Segmented friction torque modeling compensation method, system and apparatus for permanent magnet synchronous motor

Abstract

The present application relates to permanent magnet synchronous motor friction torque compensation technical field, especially in kind, a kind of permanent magnet synchronous motor segmented type friction torque modeling compensation method, system and device.It includes the following steps: collecting motor operating data, obtains motor operating data set;In extremely low speed interval, using the friction torque model based on dynamic learning rate neural network;In non-extremely low speed interval, using polynomial equation model;The friction torque model constructed is embedded in motor control system, according to the interval of current speed selects corresponding friction torque model.The present application is a kind of permanent magnet synchronous motor segmented type friction torque modeling and compensation method and system, by the concept of segmented modeling based on speed threshold, the motor operating interval is divided into extremely low speed interval and non-extremely low speed interval two operating conditions, and different modeling methods are used for different working conditions.Effectively reduce the low-speed crawling phenomenon, it is of great significance to improve the control accuracy and running stability of system.

Description

Technical Field

[0001] This invention relates to the field of friction torque compensation technology for permanent magnet synchronous motors, and in particular to a segmented friction torque modeling and compensation method, system and device for permanent magnet synchronous motors. Background Technology

[0002] In recent years, permanent magnet synchronous motor servo systems have been widely used in various technical fields due to their advantages such as high efficiency, high power density, and wide speed range. However, in the control scenarios of permanent magnet synchronous motor servo systems, due to the switching characteristics of static and dynamic friction and the nonlinear characteristics of viscous friction, the system is prone to low-speed crawling when the speed is below the critical value. This manifests as speed fluctuations, divergence in position tracking errors, and unstable motion, which will seriously affect the control accuracy of the system.

[0003] Friction torque compensation methods are generally divided into model-based and model-free methods. Among model-based methods, common friction models are further divided into static and dynamic models. Currently, dynamic models (with the LuGre model being the most prominent) are the most widely used. However, traditional friction models (such as the LuGre model) suffer from problems such as insufficient accuracy in describing the complex friction characteristics of low-speed, high-torque permanent magnet synchronous motors, especially in the low-speed domain and under varying operating conditions. These problems include insufficient accuracy in describing friction torque, limited ability to capture dynamic characteristics, poor adaptability, and complex parameter identification processes. Model-free compensation methods bypass the physical modeling of friction torque, treating friction as an unknown disturbance or a nonlinear black box, and directly suppressing its influence through control algorithms. However, model-free compensation methods suffer from problems such as heavy reliance on the accuracy of the system model and insufficient robustness.

[0004] In summary, traditional friction compensation methods all have certain drawbacks. Those skilled in the art have made improvements, such as using a data-driven and physical model fusion approach, integrating neural networks with the traditional LuGre model. Leveraging the powerful learning capabilities of neural networks, a two-dimensional mapping relationship between friction force and velocity / position is established, effectively improving the model's accuracy and adaptability under varying operating conditions. However, this method requires simultaneous optimization of neural network and physical parameters, making the identification process more complex. Furthermore, for low-speed, high-torque permanent magnet synchronous motor control systems, the inherent limitations of the LuGre model make it difficult to describe the strong nonlinearity of static-dynamic friction switching and the Stribeck effect hysteresis characteristics of low-speed, high-torque permanent magnet synchronous motors. This leads to model mismatch, resulting not only in the failure to effectively compensate for friction interference but also potentially damaging system stability due to inaccurate friction torque estimation, causing negative effects such as increased speed fluctuations or position tracking errors.

[0005] Therefore, it is necessary to propose a novel method for compensating for the frictional torque of permanent magnet synchronous motors to solve the above problems. Summary of the Invention

[0006] This invention aims to at least solve one of the technical problems existing in related technologies. To this end, this invention provides a segmented friction torque modeling and compensation method, system, and device for permanent magnet synchronous motors, which reduces low-speed creeping phenomena and is of great significance for improving the control accuracy and operational stability of the system.

[0007] This invention provides a segmented friction torque modeling and compensation method for permanent magnet synchronous motors, comprising the following steps:

[0008] S1: Collect motor operation data to obtain motor operation dataset;

[0009] S2: Based on the speed threshold set in the motor operation dataset, the motor operation range is divided into an extremely low speed range and a non-extremely low speed range;

[0010] S3: In the extremely low speed range, a friction torque model based on a dynamic learning rate neural network is used to establish a nonlinear mapping relationship between the motor friction torque and the speed, thus obtaining the friction torque model in the extremely low speed range.

[0011] S4: In the non-extremely low speed range, the friction torque model is described by the optimal order polynomial equation to establish a nonlinear mapping relationship between the friction torque model and the rotational speed, thus obtaining the friction torque model in the non-extremely low speed range.

[0012] S5: Embed the friction torque model in the extremely low speed range and the friction torque model in the non-extremely low speed range into the motor control system. Select the corresponding friction torque model according to the current speed range to obtain the realized friction torque. Convert the realized friction torque into compensation current and compensate for it in the motor terminal current.

[0013] According to the present invention, a segmented friction torque modeling and compensation method for a permanent magnet synchronous motor is provided, wherein the motor operating data includes: obtaining the rotor mechanical angular position through an incremental photoelectric encoder, calculating the motor speed through differentiation, and obtaining the friction torque of the motor through a torque sensor.

[0014] According to the present invention, a segmented friction torque modeling and compensation method for a permanent magnet synchronous motor is provided, wherein step S2 includes:

[0015] A speed threshold is set based on the motor speed and actual operating conditions. When the system is in an extremely low speed range, nonlinear friction has the greatest impact on the system; when When the system is in a non-extremely low speed range, it is defined as the system being in this range.

[0016] in, This refers to the actual motor speed. For speed threshold, It is an absolute value.

[0017] According to the segmented friction torque modeling and compensation method for permanent magnet synchronous motors provided by the present invention, the expression of the friction torque model of the dynamic learning rate neural network in step S3 is as follows:

[0018]

[0019] in, This is the output vector of the first hidden layer. It is the hyperbolic tangent function. These are the weight matrices for the first hidden layer. The motor speed is the input to the neural network. These are the bias vectors of the first hidden layer. This is the output vector of the second hidden layer. This is the weight matrix of the second hidden layer. This is the bias vector for the second hidden layer. The frictional torque value estimated by the neural network. This is the weight matrix of the output layer. This is the bias vector for the output layer. The layer number of the first hidden layer. This represents the total number of the first hidden layer. The layer number of the output layer. This represents the total number of output layers. The total number of the second hidden layers is the same as the total number of output layers.

[0020] According to the segmented friction torque modeling and compensation method for permanent magnet synchronous motors provided by the present invention, S4, the friction model for fitting the optimal order polynomial equation in the non-extremely low speed range includes:

[0021] The friction torque in the non-extremely low-speed range is modeled using polynomial equations. The fitting results of several orders of polynomials are calculated in parallel, and the residuals of each order of fitting are calculated. By comparing the magnitude of the residuals of each order of fitting, the order of polynomial with the smallest residual is selected as the optimal order. The friction torque model in the non-extremely low-speed range is established using the optimal order polynomial equation.

[0022] According to the present invention, a segmented friction torque modeling and compensation method for a permanent magnet synchronous motor is provided, wherein the polynomial equation is:

[0023]

[0024] in, The friction torque value is estimated by a polynomial. This represents the total number of terms in the polynomial. For polynomial coefficients, To estimate the input motor speed using a polynomial, Let be the degree of the polynomial.

[0025] According to the present invention, a segmented friction torque modeling and compensation method for a permanent magnet synchronous motor is provided, wherein step S5 includes:

[0026] Used to convert the estimated frictional torque into a compensation current. To achieve frictional torque The feedforward compensation formula is:

[0027]

[0028] in, This is the torque coefficient.

[0029] This invention also provides a segmented friction torque modeling and compensation system for permanent magnet synchronous motors, comprising:

[0030] Data acquisition module: used to collect motor operation data and obtain motor operation dataset;

[0031] Speed ​​differentiation module: used to divide the motor operating range into extremely low speed range and non-extreme low speed range according to the speed threshold set in the motor operation dataset;

[0032] Ultra-low speed prediction module: Used to establish a nonlinear mapping relationship between motor friction torque and speed in the ultra-low speed range using a friction torque model based on a dynamic learning rate neural network, and obtain the friction torque model in the ultra-low speed range;

[0033] Non-extremely low speed prediction module: Used to describe the friction torque model in the non-extremely low speed range using the optimal order polynomial equation to establish the nonlinear mapping relationship between the friction torque model and the rotational speed, and obtain the friction torque model in the non-extremely low speed range.

[0034] Motor compensation module: used to embed the friction torque model of the extremely low speed range and the friction torque model of the non-extremely low speed range into the motor control system, select the corresponding friction torque model according to the current speed range, obtain the realized friction torque, convert the realized friction torque into compensation current, and compensate the current at the motor end.

[0035] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the segmented friction torque modeling and compensation method for a permanent magnet synchronous motor as described above.

[0036] The above-described one or more technical solutions in the embodiments of the present invention have at least one of the following technical effects:

[0037] This invention provides a segmented friction torque modeling and compensation method, system, and device for permanent magnet synchronous motors. By dividing the motor's operating range into two conditions—an extremely low-speed range and a non-extremely low-speed range—based on a speed threshold-based segmented modeling approach, different modeling methods are employed for each condition. In the extremely low-speed range, a friction torque modeling method based on a dynamic learning rate neural network solves the problems of poor adaptability and complex parameter identification processes caused by the narrow extremely low-speed range of the traditional LuGre model. In the non-extremely low-speed range, an optimal-order polynomial equation is used to describe the nonlinear mapping relationship between friction torque and speed, ensuring accurate estimation of friction torque in the non-low-speed range while reducing computational complexity. Furthermore, the constructed friction torque model is embedded into the control system. The corresponding friction torque model is selected based on the current speed range to obtain an estimated friction torque value. This estimated value is then converted into a compensation current, which is used for compensation at the motor terminals, achieving dynamic compensation of the friction torque. This effectively reduces low-speed creeping and significantly improves the system's control accuracy and operational stability.

[0038] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0040] Figure 1 The flowchart illustrates a segmented friction torque modeling and compensation method for a permanent magnet synchronous motor, as provided in an embodiment of the present invention.

[0041] Figure 2 This is a schematic diagram of a segmented friction torque modeling and compensation system for a permanent magnet synchronous motor provided in an embodiment of the present invention.

[0042] Figure 3 This is a schematic diagram of the structure of the electronic device provided by the present invention.

[0043] Figure 4 This is a waveform diagram of the compensated rotational speed of the traditional LuGre model at 4 rpm.

[0044] Figure 5 This is a waveform diagram of the compensated rotational speed of the present invention at 4 rpm.

[0045] Figure 6This is a waveform diagram of the compensated rotational speed of the traditional LuGre model at 6 rpm.

[0046] Figure 7 This is a waveform diagram of the compensated rotational speed of the present invention at 6 rpm.

[0047] Figure 8 This is a waveform diagram of the compensated rotational speed of the traditional LuGre model at 9 rpm.

[0048] Figure 9 This is a waveform diagram of the compensated rotational speed of the present invention at 9 rpm.

[0049] Figure 10 It is a comparison chart of residuals of different orders.

[0050] Figure 11 This is the fitting curve when the fitting order is 7.

[0051] Figure 12 This is the fitting curve when the fitting order is 11.

[0052] Figure 13 This is a flowchart of a dynamic learning rate strategy.

[0053] Figure label:

[0054] 101. Data acquisition module; 102. Speed ​​differentiation module; 103. Extremely low speed prediction module; 104. Non-extremely low speed prediction module; 105. Motor compensation module; 810. Processor; 820. Communication interface; 830. Memory; 840. Communication bus. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention. The following embodiments are used to illustrate this invention but cannot be used to limit the scope of this invention.

[0056] In the description of the embodiments of the present invention, it should be noted that the terms "center," "longitudinal," "lateral," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing the embodiments of the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the embodiments of the present invention. In addition, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0057] In the description of the embodiments of the present invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "connected" and "linked" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms in the embodiments of the present invention based on the specific circumstances.

[0058] In embodiments of the present invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature is in indirect contact with the second feature through an intermediate medium. Furthermore, "above," "on top of," and "over" the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply that the first feature is at a lower horizontal level than the second feature.

[0059] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0060] The following is combined Figures 1 to 13 This invention is described.

[0061] Example

[0062] The present invention will now be described in further detail with reference to embodiments and accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. For ease of description, the accompanying drawings show only the parts relevant to the invention, not all of the structures.

[0063] This invention provides a segmented friction torque modeling and compensation method for permanent magnet synchronous motors. The method can solve the problems of poor adaptability and complex parameter identification process that occur when using the traditional LuGre model for low-speed, high-torque permanent magnet synchronous motor systems.

[0064] like Figure 1 As shown in the figure, the segmented friction torque modeling and compensation method for a permanent magnet synchronous motor provided by this embodiment of the invention includes the following steps:

[0065] S1: Collect motor operation data to obtain motor operation dataset;

[0066] S2: Based on the speed threshold set in the motor operation dataset, the motor operation range is divided into an extremely low speed range and a non-extremely low speed range;

[0067] S3: In the extremely low speed range, a friction torque model based on a dynamic learning rate neural network is used to establish a nonlinear mapping relationship between the motor friction torque and the speed, thus obtaining the friction torque model in the extremely low speed range.

[0068] S4: In the non-extremely low speed range, the friction torque model is described by the optimal order polynomial equation to establish a nonlinear mapping relationship between the friction torque model and the rotational speed, thus obtaining the friction torque model in the non-extremely low speed range.

[0069] S5: Embed the friction torque model in the extremely low speed range and the friction torque model in the non-extremely low speed range into the motor control system. Select the corresponding friction torque model according to the current speed range to obtain the realized friction torque. Convert the realized friction torque into compensation current and compensate for it in the motor terminal current.

[0070] Specifically, the steps of this embodiment of the invention are as follows:

[0071] S1: Collect data.

[0072] Within the speed range of -25 r / min to 25 r / min, the rotor mechanical angular position obtained by the incremental photoelectric encoder and the friction torque obtained by the high-precision torque sensor are acquired synchronously, and the position signal is differentiated to obtain the motor speed.

[0073] S2: Based on the set speed threshold, the motor operating range is divided into extremely low speed range and non-extreme low speed range.

[0074] A speed threshold is set based on motor parameters and actual operating conditions. ,when When the system is in an extremely low speed range, nonlinear friction has the greatest impact on the system; when When the system is in a non-extremely low speed range, it is defined as being in this range. (This refers to the speed threshold in this embodiment of the invention.) Taking 2% of the rated speed, the rated speed in this embodiment of the invention is 100 r / min, and the speed threshold in this embodiment of the invention is 2 r / min. Wherein, This represents the actual motor speed.

[0075] S3: In the extremely low speed range, a friction model based on a dynamic learning rate neural network is used to establish a nonlinear mapping relationship between the motor friction torque and the speed.

[0076] Approximately 100,000 valid data samples were extracted in the extremely low speed range (-2 r / min to 2 r / min) to form the original dataset. To further improve data quality, preprocessing operations such as filtering and normalization were performed on the original data. Finally, the dataset was divided into training, validation, and test sets according to a certain ratio.

[0077] Using the rotor angular velocity as input and the estimated friction torque as output, the neural network expression is:

[0078]

[0079] in, This is the output vector of the first hidden layer. It is the hyperbolic tangent function. These are the weight matrices for the first hidden layer. The motor speed is the input to the neural network. These are the bias vectors of the first hidden layer. This is the output vector of the second hidden layer. This is the weight matrix of the second hidden layer. This is the bias vector for the second hidden layer. The frictional torque value estimated by the neural network. This is the weight matrix of the output layer. This is the bias vector for the output layer. The layer number of the first hidden layer. This represents the total number of the first hidden layer. The layer number of the output layer. This represents the total number of output layers. The total number of the second hidden layers is the same as the total number of output layers.

[0080] A dynamic learning rate strategy based on loss function feedback and momentum was used to train the neural network to obtain a friction torque model in the extremely low speed range.

[0081] like Figure 13 As shown, specifically, dynamic learning rate is a dynamic learning rate strategy based on a combination of loss function feedback and momentum. During network training, the learning rate is dynamically adjusted according to the current error of the loss function and a set threshold. This adjustment mechanism is implemented through the following steps:

[0082] Loss function calculation: In each iteration, the learning rate is adjusted by calculating the error between the current model output and the true value. This embodiment of the invention uses the mean squared error (MSE) as the loss function error, and the formula for the loss function error is:

[0083]

[0084] in, Let be the error of the loss function in the nth iteration, where n is the number of iterations and N is the total number of training samples. Let be the true frictional torque of the t-th sample. Let t be the friction torque value estimated by the neural network for the t-th sample, where t is the sample number.

[0085] Dynamic learning rate adjustment: The learning rate of the neural network is dynamically adjusted based on a comparison between the loss function error and a preset error threshold. When the error is large, the learning rate is increased to accelerate convergence; when the error is small and close to the set minimum threshold, the learning rate is decreased to suppress oscillations.

[0086] Based on the comparison between the current iteration's loss function error and the preset threshold, the learning rate is dynamically adjusted according to the following rules:

[0087]

[0088] in, Let n be the learning rate of the neural network in the nth iteration. The learning rate of the neural network is the (n+1)th iteration. Learning rate growth factor This is the learning rate decay factor. The minimum threshold for the loss function error. This is the maximum threshold for the loss function error.

[0089] Momentum Update Mechanism: Stochastic gradient descent with momentum term is used to update the weights and biases of the neural network. Through this momentum mechanism, the learning process becomes smoother, avoiding getting trapped in local optima. In this embodiment of the invention, stochastic gradient descent with momentum term is used to update the neural network parameters. The model parameter update expression is:

[0090]

[0091] in, For the iteratively updated neural network weights or biases, These are the neural network weights or biases before the iterative update. The gradient of the loss function with respect to the parameters of the neural network model. The momentum coefficient, This refers to the weights or biases of the neural network from the previous iteration. , and The representative neural network weights or biases include , , , , and .

[0092] Finally, determine whether the requirements for ending the training have been met. If the requirements are met, the training is complete; otherwise, continue training.

[0093] S4: In the non-extremely low speed range, the nonlinear mapping relationship between friction torque and rotational speed is described by the optimal order polynomial equation.

[0094] In the non-extremely low speed range, firstly according to Data with large errors are removed in principle. Then, the collected data is weighted and averaged. Subsequently, the friction torque in the non-extremely low-speed range is modeled using an optimal order polynomial equation. The polynomial equation is as follows:

[0095]

[0096] in, The friction torque value is estimated by a polynomial. This represents the total number of terms in the polynomial. For polynomial coefficients, To estimate the input motor speed using a polynomial, Let be the degree of the polynomial.

[0097] Simultaneous calculation for The fitting results of the first-order polynomials are obtained, and the residuals of each order of fitting are calculated according to the definition of polynomial fitting residuals.

[0098] Since the motor operates in the non-low-speed range, the friction is in a fully fluid lubrication stage, and the friction increases with the speed, without any creeping phenomenon. To reduce the overall training load, neural network training is not performed in the non-low-speed range, so not much data was collected, only a dozen or so speed feature points were collected.

[0099] like Figure 10As shown, it can be seen that the deviation of the fitted curve decreases as the polynomial increases. However, considering that excessively high orders in polynomial fitting can easily lead to overfitting, it is necessary to control the fitting order to avoid the predicted curve deviating from the actual characteristics. Conversely, lower orders will lead to increased model prediction errors. Figure 11 and Figure 12 As shown in the fitting results, the total number of polynomial terms in the embodiments of the present invention is set to 5 to 10.

[0100] By comparing the residuals of fitting at each order, the polynomial order with the smallest residual is selected as the optimal order, and the friction torque model in the non-extremely low speed range is established using the optimal order polynomial equation.

[0101] S5: Embed the constructed friction torque model into the motor control system. Select the corresponding friction torque model according to the current speed range to obtain the estimated value of the friction torque. Convert the estimated value of the friction torque into a compensation current and compensate it at the motor terminal current to achieve dynamic compensation of the friction torque. The formula is:

[0102]

[0103]

[0104] in, The torque coefficient, To compensate for the current, To achieve frictional torque, For permanent magnet flux linkage in motors, This represents the number of pole pairs of the motor. This represents the moment of inertia of the motor.

[0105] Specifically, Figure 4 This is a waveform diagram of the compensated rotational speed of the traditional LuGre model at 4 rpm. Figure 5 These are experimental waveforms of the compensated rotational speed of this invention at 4 rpm. In the waveforms, red represents the waveform at the given rotational speed, and blue represents the compensated waveform. Based on these two figures, it is clear that this invention provides a better compensation effect than the traditional model.

[0106] Specifically, Figure 6 This is a waveform diagram of the compensated rotational speed of the traditional LuGre model at 6 rpm. Figure 7 These are experimental waveforms of the compensated rotational speed of this invention at 6 rpm. In the waveforms, red represents the waveform at the given rotational speed, and blue represents the compensated waveform. Based on these two figures, it is clear that this invention provides a better compensation effect than the traditional model.

[0107] Specifically, Figure 8 This is a waveform diagram of the compensated rotational speed of the traditional LuGre model at 9 rpm. Figure 9These are experimental waveforms of the compensated rotational speed of this invention at 9 rpm. In the waveforms, red represents the waveform at the given rotational speed, and blue represents the compensated waveform. Based on these two figures, it is clear that this invention provides a better compensation effect than the traditional model.

[0108] Table 1. Comparison of Root Mean Square of Tracking Errors

[0109]

[0110] Table 1 is a comparison table of the root mean square of the tracking error in the above four cases. It can be clearly seen that the compensation effect of the present invention is better than that of the traditional model.

[0111] like Figure 2 As shown in the figure, the segmented friction torque modeling and compensation system for a permanent magnet synchronous motor provided in this embodiment of the invention mainly includes the following parts:

[0112] Data acquisition module 101: Used to collect motor operation data and obtain motor operation dataset;

[0113] Speed ​​differentiation module 102: used to divide the motor operating range into an extremely low speed range and a non-extreme low speed range according to the speed threshold set in the motor operating dataset;

[0114] Ultra-low speed prediction module 103: used to establish a nonlinear mapping relationship between motor friction torque and speed in the ultra-low speed range using a friction torque model based on a dynamic learning rate neural network, and obtain the friction torque model in the ultra-low speed range;

[0115] Non-extremely low speed prediction module 104: used to describe the friction torque model in the non-extremely low speed range using the optimal order polynomial equation to establish the nonlinear mapping relationship between the friction torque model and the rotational speed, and obtain the friction torque model in the non-extremely low speed range.

[0116] Motor compensation module 105: used to embed the friction torque model of the extremely low speed range and the friction torque model of the non-extremely low speed range into the motor control system, select the corresponding friction torque model according to the current speed range, obtain the realized friction torque, convert the realized friction torque into compensation current, and compensate the current at the motor end.

[0117] Figure 3 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 3 As shown, the electronic device may include: a processor 810, a communication interface 820, a memory 830, and a communication bus 840, wherein the processor 810, the communication interface 820, and the memory 830 communicate with each other through the communication bus 840. The processor 810 can call logic instructions in the memory 830 to execute a backstepping sliding mode control method for a power grid frequency control system, the method including:

[0118] S1: Collect motor operation data to obtain motor operation dataset;

[0119] S2: Based on the speed threshold set in the motor operation dataset, the motor operation range is divided into an extremely low speed range and a non-extremely low speed range;

[0120] S3: In the extremely low speed range, a friction torque model based on a dynamic learning rate neural network is used to establish a nonlinear mapping relationship between the motor friction torque and the speed.

[0121] S4: In the non-extremely low speed range, the nonlinear mapping relationship between friction torque and rotational speed is described by the optimal order polynomial equation.

[0122] S5: Embed the constructed friction torque model into the motor control system, select the corresponding friction torque model according to the current speed range, obtain the estimated value of the friction torque, convert the estimated value of the friction torque into a compensation current, and compensate the current at the motor end.

[0123] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0124] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

[0125] It should be noted that the embodiments of this disclosure can be implemented using hardware, software, or a combination of both. The hardware portion can be implemented using dedicated logic; the software portion can be stored in memory and executed by a suitable instruction execution system, such as a microprocessor or dedicated-design hardware. Those skilled in the art will understand that the above-described devices and methods can be implemented using computer-executable instructions and / or included in processor control code, for example, such code provided on a programmable memory or a data carrier such as an optical or electronic signal carrier.

[0126] Furthermore, although the operation of the methods of this disclosure is described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. Rather, the steps depicted in the flowcharts may be performed in a different order. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps. It should also be noted that the features and functions of two or more devices according to this disclosure may be embodied in one device. Conversely, the features and functions of one device described above may be further divided and embodied by multiple devices.

[0127] While this disclosure has been described with reference to several specific embodiments, it should be understood that this disclosure is not limited to the specific embodiments disclosed. This disclosure is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims.

Claims

1. A segmented friction torque modeling and compensation method for a permanent magnet synchronous motor, characterized in that, Includes the following steps: S1: Collect motor operation data to obtain motor operation dataset; S2: Based on the speed threshold set in the motor operation dataset, the motor operation range is divided into an extremely low speed range and a non-extremely low speed range; S3: In the extremely low speed range, a friction torque model based on a dynamic learning rate neural network is used to establish a nonlinear mapping relationship between the motor friction torque and the rotational speed, thus obtaining the friction torque model for the extremely low speed range; the expression for the friction torque model based on the dynamic learning rate neural network in step S3 is: in, This is the output vector of the first hidden layer. It is the hyperbolic tangent function. These are the weight matrices for the first hidden layer. The motor speed is the input to the neural network. These are the bias vectors of the first hidden layer. This is the output vector of the second hidden layer. This is the weight matrix of the second hidden layer. This is the bias vector for the second hidden layer. The frictional torque value estimated by the neural network. The weight matrix of the output layer. This is the bias vector for the output layer. The layer number of the first hidden layer. This represents the total number of the first hidden layer. The layer number of the output layer. This represents the total number of output layers, and the total number of the second hidden layers is the same as the total number of output layers. Dynamic learning rate is achieved through the following steps: Loss function calculation: In each iteration, the learning rate is adjusted by calculating the error between the current model output and the true value. The formula for the loss function error is: in, Let be the error of the loss function in the nth iteration, where n is the number of iterations and N is the total number of training samples. Let be the true frictional torque of the t-th sample. Let t be the friction torque value estimated by the neural network for the t-th sample, where t is the sample number; Dynamically adjust the learning rate: Based on the comparison between the loss function error and a preset error threshold, the learning rate of the neural network is dynamically adjusted. The formula is as follows: in, Let n be the learning rate of the neural network in the nth iteration. The learning rate of the neural network is the (n+1)th iteration. Learning rate growth factor This is the learning rate decay factor. The minimum threshold for the loss function error. This represents the maximum threshold for the loss function error. Momentum update mechanism: Combining stochastic gradient descent with momentum terms, it is used to update the weights and biases of the neural network. The neural network parameters are updated using stochastic gradient descent with momentum terms. The model parameter update expression is: in, For the iteratively updated neural network weights or biases, These are the neural network weights or biases before the iterative update. The gradient of the loss function with respect to the parameters of the neural network model. The momentum coefficient, Here, represents the weights or biases of the neural network from the previous iteration, where... , and The representative neural network weights or biases include , , , , and ; S4: In the non-extremely low speed range, the friction torque model is described by the optimal order polynomial equation to establish a nonlinear mapping relationship between the friction torque model and the rotational speed, thus obtaining the friction torque model in the non-extremely low speed range. S5: Embed the friction torque model in the extremely low speed range and the friction torque model in the non-extremely low speed range into the motor control system. Select the corresponding friction torque model according to the current speed range to obtain the realized friction torque. Convert the realized friction torque into compensation current and compensate it at the motor end current.

2. The segmented friction torque modeling and compensation method for a permanent magnet synchronous motor according to claim 1, characterized in that, The motor operating data includes: the rotor mechanical angular position obtained by an incremental photoelectric encoder, the motor speed calculated by differentiation, and the motor friction torque obtained by a torque sensor.

3. The segmented friction torque modeling and compensation method for a permanent magnet synchronous motor according to claim 1, characterized in that, Step S2 includes: Set the speed threshold according to the motor speed and actual operating conditions, when When, it is defined as the system being in the extremely low speed range; when When the system is in a non-extremely low speed range, it is defined as such. in, This is the actual motor speed. For speed threshold, It is an absolute value.

4. The segmented friction torque modeling and compensation method for a permanent magnet synchronous motor according to claim 1, characterized in that, The friction model fitted by the optimal order polynomial equation in step S4 includes: The friction torque in the non-extremely low-speed range is modeled using polynomial equations. The fitting results of several orders of polynomials are calculated in parallel, and the residuals of each order of fitting are calculated. By comparing the magnitude of the residuals of each order of fitting, the order of polynomial with the smallest residual is selected as the optimal order. The friction torque model in the non-extremely low-speed range is established using the optimal order polynomial equation.

5. The segmented friction torque modeling and compensation method for a permanent magnet synchronous motor according to claim 1, characterized in that, The polynomial equation is: in, The friction torque value is estimated by a polynomial. This represents the total number of terms in the polynomial. For polynomial coefficients, To estimate the input motor speed using a polynomial, Let be the degree of the polynomial.

6. The segmented friction torque modeling and compensation method for a permanent magnet synchronous motor according to claim 1, characterized in that, Step S5 includes: The estimated value of the frictional torque is converted into a compensation current. To achieve frictional torque The feedforward compensation formula is: in, This is the torque coefficient.

7. A segmented friction torque modeling and compensation system for a permanent magnet synchronous motor, used to execute the segmented friction torque modeling and compensation method for a permanent magnet synchronous motor as described in any one of claims 1 to 6, characterized in that, include: Data acquisition module: used to collect motor operation data and obtain motor operation dataset; Speed ​​differentiation module: used to divide the motor operating range into extremely low speed range and non-extreme low speed range according to the speed threshold set in the motor operation dataset; Ultra-low speed prediction module: Used to establish a nonlinear mapping relationship between motor friction torque and speed in the ultra-low speed range using a friction torque model based on a dynamic learning rate neural network, and obtain the friction torque model in the ultra-low speed range; Non-extremely low speed prediction module: Used to describe the friction torque model in the non-extremely low speed range using the optimal order polynomial equation to establish the nonlinear mapping relationship between the friction torque model and the rotational speed, and obtain the friction torque model in the non-extremely low speed range. Motor compensation module: used to embed the friction torque model of the extremely low speed range and the friction torque model of the non-extremely low speed range into the motor control system, select the corresponding friction torque model according to the current speed range, obtain the realized friction torque, convert the realized friction torque into compensation current, and compensate the current at the motor end.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the segmented friction torque modeling and compensation method for a permanent magnet synchronous motor as described in any one of claims 1 to 6.

Images