Data-driven predictive current modeling method, device, medium and equipment for permanent magnet synchronous motor

Abstract

Embodiments of the present application disclose a data-driven predictive current modeling method, device, medium and equipment for a permanent magnet synchronous motor, relating to the field of converter technology, aiming to solve the problem that the predictive current model in the prior art is susceptible to noise interference, resulting in a reduction in prediction accuracy. First, considering the influence of electromagnetic cross-saturation effect in actual situations, the present application obtains a data-driven current prediction model for the motor. Secondly, a ridge regression-based calculation model for the parameter vector in the model is constructed through historical data. Then, by evaluating the severity of the noise in the data, a dynamic adjustment of the ridge parameter is established to achieve adaptive punishment for the data noise, determine the target solution of the penalty term, and further back-calculate the parameter vector to obtain the final prediction model, thereby improving the current prediction accuracy under noise conditions.

Description

Technical Field

[0001] The present application relates to the technical field of converters, and particularly relates to a data-driven predictive current modeling method, device, medium and equipment for a permanent magnet synchronous motor. Background Art

[0002] Model Predictive Control (MPC) has received extensive attention in the field of motor drive due to its advantages such as fast dynamic response and strong control flexibility, and is a promising alternative to traditional control methods. Traditional MPC predicts the system behavior based on the motor model, which means that the controller requires accurate motor parameters to achieve satisfactory control performance. However, under actual working conditions, the key motor parameters are highly sensitive, and currently, data-driven technologies have received increasing attention in solving the problem of MPC parameter dependence.

[0003] Among them, the least squares method has been proven to be a potential alternative to traditional MPC. The basic principle of the least squares method is to find the optimal estimate of the model parameters by minimizing the sum of the squares of the errors between the predicted values and the actual values. Its core idea is to make the prediction results of the model match the observed data as much as possible. Therefore, this data-based scheme can identify a more accurate current prediction model in real time, thereby improving the predictive control performance. Due to the potential noise in current sensor measurements, the dependence of the least squares method on the system input-output data poses a challenge. The predictive current model established by the least squares method under noise interference is vulnerable to interference, resulting in poor predictive current accuracy and thus affecting the control performance. Summary of the Invention

[0004] The main purpose of the present application is to provide a data-driven predictive current modeling method, device, medium and equipment for a permanent magnet synchronous motor, aiming to solve the problem that the predictive current model in the prior art is vulnerable to noise interference, resulting in reduced prediction accuracy.

[0005] To achieve the above object, the technical solutions adopted in the embodiments of the present application are as follows:

[0006] In a first aspect, an embodiment of the present application provides a data-driven predictive current modeling method for a permanent magnet synchronous motor, including the following steps:

[0007] Obtain a data-driven current prediction model of the target motor based on the cross-saturation effect; wherein, the data-driven current prediction model includes a parameter vector;

[0008] According to the historical data collected in the historical control cycle, establish a calculation model of the parameter vector based on ridge regression; wherein, the calculation model includes a penalty term;

[0009] Determine the target solution of the penalty term according to the severity of the noise in the historical data;

[0010] Obtain the target penalty term and the target parameter vector according to the target solution, so as to establish a target data-driven current prediction model.

[0011] In a possible implementation manner of the first aspect, before determining the target solution of the penalty term according to the noise severity of the historical data, the method further includes:

[0012] Construct matrices for the data corresponding to the input data and the output data in the historical data respectively;

[0013] Obtain the sum of squared residuals and the total sum of squares according to the matrices;

[0014] Obtain the multicollinearity of the input data according to the sum of squared residuals and the total sum of squares, so as to characterize the noise severity of the historical data.

[0015] In a possible implementation manner of the first aspect, determining the target solution of the penalty term according to the noise severity of the historical data includes:

[0016] Obtain the relationship between the multicollinearity and the penalty term according to the positive correlation between the multicollinearity and the noise severity of the historical data;

[0017] Map the relationship between the multicollinearity and the penalty term based on the mapping function, and determine the target solution of the penalty term.

[0018] In a possible implementation manner of the first aspect, obtaining the multicollinearity of the input data according to the sum of squared residuals and the total sum of squares, so as to characterize the noise severity of the historical data, includes:

[0019] Obtain the variance inflation factor of the input data according to the sum of squared residuals and the total sum of squares to measure the multicollinearity of the input data, so as to characterize the noise severity of the historical data.

[0020] In a possible implementation manner of the first aspect, before obtaining the data-driven current prediction model based on the cross-saturation effect of the target motor, the method further includes:

[0021] Obtain the input data and the output data respectively according to the stator voltage and the stator current in the current control of the target motor;

[0022] Establish a data-driven current prediction model based on the cross-saturation effect of the target motor according to the input data, the output data, and the parameter vector.

[0023] In a possible implementation manner of the first aspect, establishing a calculation model of the parameter vector based on ridge regression according to the historical data collected in the historical control period includes:

[0024] Obtain the historical input data and the historical output data according to the historical data collected in the historical control period;

[0025] Construct a past subset and a future subset respectively according to historical input data and historical output data;

[0026] Establish a calculation model of the parameter vector based on ridge regression according to the past subset and the future subset.

[0027] In a possible implementation manner of the first aspect, establishing a calculation model of the parameter vector based on ridge regression according to the past subset and the future subset includes:

[0028] Estimate the parameter vector by applying the solution of the least squares method according to the past subset and the future subset, so as to establish a calculation model of the parameter vector based on ridge regression.

[0029] In a second aspect, an embodiment of the present application provides a permanent magnet synchronous motor data-driven predictive current modeling device, including:

[0030] An acquisition module, which is used to acquire a data-driven current prediction model of a target motor based on cross-saturation effect; wherein, the data-driven current prediction model includes a parameter vector;

[0031] A building module, which is used to establish a calculation model of the parameter vector based on ridge regression according to historical data collected in a historical control period; wherein, the calculation model includes a penalty term;

[0032] A determination module, which is used to determine a target solution of the penalty term according to the noise severity of historical data;

[0033] A target module, which is used to obtain a target penalty term and a target parameter vector according to the target solution, so as to establish a target data-driven current prediction model.

[0034] In a third aspect, an embodiment of the present application provides a computer-readable storage medium, storing a computer program, which when loaded and executed by a processor, implements the permanent magnet synchronous motor data-driven predictive current modeling method provided in any one of the above first aspects.

[0035] In a fourth aspect, an embodiment of the present application provides an electronic device, including a processor and a memory, wherein,

[0036] The memory is used to store a computer program;

[0037] The processor is used to load and execute the computer program, so that the electronic device executes the permanent magnet synchronous motor data-driven predictive current modeling method provided in any one of the above first aspects.

[0038] Compared with the prior art, the beneficial effects of the present application are:

[0039] A method, device, medium, and equipment for data-driven predictive current modeling of a permanent magnet synchronous motor proposed in an embodiment of the present application. The method includes: obtaining a data-driven current prediction model of a target motor based on cross-saturation effect; wherein, the data-driven current prediction model includes a parameter vector; establishing a calculation model of the parameter vector based on ridge regression according to historical data collected in a historical control period; wherein, the calculation model includes a penalty term; determining a target solution of the penalty term according to the severity of noise in the historical data; obtaining a target penalty term and a target parameter vector according to the target solution to establish a target data-driven current prediction model. The present application first takes into account the influence of electromagnetic cross-saturation effect in actual situations to obtain a data-driven current prediction model of the motor. Secondly, a calculation model of the parameter vector in the model based on ridge regression is constructed through historical data. Then, by evaluating the severity of noise in the data, dynamic adjustment of the ridge parameter is established to achieve adaptive penalty for data noise, determining the target solution of the penalty term and further back-calculating the parameter vector to obtain the final prediction model, thereby improving the current prediction accuracy under noise conditions. Description of the Drawings

[0040] Figure 1 Schematic diagram of least squares overfitting caused by noise;

[0041] Figure 2 Fitting schematic diagram of ridge regression;

[0042] Figure 3 Schematic diagram of the optimal solution space of ridge regression;

[0043] Figure 4 Schematic diagram of the structure of an electronic device for the hardware operating environment involved in an embodiment of the present application;

[0044] Figure 5 Schematic flow chart of the data-driven predictive current modeling method for a permanent magnet synchronous motor provided by an embodiment of the present application;

[0045] Figure 6 Comparison schematic diagram of dq-axis predictive current, dq-axis actual current, and dq-axis predictive current residual waveforms based on the least squares method;

[0046] Figure 7 Comparison schematic diagram of dq-axis predictive current, dq-axis actual current, and dq-axis predictive current residual waveforms of the data-driven predictive current modeling method for a permanent magnet synchronous motor provided by an embodiment of the present application;

[0047] Figure 8 Comparison schematic diagram of speed, dq-axis actual current, and a-phase current waveforms based on the least squares method;

[0048] Figure 9Schematic diagram for comparing the rotational speed, actual dq-axis current, and a-phase current waveform of the permanent magnet synchronous motor data-driven predictive current modeling method provided by the embodiments of the present application;

[0049] Figure 10 Module schematic diagram of the permanent magnet synchronous motor data-driven predictive current modeling device provided by the embodiments of the present application;

[0050] Markings in the figure: 101 - Processor, 102 - Communication bus, 103 - Network interface, 104 - User interface, 105 - Memory. Specific embodiments

[0051] It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0052] The main solution provided by the embodiments of the present application is: obtaining a data-driven current prediction model of a target motor based on cross-saturation effects; wherein, the data-driven current prediction model includes a parameter vector; establishing a calculation model of the parameter vector based on ridge regression according to historical data collected in historical control cycles; wherein, the calculation model includes a penalty term; determining the target solution of the penalty term according to the severity of the noise in the historical data; obtaining a target penalty term and a target parameter vector according to the target solution to establish a target data-driven current prediction model.

[0053] Model predictive control (MPC) has received extensive attention in the field of motor drive due to its advantages such as fast dynamic response and strong control flexibility, and is a promising alternative to traditional control methods (such as field-oriented control and direct torque control). Traditional MPC predicts the system behavior based on the motor model, which means that the controller requires accurate motor parameters to achieve satisfactory control performance. However, under actual working conditions, key motor parameters are highly sensitive. For example, the resistance of a permanent magnet synchronous motor (PMSM) may change due to temperature changes, and the value of the inductance may also be affected by magnet saturation and cross-saturation effects. In practical applications, such interference brings difficulties to the accurate modeling of the motor and the acquisition of accurate parameters. Therefore, the traditional MPC considering fixed motor parameters has a problem of parameter mismatch, resulting in performance degradation.

[0054] In the context of digitization, networking, and intelligence as the core, data-driven technologies have attracted wide attention due to the concept of deeply integrating new-generation information technologies and mathematical models, providing effective solutions for the modeling of strongly coupled time-varying systems. Among them, the least squares technique is widely used in system identification and intelligent control to predict and explain the relationships between variables. Generally speaking, different solutions have been proposed to improve the robustness of MPC parameters for PMSM drives. Among them, the least squares (LS) method has been proven to be a potential alternative to traditional MPC. The basic principle of the least squares method is to find the optimal estimate of the model parameters by minimizing the sum of the squares of the errors between the predicted values and the actual values. Its core idea is to make the prediction results of the model match the observed data as much as possible. Therefore, this data-based solution can identify a more accurate current prediction model in real time, thereby improving the prediction control performance.

[0055] The anti-interference performance is a key issue faced by data-driven technologies. Due to the potential noise in current sensor measurements, the dependence of the least squares method on system input-output data poses a challenge. The sources of noise are diverse, including manufacturing errors, temperature changes, and sensor aging. On the other hand, the electromagnetic field in the motor operating environment will interfere with the signal transmission, resulting in noise in the sampled signals. These interferences may come from the electromagnetic emissions of the motor itself or from electromagnetic interference caused by other devices. In many industrial applications, motors often operate in an environment with obvious electromagnetic noise, and the electromagnetic noise is easily coupled with the sampled signals, resulting in continuous measurement errors and affecting the model performance. Therefore, in order to ensure the robustness of the control system, the uncertainty of the observed data in the least squares method must be considered.

[0056] The schematic diagram of the least squares overfitting caused by noise is shown in the appendix Figure 1 As shown, the embodiment of the present application applies the ridge regression technique to the field of permanent magnet synchronous motor drives. As shown in the appendix Figure 2 The fitting schematic diagram of ridge regression is shown in the appendix Figure 3 The schematic diagram of the optimal solution space of ridge regression is shown in the appendix. The severity of the noise is evaluated and the dynamic ridge coefficient adjustment is applied to achieve adaptive punishment for data noise, thereby improving the robustness.

[0057] Therefore, the present application provides a means. First, considering the influence of electromagnetic cross-saturation effects in actual situations, a data-driven current prediction model for the motor is obtained. Secondly, a ridge regression-based calculation model for the parameter vector in the model is constructed through historical data. Then, by evaluating the severity of the noise in the data, a dynamic adjustment of the ridge parameter is established to achieve adaptive punishment for data noise, determine the target solution of the penalty term, and further back-calculate the parameter vector to obtain the final prediction model, thereby improving the current prediction accuracy under noise conditions.

[0058] Refer to the appendix Figure 4 , the appendix Figure 4 is a schematic structural diagram of an electronic device for the hardware operating environment involved in the solution of the embodiment of the present application. The electronic device may include: a processor 101, such as a central processing unit (CPU), a communication bus 102, a user interface 104, a network interface 103, and a memory 105. Among them, the communication bus 102 is used to realize the connection and communication between these components. The user interface 104 may include a display screen (Display), an input unit such as a keyboard (Keyboard). Optionally, the user interface 104 may further include a standard wired interface and a wireless interface. The network interface 103 may optionally include a standard wired interface and a wireless interface (such as a wireless fidelity (WI-FI) interface). The memory 105 may optionally be a storage device independent of the aforementioned processor 101. The memory 105 may be a high-speed random access memory (Random Access Memory, RAM) memory, or may be a stable non-volatile memory (Non-Volatile Memory, NVM), such as at least one disk memory; the processor 101 may be a general-purpose processor, including a central processor, a network processor, etc., or may also be a digital signal processor, an application-specific integrated circuit, a field programmable gate array, or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0059] Those skilled in the art can understand that the structure shown in the appendix Figure 4 does not constitute a limitation on the electronic device, and may include more or fewer components than shown in the figure, or combine certain components, or arrange different components.

[0060] As shown in the appendix Figure 4 , the memory 105 as a storage medium may include an operating system, a network communication module, a user interface module, and a permanent magnet synchronous motor data-driven predictive current modeling device.

[0061] In the electronic device shown in the appendix Figure 4 , the network interface 103 is mainly used for data communication with a network server; the user interface 104 is mainly used for data interaction with a user; the processor 101 and the memory 105 in the present application may be provided in the electronic device. The electronic device calls the permanent magnet synchronous motor data-driven predictive current modeling device stored in the memory 105 through the processor 101, and executes the permanent magnet synchronous motor data-driven predictive current modeling method provided by the embodiment of the present application.

[0062] Refer to the appendix Figure 5, based on the hardware device of the foregoing embodiments, an embodiment of the present application provides a data-driven predictive current modeling method for a permanent magnet synchronous motor, including the following steps:

[0063] S10: Obtain a data-driven current prediction model of the target motor based on the cross-saturation effect; wherein, the data-driven current prediction model includes a parameter vector.

[0064] In the specific implementation process, the target motor is the motor for which a data-driven current prediction model needs to be established, which is a permanent magnet synchronous motor. Cross-saturation means that the magnetic flux distribution in the motor magnetic circuit is uneven, resulting in some magnetic flux lines reaching the magnetic saturation state simultaneously while other lines do not reach the saturation state. Data-driven refers to the process of using the collected, analyzed, and interpreted data to guide decision-making, optimize processes, and achieve goals.

[0065] In one embodiment, before obtaining the data-driven current prediction model of the target motor based on the cross-saturation effect, the method further includes:

[0066] Obtain input data and output data respectively according to the stator voltage and stator current in the current control of the target motor;

[0067] Establish a data-driven current prediction model of the target motor based on the cross-saturation effect according to the input data, output data, and parameter vector.

[0068] In the specific implementation process, in the current control of the permanent magnet synchronous motor, the stator voltage and stator current are the input and output of the system respectively. The state space considering the cross-saturation effect is expressed as:

[0069] y(k + 1) = θx(k + 1)

[0070]

[0071] where k is the sampling time, x is the regression matrix, u = [u d , u q T is the system input voltage, y = [i d , i q T is the system output current, θ is the parameter vector, and a, b, and e are all elements in the parameter vector, corresponding to the coefficients of the current terms and voltage terms and the constant term under the dq axis respectively.

[0072] S20: Establish a calculation model of the parameter vector based on ridge regression according to the historical data collected in the historical control period; wherein, the calculation model includes a penalty term.

[0073] In the specific implementation process, the parameter vector is estimated based on historical data collected over multiple past control cycles. In the predictive current control environment of the PMSM drive, the dq-axis voltages applied within each control cycle are determined by the controller and are thus stored as input data. On the other hand, the dq-axis currents are sampled at the beginning of each control cycle and stored as output data. That is, the historical data is divided into historical input data and historical output data. Based on the historical data collected in historical control cycles, a calculation model for the parameter vector based on ridge regression is established, including:

[0074] Obtain historical input data and historical output data according to the historical data collected in historical control cycles;

[0075] Construct past subsets and future subsets respectively according to the historical input data and historical output data;

[0076] Establish a calculation model for the parameter vector based on ridge regression according to the past subsets and future subsets.

[0077] In the specific implementation process, the data sets of historical input data and historical output data are respectively denoted as U ds and Y ds , and a set of subsets can be constructed to define two past subsets and future subsets for the input and output data:

[0078] U p =[u1 u2... u N-1

[0079] U f =[u2 u3... u N

[0080] Y p =[y1 y2... y N-1

[0081] Y f =[y2 y3... y N

[0082] The subscripts p and f represent past and future respectively. U p is the past subset of historical input data, U f is the future subset of historical input data, Y p is the past subset of historical output data, Y f is the future subset of historical output data. The number of columns of these matrices is N - 1. At the sampling moment k, according to the collected data, the calculation formula for the ridge regression parameter vector θ can be established, that is, the calculation model:

[0083] θ=Y f H T (HH​​​​T + λI) -1

[0084] Where:

[0085]

[0086] I is the identity matrix, and H T is the transpose of H. If HH T is a singular matrix, then the added λI term can ensure that HH T + λI is full rank, making the matrix invertible and simultaneously suppressing the influence of noise.

[0087] In one embodiment, a calculation model of the parameter vector based on ridge regression is established according to the past subset and the future subset, including:

[0088] Estimate the parameter vector by applying the solution of the least squares method according to the past subset and the future subset, so as to establish a calculation model of the parameter vector based on ridge regression.

[0089] In the specific implementation process, the constructed past subset and future subset are used to estimate the parameter vector by applying the solution of the least squares method. Generally speaking, the goal of the least squares method is to achieve global fitting optimization, which can give an accurate unbiased estimate in the absence of noise. However, the data is not always ideal in practical applications. In a permanent magnet synchronous motor drive, measurement noise is generated due to current induction, and in some low-speed and low-load cases, this noise effect will be more prominent. In this case, data noise will cause the problem of multicollinearity. The traditional least squares method will lead to unreliable results due to overfitting. Therefore, the sensitivity of the least squares method to noisy data will reduce the current prediction accuracy of the PMSM drive. Ridge regression is introduced to solve the above multicollinearity problem. By adding a regularization term to the loss function, the problem of high correlation between independent variables is solved, making the prediction performance of the model more stable:

[0090] In the traditional LS regression, the loss function J is defined as:

[0091] J(θ) = ||Y f - θH|| 2

[0092] In the ridge regression model, a penalty term of the L2 norm is added to the loss function to solve the multicollinearity problem

[0093] J(θ) = ||Y f - θH|| 2 + λ ||θ 2

[0094] Among them, λ ∈ [0, ∞] is the ridge coefficient for adjusting the penalty. The main idea of ridge regression is to add an additional penalty reduction term to the loss function of traditional LS, and then narrow the optimal solution space. The above formula can be further written as:

[0095] J(θ) = (Y f - θH) T (Y f - θH) + λθ T θ

[0096] = Y f T Y f - Y f T θH - H T θ T Y f + H T θ T θH + λθ T θ

[0097] To minimize J(θ), apply the relationship to find the minimum point, and then the following relationship can be derived:

[0098] 0 - H T Y f - H T Y + 2HH T θ + 2λθ = 0

[0099] The parameter vector under ridge regression can be expressed as:

[0100] θ = Y f H T (HH T + λI) -1

[0101] S30: Determine the target solution of the penalty term according to the noise severity of historical data.

[0102] In the specific implementation process, the final solution of the model is transformed into the final solution of the parameter vector through the established prediction model, and then adjusted according to the noise severity, and the final solution of the model is transformed into the final solution of the penalty term. Specifically: Determine the target solution of the penalty term according to the noise severity of historical data, including:

[0103] Obtain the relationship between multicollinearity and the penalty term according to the positive correlation between multicollinearity and the noise severity of historical data;

[0104] Based on the mapping function, map the relationship between multicollinearity and the penalty term, and determine the target solution of the penalty term.

[0105] In the specific implementation process, a ridge coefficient is introduced to compensate for the multicollinearity caused by data noise. Since multicollinearity is positively correlated with the severity of data noise, the greater the multicollinearity, the larger the λ should be used to enhance the penalty intensity. Therefore, the relationship between the penalty intensity λ and the severity of multicollinearity VIF iq can be mapped by a sigmoid function. Therefore, the optimal λ, that is, the target solution of the penalty term, can be obtained online according to the evaluation result of data noise, and the relational expression is given as:

[0106]

[0107] where e is the natural constant, and α and β are fixed coefficients used to horizontally and vertically stretch the sigmoid function. The optimal values of α and β can be searched through offline cross-validation of experimental data, so as to skip the unwanted penalty range with relatively low severity of multicollinearity and limit the penalty intensity within an appropriate range.

[0108] In one embodiment, before determining the target solution of the penalty term according to the severity of the noise of historical data, the method further includes:

[0109] Construct matrices for the data corresponding to the input data and the output data in the historical data respectively;

[0110] Obtain the sum of squared residuals and the total sum of squares according to the matrices;

[0111] Obtain the multicollinearity of the input data according to the sum of squared residuals and the total sum of squares to characterize the severity of the noise of the historical data.

[0112] In the specific implementation process, in the data-driven current prediction model, elements i d 、i q 、u d and u q in the input voltage and the output current are involved. The data corresponding to each variable are collected into a matrix, and the matrix can be represented as X. Then, the regression coefficient b of i q can be calculated as:

[0113]

[0114] where X iq is the row in X containing the data of i q , X no_iq is the remaining rows not containing the data of i q , and are the matrix transposes of X iq and X no_iq respectively. Furthermore, i qThe correlation coefficient with other variables. The sum of squared residuals SSR can be expressed as:

[0115]

[0116] where is the mean value of the data. The total sum of squares SST can be expressed as:

[0117]

[0118] Therefore, the variance inflation factor VIF of i q can be expressed as

[0119]

[0120] where R 2 = SSR / SST.

[0121] The VIF value represents the severity of multicollinearity related to data noise, that is: based on the sum of squared residuals and the total sum of squares, the multicollinearity of the input data is obtained to characterize the severity of the noise of historical data, including:

[0122] Based on the sum of squared residuals and the total sum of squares, the variance inflation factor of the input data is obtained to measure the multicollinearity of the input data, so as to characterize the severity of the noise of historical data.

[0123] When the VIF exceeds the threshold, the traditional LS regression will be unreliable. Ridge regression is introduced to improve the prediction accuracy. In the embodiments of the present application, the average value of the VIF is used to evaluate the severity of data noise.

[0124] S40: Obtain the target penalty term and the target parameter vector according to the target solution, so as to establish the target data-driven current prediction model.

[0125] In the specific implementation process, according to the foregoing steps, by determining the target solution of the penalty term, the penalty term can be determined and then the target solution of the parameter vector can be inversely deduced, that is, the target parameter vector. Then, the parameter vector in the initially established model is updated to obtain the final target data-driven current model.

[0126] In this embodiment, first, considering the influence of the cross-saturation effect on the electromagnetic in actual situations, a data-driven current prediction model of the motor is obtained. Secondly, a calculation model based on ridge regression of the parameter vector in the model is constructed through historical data. Then, by evaluating the severity of the noise in the data, a dynamic adjustment of the ridge parameter is established to achieve an adaptive penalty for the data noise, determine the target solution of the penalty term, and further inversely deduce the parameter vector to obtain the final prediction model, thereby improving the current prediction accuracy under noise conditions.

[0127] To verify the effectiveness of the method provided by the embodiments of the present application, ridge regression and the least squares method are compared under different noise levels. The comparison of the dq-axis predicted current residuals is shown in Table 1 as follows:

[0128] Table 1 Comparison of the sizes of the predicted residual variances at a rotational speed of 300 rpm

[0129]

[0130] Among them: under different noise levels, the predicted residuals of the method of the present application in the two indexes of rd(A) and rq(A) are significantly smaller than those of the traditional methods (MPC and LS), reflecting higher prediction accuracy and robustness. Further, as data-driven methods, the LS method and the method of the present application can accurately identify the current prediction model online. Compared with the model-based MPC method, they have significant advantages when no additional noise (var = 0) and low noise (var = 0.02) are added. However, as the noise level increases (var = 0.05), due to the negative impact of noise on the prediction model, the predicted residuals of the LS method are greater than those of the MPC method, while the method of the present application can still maintain significant advantages under the same conditions.

[0131] Referring to the attached Figure 6 - attached Figure 9 , for the comparison between the least squares method and the method provided by the embodiments of the present application, where Figure 6 is a schematic diagram of the comparison of the dq-axis predicted current, dq-axis actual current, and dq-axis predicted current residual waveforms based on the least squares method, Figure 7 is a schematic diagram of the comparison of the dq-axis predicted current, dq-axis actual current, and dq-axis predicted current residual waveforms of the permanent magnet synchronous motor data-driven predictive current modeling method provided by the embodiments of the present application, Figure 8 is a schematic diagram of the comparison of the rotational speed, dq-axis actual current, and a-phase current waveforms based on the least squares method, Figure 9 is a schematic diagram of the comparison of the rotational speed, dq-axis actual current, and a-phase current waveforms of the permanent magnet synchronous motor data-driven predictive current modeling method provided by the embodiments of the present application. Among them, Measured represents the actually measured data, and Predicted represents the predicted data. From the comparison of the attached drawings, it can be seen that under the working conditions of noise interference, the method of the present application has better predictive current performance than the traditional least squares modeling, which is reflected in that under the closed-loop control of the predictive current, the current ripple is reduced and the robustness is improved.

[0132] Referring to the attached Figure 10 , based on the same inventive concept as in the foregoing embodiments, the embodiments of the present application further provide a permanent magnet synchronous motor data-driven predictive current modeling device, including:

[0133] An acquisition module, which is used to acquire a data-driven current prediction model of a target motor based on cross-saturation effect; wherein, the data-driven current prediction model includes a parameter vector.

[0134] A building module, which is used to build a calculation model of the parameter vector based on ridge regression according to historical data collected in historical control cycles; wherein, the calculation model includes a penalty term.

[0135] A determination module, which is used to determine a target solution of the penalty term according to the severity of noise in the historical data.

[0136] A target module, which is used to obtain a target penalty term and a target parameter vector according to the target solution, so as to build a target data-driven current prediction model.

[0137] Those skilled in the art should understand that the division of each module in the embodiment is only a logical function division. In actual application, they can be fully or partially integrated into one or more actual carriers, and these modules can all be implemented in the form of software called by a processing unit, or all be implemented in the form of hardware, or be implemented in the form of a combination of software and hardware. It should be noted that each module in the permanent magnet synchronous motor data-driven predictive current modeling device in this embodiment corresponds one by one to each step in the permanent magnet synchronous motor data-driven predictive current modeling method in the foregoing embodiment. Therefore, the specific implementation manner of this embodiment can refer to the implementation manner of the foregoing permanent magnet synchronous motor data-driven predictive current modeling method, and will not be elaborated here.

[0138] Based on the same inventive concept as in the foregoing embodiment, an embodiment of the present application further provides a computer-readable storage medium, storing a computer program, which when loaded and executed by a processor, implements the permanent magnet synchronous motor data-driven predictive current modeling method provided by the embodiment of the present application.

[0139] Based on the same inventive concept as in the foregoing embodiment, an embodiment of the present application further provides an electronic device, including a processor and a memory, wherein,

[0140] The memory is used to store a computer program;

[0141] The processor is used to load and execute the computer program, so that the electronic device executes the permanent magnet synchronous motor data-driven predictive current modeling method provided by the embodiment of the present application.

[0142] In some embodiments, the computer-readable storage medium may be a memory such as FRAM, ROM, PROM, EPROM, EEPROM, flash memory, magnetic surface memory, optical disc, or CD-ROM; or may be various devices including one or any combination of the above memories. The computer may be various computing devices including intelligent terminals and servers.

[0143] In some embodiments, executable instructions may be in the form of a program, software, software module, script, or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including as a stand-alone program or as a module, component, subroutine, or other unit suitable for use in a computing environment.

[0144] As an example, executable instructions may, but need not, correspond to a file in a file system, may be stored as part of a file that stores other programs or data, such as in one or more scripts in a HyperText Markup Language (HTML) document, in a single file dedicated to the program in question, or in multiple coordinating files (e.g., files storing one or more modules, subroutines, or code portions).

[0145] By way of example, executable instructions may be deployed to be executed on one computing device, or on multiple computing devices at one site, or on multiple computing devices distributed across multiple sites and interconnected by a communication network.

[0146] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or system comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or system. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or system comprising the element.

[0147] The serial numbers of the above embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0148] Through the description of the above embodiments, those skilled in the art can clearly understand that the above embodiment methods can be implemented by means of software plus the necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better embodiment. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as a read-only memory / random access memory, a magnetic disk, or an optical disk), and includes a number of instructions for enabling a multimedia terminal device (which can be a mobile phone, a computer, a television receiver, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0149] In summary, a data-driven predictive current modeling method, device, medium, and equipment for a permanent magnet synchronous motor provided by this application. The method includes: obtaining a data-driven current prediction model of the target motor based on the cross-saturation effect; wherein, the data-driven current prediction model includes a parameter vector; establishing a calculation model of the parameter vector based on ridge regression according to historical data collected in historical control cycles; wherein, the calculation model includes a penalty term; determining the target solution of the penalty term according to the severity of the noise in the historical data; obtaining the target penalty term and the target parameter vector according to the target solution to establish the target data-driven current prediction model. This application first considers the influence of the electromagnetic cross-saturation effect in actual situations to obtain the data-driven current prediction model of the motor. Secondly, a calculation model of the parameter vector in the model based on ridge regression is constructed through historical data. Then, by evaluating the severity of the noise in the data, a dynamic adjustment of the ridge parameter is established to achieve adaptive penalty for the data noise, determine the target solution of the penalty term, and further reverse-infer the parameter vector to obtain the final prediction model, thereby improving the current prediction accuracy under noisy conditions.

[0150] The above are only the preferred embodiments of this application and are not intended to limit this application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of this application shall be included within the protection scope of this application.

Claims

1. A data-driven predictive current modeling method for a permanent magnet synchronous motor, characterized in that, Including the following steps: Obtain a data-driven current prediction model of the target motor based on the cross-saturation effect; wherein, the data-driven current prediction model includes a parameter vector; Establish a calculation model of the parameter vector based on ridge regression according to historical data collected in a historical control period; wherein, the calculation model includes a penalty term; Determine the target solution of the penalty term according to the noise severity of the historical data; before determining the target solution of the penalty term according to the noise severity of the historical data, the method further includes: Construct matrices for the data corresponding to the input data and the output data in the historical data respectively; Obtain the sum of squared residuals and the total sum of squares according to the matrices; Obtain the multicollinearity of the input data according to the sum of squared residuals and the total sum of squares to characterize the noise severity of the historical data; The determining the target solution of the penalty term according to the noise severity of the historical data includes: Obtain the relationship between the multicollinearity and the penalty term according to the positive correlation between the multicollinearity and the noise severity of the historical data; Map the relationship between the multicollinearity and the penalty term based on a mapping function and determine the target solution of the penalty term; Obtain a target penalty term and a target parameter vector according to the target solution to establish a target data-driven current prediction model.

2. The data-driven predictive current modeling method for a permanent magnet synchronous motor according to claim 1, characterized in that The obtaining the multicollinearity of the input data according to the sum of squared residuals and the total sum of squares to characterize the noise severity of the historical data includes: Obtain the variance inflation factor of the input data according to the sum of squared residuals and the total sum of squares to measure the multicollinearity of the input data and characterize the noise severity of the historical data.

3. The data-driven predictive current modeling method for a permanent magnet synchronous motor according to claim 1, wherein Before obtaining the data-driven current prediction model of the target motor based on the cross-saturation effect, the method further includes: Obtain input data and output data respectively according to the stator voltage and stator current in the current control of the target motor; Establish a data-driven current prediction model of the target motor based on the cross-saturation effect according to the input data, the output data and the parameter vector.

4. The data-driven predictive current modeling method for a permanent magnet synchronous motor according to claim 1, characterized in that, The establishing a calculation model of the parameter vector based on ridge regression according to historical data collected in a historical control period includes: Obtain historical input data and historical output data according to historical data collected in a historical control period; Construct a past subset and a future subset respectively according to the historical input data and the historical output data; Establish a calculation model of the parameter vector based on ridge regression according to the past subset and the future subset.

5. The data-driven predictive current modeling method for a permanent magnet synchronous motor according to claim 4, characterized in that The establishing a calculation model of the parameter vector based on ridge regression according to the past subset and the future subset includes: Estimate the parameter vector by applying the solution of the least squares method according to the past subset and the future subset to establish a calculation model of the parameter vector based on ridge regression.

6. A data-driven predictive current modeling device for a permanent magnet synchronous motor, characterized in that, Including: An obtaining module, configured to obtain a data-driven current prediction model of the target motor based on the cross-saturation effect; wherein, the data-driven current prediction model includes a parameter vector; A building module, which is used to establish a calculation model of the parameter vector based on ridge regression according to historical data collected in historical control cycles; wherein, the calculation model includes a penalty term; A determination module, which is used to determine the target solution of the penalty term according to the noise severity of the historical data; before determining the target solution of the penalty term according to the noise severity of the historical data, it further includes: Construct matrices for the data corresponding to the input data and the output data in the historical data respectively; Obtain the sum of squared residuals and the total sum of squares according to the matrices; Obtain the multicollinearity of the input data according to the sum of squared residuals and the total sum of squares to characterize the noise severity of the historical data; The determining the target solution of the penalty term according to the noise severity of the historical data includes: Obtain the relationship between the multicollinearity and the penalty term according to the positive correlation between the multicollinearity and the noise severity of the historical data; Map the relationship between the multicollinearity and the penalty term based on a mapping function and determine the target solution of the penalty term; A target module, which is used to obtain a target penalty term and a target parameter vector according to the target solution to establish a target data-driven current prediction model.

7. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is loaded and executed by a processor, it implements the permanent magnet synchronous motor data-driven predictive current modeling method according to any one of claims 1-5.

8. An electronic device, characterized in that, Including a processor and a memory, wherein, The memory is used to store a computer program; The processor is used to load and execute the computer program so that the electronic device executes the permanent magnet synchronous motor data-driven predictive current modeling method according to any one of claims 1-5.

Images