The voltage screening model established is as follows,

[0018] (2),

[0019] (3),

[0020] (4),

[0021] Where, is the decision function, H for Heaviside step function, is the weight factor, is the threshold factor, for The cumulative function of for M Function in k The value of the moment, for M Function in k-1 moment value, for Cumulative function of changing values, for V Function in k The value of the moment, For V function k- The value at moment 1, for k The square of the voltage vector at the moment;

[0022] Among them, >0, k Voltage vector at each moment 、 k Current sampling at all times and Current sampling at all times And speed information , input into the parameter identification.

[0023] In the voltage screening model, the square of the voltage vector is used as the screening target, and its spectral density equation is: (5), where for The autocorrelation equation of is the sampling time interval, is the frequency, is a mathematical constant, is the square value of the voltage vector;

[0024] in, The estimated equation is (6), where L is the observation window length of the spectral density function, k To observe the time, The voltage vector is The square value of the moment; then, use The covariance of The estimated equation is adjusted to (7), where for The covariance of is the expected symbol, is the expected value of the voltage vector; thus, formula (7) is used as the theoretical basis for constructing the voltage screening model.

[0025] In step 3, the objective function with constraints is established as follows,

[0026] (8),

[0027] ,

[0028] Where, To minimize the stator current of the objective function, is the stator sampling current, j is the time in the observation window, In the observation window j The stator sampling current at time , In the observation window j The estimated stator current at time , is the artificially designed observation window length parameter, Corresponding to exist k The value of the moment, for k Estimated stator current at time +1, is the maximum allowable current, where All equations that follow are constraints that need to be satisfied in the process of minimizing the objective function.

[0029] In step 2, the next Substitute into the objective function, and at the same time substitute the current output voltage, the current stator sampling current and the current speed information into the equality constraints in the constraint conditions for calculation, and the calculation is obtained k The stator estimated current at time +1 is then substituted into the objective function to obtain the unconstrained minimization problem model as follows: (9), where

[0030] ,

[0031] ,

[0032] ,

[0033] ,

[0034] ,

[0035] ,

[0036] ,

[0037] Where, for k A large matrix consisting of the sampled currents in the observation window at time +1, for k A large matrix consisting of the sampled currents within the observation window at any moment, for k Parameter matrix within the time observation window A large matrix composed of for k Parameter matrix within the time observation window A large matrix composed of for k Parameter matrix within the time observation window A large matrix composed of for k Voltage vector within the observation window at any moment A large matrix composed of for k Electrical angular velocity within the observation window A large vector of

[0038] The Gauss-Newton method is used to solve the unconstrained minimization problem model, and the solution equation is: (10), where for Any parameter matrix in is the corresponding parameter matrix k +1 moment value, is the Jacobian matrix of formula (9), for The transpose of

[0039] In a shorter period of time unchanged, so formula (10) is adjusted to the following expression to obtain the parameter matrix correction formula, which is:

[0040] (11),

[0041] Where, In the observation window j The current vector difference at the moment, In the observation window j time d Shaft current difference, In the observation window j time q Shaft current difference, In the observation window j time The value of any parameter matrix in ; where, , , , where In the observation window j time d Shaft current sampling value, In the observation window j time d Estimated shaft current, In the observation window j timeq Shaft current sampling value, In the observation window j Estimated value of the q-axis current at time t.

[0042] Formula (11) is used to obtain the correction of the parameter matrix B, and then the estimated value of the inductance parameter is estimated according to the estimation formula of the inductance parameter, which is: (12), where To estimate the inductance, To estimate the inverse matrix of the parameter matrix B.

[0043] The estimation formula is obtained by using the formula (1) Reverse deduction to obtain.

[0044] After adopting the above method, the present invention has the following beneficial effects: the present invention adopts data screening and rolling time domain parameter identification, adopts a voltage screening model in data screening, pre-screens the sampled data, screens out the data segments with the most identification value, significantly reduces the estimation variance, and accelerates the inductance convergence, thereby improving the identification accuracy and convergence speed; at the same time, the pre-screening process suppresses low signal-to-noise ratio samples, and combines the rolling time domain observation window to smoothly update the abnormal estimation to avoid jumps or drifts, thereby enhancing noise robustness and algorithm stability. At the same time, the rolling time domain parameter identification updates the identification data set through the sliding observation window to ensure that the model parameters are synchronized with the operating state of the permanent magnet synchronous motor, and provides high-precision inductance parameter information for subsequent current prediction, maximum torque control, flux observation and fault detection control algorithms, thereby realizing full-condition and adaptive parameter tracking; compared with the prior art, the present invention has the characteristics of easy implementation, small computational complexity, fast convergence speed and high estimation accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 This is a control block diagram of the present invention. DETAILED DESCRIPTION

[0046] In order to further explain the technical solution of the present invention, the present invention is described in detail below through specific embodiments.

[0047] A method for identifying inductance parameters of a permanent magnet synchronous motor based on data pre-screening includes establishing a voltage-current discrete equation of the permanent magnet synchronous motor, data screening, and inductance parameter identification.

[0048] Specifically, the method for identifying the inductance parameters of a permanent magnet synchronous motor includes the following steps.

[0049] Step 1: Establish the voltage-current discrete equation of the permanent magnet synchronous motor.

[0050] The established voltage-current discrete equation of the permanent magnet synchronous motor is:

[0051] (1);

[0052] ,

[0053] ,

[0054] ,

[0055] Where, are parameter matrices, for k Stator sampling current at time +1; for k The stator sampling current at time , , for k time d Stator sampling current on the shaft, for k time q Stator sampling current on the shaft; for k The input voltage at the moment, , for k time d Input voltage on the axis, for k time q Input voltage on the shaft; for k Electrical angular velocity at a moment; for k The sampling noise at time , , for k time d Sampling noise on the axis, for k time q Sampling noise on the axis.

[0056] In formula (1), To control the cycle time, is the stator resistance; is a 2D rotation matrix and ; is the inductance matrix, , and They correspond to the inductance on the d / q axis respectively, is the permanent magnet flux; is the 2nd-order identity matrix and .

[0057] It should be noted that when the permanent magnet synchronous motor is running stably, the current change is mainly determined by the input voltage vector of the permanent magnet synchronous motor, which is , where the voltage vector is the above k The input voltage at the moment; it is worth mentioning that the input voltage vector of the permanent magnet synchronous motor is the output voltage of the inner loop controller described below.

[0058] Step 2: Data pre-screening. According to the voltage-current discrete equation in step 1, the change of current in steady state mainly depends on the voltage vector. Based on this, a voltage screening model is constructed. The screening data is obtained according to the voltage screening model. The screening data are the output voltage at the current moment ( ), current sampling current ( ) and the next moment sampling current ( ) and the current speed information ( ), and input it into the parameter identification.

[0059] To expand on this, the constructed voltage screening model is:

[0060] (2),

[0061] (3),

[0062] (4),

[0063] Where, is the decision function, H for Heaviside step function, is the weight factor, is the threshold factor, for The cumulative function of for M Function in k The value of the moment, for M Function in k -1 moment value, for Cumulative function of changing values, for V Function in k The value of the moment, For V function k- The value at moment 1, for k The square of the voltage vector at time .

[0064] Furthermore, according to the voltage screening model, ,exist >0, obtain the current output voltage, the current sampling current and the next sampling current and the current speed information, and input them into the parameter identification in step 3; in this embodiment, the current voltage is , the current sampling current at the current moment is and the next , the current speed information is .

[0065] It should be noted that, in this embodiment, the voltage screening model does not involve the calculation of voltage and current, and formula (2) is based on the conventional Heaviside Step function The measurement of the change amount, formula (3) mainly uses the exponential forgetting method to accumulate the voltage square value, and formula (4) is to calculate the square value of the voltage vector at the current moment (i.e. ) compared to the previous moment.

[0066] Furthermore, the above voltage screening model is derived through theoretical basis, as described below.

[0067] In order to avoid the influence of voltage sign, that is, the above formula (1) explains that voltage as the input of the motor system is the fundamental reason for the change of current and speed, so voltage is selected as the basis for data pre-screening. Therefore, in order to avoid the influence of the positive and negative signs of the voltage input vector on the screening process, the square of the voltage input vector (that is, ) as the evaluation object, in order to reasonably evaluate the strength of the input, select The spectral density is used as the evaluation index, where the voltage vector is , the spectral density equation is:

[0068] (5), where is the spectral density, for The autocorrelation equation of is the sampling time interval, is the frequency, is a mathematical constant, is the square value of the voltage vector.

[0069] in, The estimated equation is: (6), where L is the observation window length of the spectral density function, k To observe the time, The voltage vector is The square of the moment.

[0070] However, the computational complexity of online estimation of the spectral density in the observation window using formula (5) and formula (6) is too large, and the computational complexity is still significant even at a smaller observation window length. However, using the square of the voltage vector will increase the correlation between the signals, but the square of the voltage vector can be Roughly regarded as a random signal (white noise signal), the covariance can be used as the judgment criterion, that is, the spectral density is approximately equal to the covariance, where for The covariance of , so Formula 6 is readjusted to; (7), where is the expected symbol, is the expected value of the voltage vector, that is, Obtain the expected value; thus, formula (7) is used as the theoretical basis for the evaluation index.

[0071] It can be seen that due to the limitation of the computational complexity of the control system in the permanent magnet synchronous motor, a long observation window cannot be maintained, so the above-mentioned voltage screening model is established.

[0072] Step 3, parameter identification, establish an objective function based on the voltage-current discrete equation, substitute the screened data into the objective function and the equality constraints in the equality constraints respectively, update the stator sampling current and the stator estimated current in the objective function according to the screened data and the equality constraints, and substitute the equality constraints substituted into the objective function to obtain an unconstrained minimization problem model, and then use the Gauss-Newton method to obtain the solution equation, and then obtain the parameter matrix correction formula converted into a permanent magnet synchronous motor according to the solution equation, and obtain the inductance parameter estimation value according to the calculation formula of the inductance parameter.

[0073] To expand on this, we first establish an objective function with constraints, specifically:

[0074] (8),

[0075] ,

[0076] Where, To minimize the stator current of the objective function, is the stator current; j is the time in the observation window; In the observation window j Stator sampling current at time ; In the observation window j Estimated stator current at time t; It is a manually set observation window length parameter, and its value is set manually according to actual conditions; Corresponding to existk The value of the moment, for k Estimated stator current at time +1, is the maximum allowable current, All equations that follow are constraints that need to be satisfied in the process of minimizing the objective function; among them, j and k The relationship between them is: ≤ j ≤ k , and from Start traversing until k End, here .

[0077] It is worth mentioning that the above objective function discards the information outside the observation window and is mainly used to minimize the current prediction error within the observation window. In the current control process, the non-equality constraints in the constraints can be ignored, so only the equality constraints in formula (8) are valid.

[0078] In this way, the filtered data in step 2 are substituted into the objective function and the equality constraints in the constraint conditions respectively, and the stator sampling current and stator estimated current in the objective function are updated according to the filtered data and the equality constraints. At the same time, the equality constraints substituted into the filtered data are substituted into the objective function. To expand on this, the next step is Substitute into the objective function, and the current voltage , current sampling current and current speed information Substitute them into the equality constraints in the constraints and calculate them. The estimated stator current at time +1 (i.e. ), and then substitute the estimated stator current into the objective function to obtain the unconstrained minimization problem model, which is as follows:

[0079] (9), where

[0080] ,

[0081] ,

[0082] ,

[0083] ,

[0084] ,

[0085] ,

[0086] ,

[0087] Where, for +1 A large matrix consisting of the sampled currents within the observation window at any moment, for k A large matrix consisting of the sampled currents within the observation window at any moment, for k Parameter matrix within the time observation window A large matrix composed of for Parameter matrix within the time observation window A large matrix composed of for Parameter matrix within the time observation window A large matrix composed of for k Voltage vector within the observation window at any moment A large matrix composed of for Electrical angular velocity within the observation window Large vector composition.

[0088] It should be noted that, in the current control process, the non-equality constraints in the above constraints can be ignored in the actual processing process. Moreover, it can be seen in actual operation that the non-equality constraints here are ignored and do not affect the estimation of the inductance parameters in this embodiment.

[0089] Furthermore, for the above unconstrained minimization problem model, the Gauss-Newton method is used to gradually optimize.

[0090] It solves the equation: (10).

[0091] Where, for Any parameter matrix in is the corresponding parameter matrix k +1 moment value, is the Jacobian matrix of the P matrix, for The transpose of .

[0092] Among them, in a shorter time period, such as several adjacent control cycles, It can be considered to be unchanged, so formula (10) is adjusted to the following expression to obtain the parameter matrix correction formula, which is:

[0093] (11),

[0094] Where, In the observation window j The current vector difference at the moment, In the observation window j time d Shaft current difference, In the observation window j time q Shaft current difference, In the observation window j time The value of any parameter matrix in .

[0095] In formula (11), , , , where In the observation window j time d Shaft current sampling value, In the observation window j time d Estimated shaft current, In the observation window j time q Shaft current sampling value, In the observation window j Estimated value of the q-axis current at time t.

[0096] Then, the correction of the parameter matrix B is obtained using formula (11), and then the estimated value of the inductance parameter is estimated according to the estimation formula of the inductance parameter, which is: (12), where To estimate the inductance, To estimate the parameter matrix B The inverse matrix of .

[0097] To expand on this, the above estimation formula is obtained by using the formula (1) Inversely, we can get the corrected parameter matrix B by using formula (11), and then we can get the inverse of the parameter matrix B. .

[0098] In this embodiment, if Figure 1 As shown, the control system of the permanent magnet synchronous motor includes an inner loop controller and an inverter. The use of a current sensor to sample the phase current of the motor and the control voltage output by the inner loop controller are conventional operations in the field of motor control and will not be described in detail.

[0099] Furthermore, in this embodiment, the control system can combine the inductance parameter estimate output in this embodiment with the torque feedback value (i.e., the actual torque value) and torque reference value of the permanent magnet synchronous motor, and process the input variables in a conventional manner to calculate an optimal current reference value. This optimal current reference value can be output to the inner-loop controller for processing, thereby outputting a switching signal to the inverter. This operation is conventional and will not be described in detail. In this embodiment, the inner-loop controller uses a conventional controller. This inner-loop control can implement classic field-oriented control and can implement standard conventional processing in the field of motor control, so it will not be described in detail.

[0100] It should be noted that conventional operations can be used to obtain the above-mentioned optimal current reference value, such as using a calibration method, that is, using the calibrated parameters to calculate the optimal reference current, or using a table lookup method, that is, directly looking up the table output according to the motor operating status and target torque; preferably, in this embodiment, the optimal current reference value can also be obtained by using a method mentioned in a current prediction control method based on a permanent magnet synchronous motor applied by the applicant.

[0101] In addition, in this embodiment, the processing of the inner loop controller and the acquisition of the torque feedback value (ie, the actual torque value) and the torque reference value of the permanent magnet synchronous motor are conventional operations in the field of motor control, and thus will not be described in detail.

[0102] The present invention discloses a method for identifying the inductance parameters of a permanent magnet synchronous motor based on data pre-screening. The present invention introduces data pre-screening on the basis of a traditional online identification method, uses the covariance information of the square change value of the input voltage as an evaluation index, and screens out the data segments with the greatest identification value, effectively avoiding the interference of invalid or low-excitation data on the identification accuracy. At the same time, the parameter identification in the rolling time domain updates the identification data set through a sliding observation window, ensuring that the model parameters are synchronized with the operating state of the permanent magnet synchronous motor, thereby adapting to the dynamic drift of the parameters of the permanent magnet synchronous motor with factors such as magnetic saturation and operating point. Compared with the prior art, the present invention realizes continuous observation and dynamic tracking of the inductance parameters of the permanent magnet synchronous motor, provides high-precision inductance parameter information for subsequent control algorithms such as current prediction, maximum torque control, flux observation and fault detection, and significantly improves the dynamic response performance, steady-state control accuracy and robustness of the control system of the permanent magnet synchronous motor.

[0103] In addition, the present invention improves identification accuracy and convergence speed, enhances noise robustness and algorithm stability, reduces control system changes and computational complexity, achieves full-operating-condition and adaptive parameter tracking, and improves engineering applicability and maintenance convenience.

[0104] To elaborate, in terms of improving the identification accuracy and convergence speed, the sampling data is pre-screened by real-time calculation of the input-output change, covariance evolution or information gain index, and only high-information data is retained for estimation, which significantly reduces the estimation variance and accelerates the convergence of inductance.

[0105] In terms of enhancing noise robustness and algorithm stability, the pre-screening process naturally suppresses low signal-to-noise ratio samples, and combines the rolling time domain sliding window to smoothly update the anomaly estimate to avoid sudden jumps or drifts.

[0106] It reduces the changes to the control system and the amount of calculation. There is no need to inject high-frequency signals or modify hardware, and it only relies on existing voltage and current sampling. The identification core adopts a recursive form, and the amount of calculation is the same as that of standard least squares, and can be directly deployed in the MCU / DSP of common control systems.

[0107] In terms of achieving full working conditions and adaptive parameter tracking, the sliding observation window slides in real time with the working conditions, and can continuously capture the dynamic changes in inductance caused by temperature rise, magnetic saturation and load changes, providing the latest parameter support for conventional MPC, MTPA, FW / MTV and other control algorithms.

[0108] In terms of improving engineering applicability and maintenance convenience, this method is highly versatile and applicable to control systems with different power levels and different sampling frequencies. It does not require shutdown for recalibration, significantly reducing operation and maintenance costs.

[0109] The above description is only a preferred embodiment of this embodiment, and all equivalent changes and modifications made within the scope of the claims of the present invention should fall within the scope of the claims of the present invention.

Claims

1. A method for identifying inductance parameters of a permanent magnet synchronous motor based on data pre-screening, characterized in that: The steps include: Step 1: Establish the voltage-current discrete equation of the permanent magnet synchronous motor as follows: （1）， ， ， ， Where, are parameter matrices, for k Stator sampling current at time +1; for k The stator sampling current at time , , for k time d Stator sampling current on the shaft, for k time q Stator sampling current on the shaft; for k The input voltage at the moment, , for k time d Input voltage on the axis, for k time q Input voltage on the shaft; for k Electrical angular velocity at a moment; for k The sampling noise at time , , for k time d Sampling noise on the axis, for k time q Sampling noise on the axis; To control the cycle time, is the stator resistance; is a 2D rotation matrix and ; is the inductance matrix, ,in and They correspond to the inductance on the d / q axis respectively, is the permanent magnet flux; is the 2nd-order identity matrix and ; Step 2: Data screening. According to the voltage-current discrete equation in step 1, the change in current in steady state mainly depends on the input voltage vector of the permanent magnet synchronous motor. Based on this, a voltage screening model is established. Screening data is obtained according to the voltage screening model. The screening data are the output voltage at the current moment, the stator sampling current at the current moment, the stator sampling current at the next moment, and the speed information at the current moment, and are input into the parameter identification. Step 3, parameter identification, establish an objective function based on the voltage-current discrete equation, substitute the filtered data into the objective function and the equality constraints in the equality constraints respectively, update the stator sampling current and the stator estimated current in the objective function according to the filtered data and the equality constraints, and substitute the equality constraints substituted into the objective function to obtain an unconstrained minimization problem model, then use the Gauss-Newton method to obtain the solution equation, and then convert it into the parameter matrix correction formula of the permanent magnet synchronous motor, and then obtain the inductance parameter estimation value according to the calculation formula of the inductance parameter.

2. The method for identifying inductance parameters of a permanent magnet synchronous motor based on data pre-screening according to claim 1 is characterized in that: The voltage screening model established is as follows, （2）， （3）， （4）， Where, is the decision function, H for Heaviside step function, is the weight factor, is the threshold factor, for The cumulative function of for M Function in k The value of the moment, for M Function in k -1 moment value, for Cumulative function of changing values, for V Function in k The value of the moment, For V function k- The value at moment 1, for k The square of the voltage vector at the moment; Among them, >0, k Voltage vector at each moment 、 k Current sampling at all times and Current sampling at all times And speed information , input into the parameter identification.

3. The method for identifying inductance parameters of a permanent magnet synchronous motor based on data pre-screening according to claim 2, characterized in that: In the voltage screening model, the square of the voltage vector is used as the screening target, and its spectral density equation is: (5), where for The autocorrelation equation of is the sampling time interval, is the frequency, is a mathematical constant, is the square value of the voltage vector; in, The estimated equation is (6), where L is the observation window length of the spectral density function, k To observe the time, The voltage vector is The square value of the moment; then, use The covariance of The estimated equation is adjusted to (7), where for The covariance of is the expected symbol, is the expected value of the voltage vector; thus, formula (7) is used as the theoretical basis for constructing the voltage screening model.

4. The method for identifying inductance parameters of a permanent magnet synchronous motor based on data pre-screening according to claim 2, characterized in that: In step 3, the objective function with constraints is established as follows, （8）， ， Where, To minimize the stator current of the objective function, is the stator sampling current, j is the time in the observation window, In the observation window j The stator sampling current at time , In the observation window j The estimated stator current at time , is the artificially designed observation window length parameter, Corresponding to exist k The value of the moment, for k Estimated stator current at time +1, is the maximum allowable current, where All equations that follow are constraints that need to be satisfied in the process of minimizing the objective function.

5. The method for identifying inductance parameters of a permanent magnet synchronous motor based on data pre-screening according to claim 4 is characterized in that: In step 2, the stator sampling current at the next moment is substituted into the objective function, and the current output voltage, the current stator sampling current and the current speed information are substituted into the equality constraints in the constraint conditions for calculation, and the result is calculated. k The estimated stator current at time +1, then k Substituting the estimated stator current at time +1 into the objective function, the unconstrained minimization problem model is obtained as follows: (9), where ， ， ， ， ， ， ， Where, for k A large matrix consisting of the sampled currents in the observation window at time +1, for k A large matrix consisting of the sampled currents within the observation window at any moment, for k Parameter matrix within the time observation window A large matrix composed of for k Parameter matrix within the time observation window A large matrix composed of for k Parameter matrix within the time observation window A large matrix composed of for k Voltage vector within the observation window at any moment A large matrix composed of for k Electrical angular velocity within the observation window A large vector of The Gauss-Newton method is used to solve the unconstrained minimization problem model, and the solution equation is: (10), where for Any parameter matrix in is the corresponding parameter matrix k +1 moment value, is the Jacobian matrix of formula (9), for The transpose of In several adjacent control cycles unchanged, so formula (10) is adjusted to the following expression to obtain the parameter matrix correction formula, which is: （11）， Where, In the observation window j The current vector difference at the moment, In the observation window j time d Shaft current difference, In the observation window j time q Shaft current difference, In the observation window j time The value of any parameter matrix in ; where, , , , where In the observation window j time d Shaft current sampling value, In the observation window j time d Estimated shaft current, In the observation window j time q Shaft current sampling value, In the observation window j Estimated value of the q-axis current at time t.

6. The method for identifying inductance parameters of a permanent magnet synchronous motor based on data pre-screening according to claim 5, characterized in that: Formula (11) is used to obtain the correction of the parameter matrix B, and then the estimated value of the inductance parameter is estimated according to the estimation formula of the inductance parameter, which is: (12), where To estimate the inductance, To estimate the inverse matrix of the parameter matrix B.

7. The method for identifying inductance parameters of a permanent magnet synchronous motor based on data pre-screening according to claim 6, characterized in that: The estimation formula is obtained by using the formula (1) Reverse deduction to obtain.

Images