An Adaptive Integral Terminal Sliding Mode Control Method for Permanent Magnet Synchronous Motors Based on RBF Neural Network

Abstract

This invention provides an adaptive integral terminal sliding mode control method for permanent magnet synchronous motors based on RBF neural networks. First, a mathematical model of the permanent magnet synchronous motor and an adaptive integral terminal sliding mode speed loop control model are constructed. Based on this, the gradient descent method is used to automatically calibrate the controller parameters under different operating conditions. Then, a radial basis function (RBF) neural network model is constructed and trained so that the controller parameters can be automatically adjusted according to the actual operating conditions. Finally, the neural network is combined with the adaptive integral terminal sliding mode control, which can realize automatic adjustment of the speed loop controller parameters according to the target speed and load torque, significantly improving dynamic response performance while effectively reducing overshoot and improving steady-state performance.

Description

Technical Field

[0001] This invention belongs to the field of permanent magnet synchronous motor control technology, specifically relating to an adaptive integral terminal sliding mode permanent magnet synchronous motor control method based on RBF neural network. Background Technology

[0002] In the control of permanent magnet synchronous motors (PMSMs), traditional methods employ a dual-closed-loop, dual-PI vector control system composed of multiple speed and current loops. However, since PMSMs are nonlinear time-varying systems, traditional PI control is insufficient in dynamic response and steady-state performance. Furthermore, manual calibration of controller parameters is often required to adapt to various operating conditions, especially with different load torques and target speeds. Recent PMSM control technologies have introduced sliding mode variable structure control, which offers advantages such as simple structure, fast response, and strong robustness, making it more suitable for nonlinear control compared to traditional methods. However, the performance of sliding mode control often depends on reasonable controller parameter settings; improper parameter settings can lead to chattering and degraded control performance. Since the requirements for controller parameters vary under different operating conditions, parameter calibration is often complex and time-consuming. Therefore, there is an urgent need in this field for a nonlinear control method with adaptive parameter adjustment capabilities that can significantly improve PMSM control performance under multiple operating conditions while simplifying the debugging process. Summary of the Invention

[0003] In view of this, and to address the technical problems existing in this field, the present invention provides an adaptive integral terminal sliding mode permanent magnet synchronous motor control method based on RBF neural network, specifically including the following steps:

[0004] Step 1: Construct an equivalent mathematical model of the permanent magnet synchronous motor in a synchronous rotating coordinate system, including voltage equations, electromagnetic torque equations, and motor dynamics; adopt deadbeat predictive control (DPCC) for the current loop of the permanent magnet synchronous motor, and use the current at time k to predict the control voltage at time k+1 in combination with the voltage equation; adopt adaptive integral terminal sliding surface control for the speed loop; and obtain the adaptive integral terminal sliding surface speed loop control model by solving the equations simultaneously.

[0005] Step 2: Define the loss function and its related controller parameters to be optimized based on the difference between the actual speed and the ideal speed, and use the gradient descent method to automatically calibrate the controller parameters under various operating conditions;

[0006] Step 3: Establish a radial basis function (RBF) neural network model, and train the neural network using the correspondence between different operating conditions and their corresponding calibrated controller parameters, so that the trained neural network can automatically adjust to obtain the optimal controller parameters according to the operating conditions.

[0007] Step 4: Combine the trained RBF neural network with the adaptive integral terminal sliding mode speed loop control model. The RBF neural network calculates the optimal controller parameters using the load torque and target speed under different operating conditions, and then obtains the control voltage reference value by passing through the speed loop and current loop in sequence.

[0008] Furthermore, step 1 specifically establishes an equivalent mathematical model for the surface-mounted permanent magnet synchronous motor, including: establishing the following dq-axis voltage equations:

[0009]

[0010] In the formula, u d and u q These are the d-axis and q-axis stator voltage components, respectively; i d and i q These are the d-axis and q-axis stator current components, respectively; R s For stator resistance; L s Stator inductance; ψ f For rotor permanent magnet flux linkage, ω e The rotor's electric angular velocity is given by t, where t is time.

[0011] The following torque equations for a permanent magnet synchronous motor are established:

[0012]

[0013] In the formula, T e The electromagnetic torque is p; the number of pole pairs is p.

[0014] And establish the following dynamic equations for the permanent magnet synchronous motor:

[0015]

[0016] In the formula, J is the rotor's moment of inertia; ω m T is the rotor's mechanical angular velocity; l is the load torque; B is the viscous friction coefficient.

[0017] Furthermore, in step 1, deadbeat predictive control is specifically employed, and the following predictive model is established to predict the control voltage at time k+1:

[0018]

[0019] Where u d (k+1), u q (k+1) represent the predicted voltage vectors along the d and q axes at time k+1, respectively; These are the command current vectors along the d and q axes at time k, respectively; T s i is the sampling time; d(k), i q (k) represents the d-axis and q-axis sampling vectors at time k; ω e (k) represents the sampled electrical angular velocity of the motor at time k;

[0020] The adaptive integral terminal sliding surface of the speed loop is designed as follows:

[0021] s = e + s I

[0022] In the formula, s is the sliding surface, e is the mechanical angular velocity error, and s I For sliding mode integral terms;

[0023] Adaptive terminal integral sliding mode term s I for:

[0024] s I =max[-I max ,min(I,I max )]

[0025] In the formula, I max I represents the limit of the adaptive integral term; I represents the sliding mode term at the integral terminal.

[0026] The integral terminal sliding mode term I is:

[0027]

[0028] In the formula, η1 and η2 are the adaptive adjustment parameters; tanh is the hyperbolic cosine function;

[0029] Adaptive integral term limit I max for:

[0030] I max =i qmax (1-e -0.5e )

[0031] In the formula, i qmax This represents the maximum value of the motor's q-axis current.

[0032] The following adaptive reaching law is adopted:

[0033]

[0034] In the formula, η3 is also an adaptive adjustment parameter;

[0035] Differentiating with respect to the sliding surface, we get:

[0036]

[0037] The superscript · indicates the derivative of the corresponding parameter;

[0038] Since the rotational speed is a constant value, Will and Lianlide:

[0039]

[0040] Finally, by combining the dynamic equations of the permanent magnet synchronous motor and the electromagnetic torque equations, the following adaptive integral terminal sliding mode speed loop control model is obtained:

[0041]

[0042] Where η1, η2, and η3 are the parameters of the adaptive integral terminal sliding diaphragm speed loop controller, and their values ​​are all positive numbers.

[0043] Furthermore, step 2, the process of automatically calibrating the controller parameters under various operating conditions, includes the following steps:

[0044] Consider the load torque T under different operating conditions l ref and target rotational speed n ref The combination of these factors is used to define the following loss function J:

[0045]

[0046] In the formula, N is the total number of sampling points, which is determined by the controller's sampling frequency and sampling time; n is the controller's sampling rotation speed; n ideal The ideal speed curve determined based on the current operating conditions is specifically defined as follows:

[0047]

[0048] In the formula, T emax t represents the maximum electromagnetic torque of the motor. r The time required for the motor to reach the target speed.

[0049] Set the initial values ​​of the controller parameters to be optimized to... Then, the iteration begins, a startup experiment is conducted, and the initial loss function J is calculated. 0 ;

[0050] Define a learning rate δ and increment the parameter to be optimized. Conduct a startup experiment and calculate the loss function J. 1

[0051] Increment the optimized parameters Conduct a startup experiment and calculate the loss function J. 2 ;

[0052] Increment the optimized parameters Conduct a startup experiment and calculate the loss function J.3 ;

[0053] Calculate the gradient of the objective function with respect to the controller parameters.

[0054]

[0055] In the formula, J i =[J 1 J 2 J 3 ],

[0056] The initial parameter values ​​are updated as follows:

[0057]

[0058] The iteration continues until the result converges or the maximum number of iterations is reached, thus completing the automatic calibration of the controller parameters to be optimized.

[0059] Furthermore, step 3, the construction and training process of the RBF neural network, includes the following steps:

[0060] a. Define input data The output data Y = [η1η2η3] is normalized, and training and test sets are constructed.

[0061] b. Construct an RBF neural network model, set network parameters including training iterations, learning rate, minimum training error, etc., and set initial values ​​for weights and biases;

[0062] c. Train the RBF neural network model using the training set;

[0063] d. Test the RBF neural network model using the test set;

[0064] e. Determine whether the training results meet expectations by inverse normalizing the prediction results and comparing them with the true values;

[0065] f. If the prediction result matches the expected error, end the training, output the weights and biases of the neural network, and obtain the trained optimal controller parameter prediction model; otherwise, update the weights and biases, and repeat the be step until the training is complete.

[0066] Furthermore, the process of calculating the voltage reference value in step 4 includes the following steps:

[0067] a. Input the load torque and target speed into the trained RBF neural network to calculate the corresponding optimal controller parameters η1, η2, and η3;

[0068] b. Input the optimal controller parameters η1, η2, and η3 into the adaptive integral terminal speed slip mode controller model to calculate the current reference value i. q ref ;

[0069] c. Set the current reference value i s ref Input a current loop based on deadbeat predictive control to calculate the voltage reference value u. d ref and u q ref .

[0070] The adaptive integral terminal sliding mode permanent magnet synchronous motor control method based on RBF neural network provided by the present invention first constructs a mathematical model of permanent magnet synchronous motor and an adaptive integral terminal sliding mode speed loop control model. On this basis, the gradient descent method is used to automatically calibrate the controller parameters under different operating conditions. Then, a radial basis (RBF) neural network model is constructed and trained so that the controller parameters can be automatically adjusted according to the actual operating conditions. Finally, the neural network is combined with the adaptive integral terminal sliding mode, which can realize the automatic adjustment of the speed loop controller parameters according to the target speed and load torque, significantly improving the dynamic response performance, while effectively reducing overshoot and improving steady-state performance. Attached Figure Description

[0071] Figure 1 This is a control block diagram of the method provided by the present invention;

[0072] Figure 2 Flowchart for calibrating controller parameters using gradient descent method;

[0073] Figure 3 Flowchart for training and testing RBF neural networks;

[0074] Figure 4 This is a comparison diagram of the effects of the present invention and an example using a traditional PI control method;

[0075] Figure 5 This is a comparison diagram of the effects of the present invention and another example using the traditional PI control method. Detailed Implementation

[0076] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0077] like Figure 1As shown, the control system built based on this invention mainly consists of a DPCC current loop controller, a parameter-adjustable adaptive integral terminal sliding mode speed loop controller, an RBF neural network module, a Clarke / Park transform module, an SVPWM modulation module, a three-phase voltage inverter, and a permanent magnet synchronous motor model. Current sensors detect the three-phase current of the motor in real time, and the real-time d-axis and q-axis currents are obtained through coordinate transformation; speed sensors detect the real-time speed signal and rotor position signal of the motor. The target speed and load torque are calculated by the RBF neural network module to obtain the speed loop controller parameters. The control parameters and speed error are input to the speed loop controller to obtain the q-axis reference current. Using i... d =0 control strategy, the d and q axis reference current and the actual current are input to the current loop controller to obtain the d and q axis reference voltages, and the α and β axis voltages are obtained through coordinate transformation. After space vector modulation by the SVPWM module, the modulated signal is output to the inverter module, and finally drives the permanent magnet synchronous motor to run.

[0078] The corresponding control method specifically includes the following steps:

[0079] Step 1: Construct an equivalent mathematical model of the permanent magnet synchronous motor in a synchronous rotating coordinate system, including voltage equations, electromagnetic torque equations, and motor dynamics; adopt deadbeat predictive control (DPCC) for the current loop of the permanent magnet synchronous motor, and use the current at time k to predict the control voltage at time k+1 in combination with the voltage equation; adopt adaptive integral terminal sliding surface control for the speed loop; and obtain the adaptive integral terminal sliding surface speed loop control model by solving the equations simultaneously.

[0080] Step 2: Define the loss function and its related controller parameters to be optimized based on the difference between the actual speed and the ideal speed, and use the gradient descent method to automatically calibrate the controller parameters under various operating conditions;

[0081] Step 3: Establish a radial basis function (RBF) neural network model, and train the neural network using the correspondence between different operating conditions and their corresponding calibrated controller parameters, so that the trained neural network can automatically adjust to obtain the optimal controller parameters according to the operating conditions.

[0082] Step 4: Combine the trained RBF neural network with the adaptive integral terminal sliding mode speed loop control model. The RBF neural network calculates the optimal controller parameters using the load torque and target speed under different operating conditions, and then obtains the control voltage reference value by passing through the speed loop and current loop in sequence.

[0083] In a preferred embodiment of the present invention, step 1 specifically establishes an equivalent mathematical model for a surface-mounted permanent magnet synchronous motor, including: establishing the following dq-axis voltage equations:

[0084]

[0085] In the formula, u d and u q These are the d-axis and q-axis stator voltage components, respectively; i d and i q These are the d-axis and q-axis stator current components, respectively; R s For stator resistance; L s Stator inductance; ψ f For rotor permanent magnet flux linkage, ω e The rotor's electric angular velocity is given by t, where t is time.

[0086] The following torque equations for a permanent magnet synchronous motor are established:

[0087]

[0088] In the formula, T e The electromagnetic torque is p; the number of pole pairs is p.

[0089] And establish the following dynamic equations for the permanent magnet synchronous motor:

[0090]

[0091] In the formula, J is the rotor's moment of inertia; ω m T is the rotor's mechanical angular velocity; l is the load torque; B is the viscous friction coefficient.

[0092] In a preferred embodiment of the present invention, step 1 specifically employs deadbeat predictive control and establishes the following predictive model to predict the control voltage at time k+1:

[0093]

[0094] Where u d (k+1), u q (k+1) represent the predicted voltage vectors along the d and q axes at time k+1, respectively; These are the command current vectors along the d and q axes at time k, respectively; T s i is the sampling time; d (k), i q (k) represents the d-axis and q-axis sampling vectors at time k; ω e (k) represents the sampled electrical angular velocity of the motor at time k;

[0095] Designing an adaptive integral terminal sliding surface for the speed loop can improve system robustness, enhance system control performance, and reduce chattering; however, it presents challenges in parameter tuning. The sliding surface form is as follows:

[0096] s = e + s I

[0097] In the formula, s is the sliding surface, e is the mechanical angular velocity error, and s I For sliding mode integral terms;

[0098] Adaptive terminal integral sliding mode term s I for:

[0099] s I =max[-I max ,min(I,I max )]

[0100] In the formula, I max I represents the limit of the adaptive integral term; I represents the sliding mode term at the integral terminal.

[0101] The integral terminal sliding mode term I is:

[0102]

[0103] In the formula, η1 and η2 are the adaptive adjustment parameters; tanh is the hyperbolic cosine function;

[0104] The adaptive integral term limit is unaffected when the speed error is large, and also close to 0 when the speed error is close to 0, thereby reducing the chattering phenomenon on the sliding surface. The adaptive integral term limit I... max for:

[0105] I max =i qmax (1-e -0.5e )

[0106] In the formula, i qmax This represents the maximum value of the motor's q-axis current.

[0107] The design concept of the adaptive reaching law is similar to that described above. When the rotational speed error is large, its limit value is unaffected; when the rotational speed error is close to 0, its limit value is also close to 0, thereby reducing the chattering phenomenon on the sliding surface. Therefore, the following adaptive reaching law is adopted:

[0108]

[0109] In the formula, η3 is also an adaptive adjustment parameter;

[0110] Differentiating with respect to the sliding surface, we get:

[0111]

[0112] The superscript · indicates the derivative of the corresponding parameter;

[0113] Since the rotational speed is a constant value, Will and Lianlide:

[0114]

[0115] Finally, by combining the dynamic equations of the permanent magnet synchronous motor and the electromagnetic torque equations, the following adaptive integral terminal sliding mode speed loop control model is obtained:

[0116]

[0117] Where η1, η2, and η3 are the parameters of the adaptive integral terminal sliding diaphragm speed loop controller, and their values ​​are all positive numbers.

[0118] Choose Lyapunov functions:

[0119]

[0120] The stability of the system is analyzed by calculating the derivative of the Lyapunov function; if the derivative is less than 0, the system is stable.

[0121]

[0122] As shown in the above equation, the derivative of the Lyapunov function is always less than 0, indicating system stability. A drawback of adaptive integral terminal sliding mode is the complexity of parameter tuning, with different controller parameters corresponding to different operating conditions. To simplify the tuning process, the gradient descent method is used to automatically calibrate the controller parameters for each operating condition.

[0123] Multi-condition refers to the load torque T l ref and target rotational speed n ref The combination uses a motor with a load torque range of 0-10 Nm and a target speed range of 0-1000 rpm. For example... Figure 2 The diagram shown is a flowchart for calibrating controller parameters using the gradient descent method.

[0124] In a preferred embodiment of the present invention, step 2, which involves automatically calibrating the controller parameters under various operating conditions, includes the following steps:

[0125] Consider the load torque T under different operating conditions l ref and target rotational speed n ref The combination of these factors is used to define the following loss function J:

[0126]

[0127] In the formula, N is the total number of sampling points, which is determined by the controller's sampling frequency and sampling time; n is the controller's sampling rotation speed; n ideal The ideal speed curve determined based on the current operating conditions is specifically defined as follows:

[0128]

[0129] In the formula, T emax t represents the maximum electromagnetic torque of the motor. r The time required for the motor to reach the target speed.

[0130] Set the initial values ​​of the controller parameters to be optimized to... Then, the iteration begins, a startup experiment is conducted, and the initial loss function J is calculated. 0 ;

[0131] Define a learning rate δ and increment the parameter to be optimized. Conduct a startup experiment and calculate the loss function J. 1

[0132] Increment the optimized parameters Conduct a startup experiment and calculate the loss function J. 2 ;

[0133] Increment the optimized parameters Conduct a startup experiment and calculate the loss function J. 3 ;

[0134] Calculate the gradient of the objective function with respect to the controller parameters.

[0135]

[0136] In the formula, J i =[J 1 J 2 J 3 ],

[0137] The initial parameter values ​​are updated as follows:

[0138]

[0139] The iteration continues until the result converges or the maximum number of iterations is reached, thus completing the automatic calibration of the controller parameters to be optimized.

[0140] In a preferred embodiment of the present invention, step 3, the construction and training process of the RBF neural network, includes the following steps:

[0141] a. Define input data The output data Y = [η1η2η3] is normalized, and training and test sets are constructed.

[0142] b. Construct an RBF neural network model, set network parameters including training iterations, learning rate, minimum training error, etc., and set initial values ​​for weights and biases;

[0143] c. Train the RBF neural network model using the training set;

[0144] d. Test the RBF neural network model using the test set;

[0145] e. Determine whether the training results meet expectations by inverse normalizing the prediction results and comparing them with the true values;

[0146] f. If the prediction result matches the expected error, end the training, output the weights and biases of the neural network, and obtain the trained optimal controller parameter prediction model; otherwise, update the weights and biases, and repeat the be step until the training is complete.

[0147] In a preferred embodiment of the present invention, the process of calculating the voltage reference value in step 4 includes the following steps:

[0148] a. Input the load torque and target speed into the trained RBF neural network to calculate the corresponding optimal controller parameters η1, η2, and η3;

[0149] b. Input the optimal controller parameters η1, η2, and η3 into the adaptive integral terminal speed slip mode controller model to calculate the current reference value i. q ref ;

[0150] c. Set the current reference value i s ref Input a current loop based on deadbeat predictive control to calculate the voltage reference value u. d ref and u q ref .

[0151] Figure 4 Experimental diagrams of rotational speed and d-axis and q-axis current are shown for the PI speed loop and DPCC current loop (left) and the DPCC current loop (right) based on the adaptive integral terminal sliding mode of the RBF neural network under the 300RPM / 9Nm operating condition.

[0152] Figure 5 Experimental diagrams of speed and d-axis and q-axis current are shown for a PI speed loop and DPCC current loop (left) and an RBF neural network-based adaptive integral terminal sliding mode speed loop and DPCC current loop (right) under 500 RPM / 7 Nm operating conditions. The control period of the speed loop is 4e-4 s, and the control period of the current loop is 5e-5 s. Experimental results show that the traditional method suffers from overshoot and slow response speed, and is affected differently under different operating conditions. The improved control strategy is not limited by the operating conditions and can automatically adjust to the optimal control parameters, enhancing the robustness of the controller and ensuring the dynamic response performance and steady-state performance of the motor under various operating conditions.

[0153] It should be understood that the sequence number of each step in the embodiments of the present invention does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0154] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An adaptive integral terminal sliding mode permanent magnet synchronous motor control method based on RBF neural network, characterized in that: Specifically, the following steps are included: Step 1: Construct an equivalent mathematical model of the permanent magnet synchronous motor in a synchronous rotating coordinate system, including voltage equations, electromagnetic torque equations, and motor dynamics; employ deadbeat predictive control for the current loop of the permanent magnet synchronous motor, and use the current at time k to predict the control voltage at time k+1 based on the voltage equation; use an adaptive integral terminal sliding mode surface for control of the speed loop; combine the equations to obtain the adaptive integral terminal sliding mode speed loop control model; specifically, deadbeat predictive control is used, and the following prediction model is established to predict the control voltage at time k+1: Among them, u d and u q These are the d-axis and q-axis stator voltage components, respectively; i d and i q These are the d-axis and q-axis stator current components, respectively; R s For stator resistance; L s Stator inductance; ψ f For rotor permanent magnet flux linkage; ω e The rotor's electric angular velocity; u d (k+1), u q (k+1) represent the predicted voltage vectors along the d and q axes at time k+1, respectively; These are the command current vectors along the d and q axes at time k, respectively; T s i is the sampling time; d (k), i q (k) represents the d-axis and q-axis sampling vectors at time k; ω e (k) represents the sampled electrical angular velocity of the motor at time k; The adaptive integral terminal sliding surface of the speed loop is designed as follows: s=e+s I In the formula, s is the sliding surface, e is the mechanical angular velocity error, and s I For sliding mode integral terms; Adaptive terminal integral sliding mode term s I for: s I =max[-I max ,min(I,I max )] In the formula, I max I represents the limit of the adaptive integral term; I represents the sliding mode term at the integral terminal. The integral terminal sliding mode term I is: In the formula, tanh is the hyperbolic cosine function; Adaptive integral term limit I max for: I max =i qmax (1-e -0.5| yes | ) In the formula, i qmax This represents the maximum value of the motor's q-axis current. The following adaptive reaching law is adopted: In the formula, t represents time; Differentiating with respect to the sliding surface, we get: The superscript · indicates the derivative of the corresponding parameter; Since the rotational speed is a constant value, ω m Let s be the rotor's mechanical angular velocity, and let s be... Lianlide: Finally, by combining the dynamic equations of the permanent magnet synchronous motor and the electromagnetic torque equations, the following adaptive integral terminal sliding mode speed loop control model is obtained: Where η1, η2, and η3 are the parameters of the adaptive integral terminal sliding mode speed loop controller, and their values ​​are all positive numbers; J is the rotor moment of inertia; and p is the number of pole pairs. T l This is the load torque; Step 2: Define the loss function and its related controller parameters to be optimized based on the difference between the actual speed and the ideal speed. Then, automatically calibrate the controller parameters under various operating conditions using the gradient descent method. The process includes the following steps: Consider the load torque corresponding to different operating conditions and target rotational speed n ref The combination of these factors is used to define the following loss function Θ: In the formula, N is the total number of sampling points, which is determined by the controller's sampling frequency and sampling time; n is the controller's sampling rotation speed; n ideal The ideal speed curve determined based on the current operating conditions is specifically defined as follows: In the formula, T emax t represents the maximum electromagnetic torque of the motor. r The time required for the motor to reach the target speed. Set the initial values ​​of the controller parameters to be optimized to... Then, the iteration begins, a startup experiment is conducted, and the initial loss function Θ is calculated. 0 ; Define a learning rate δ and increment the parameter to be optimized. Conduct a startup experiment and calculate the loss function Θ. 1 Increment the optimized parameters Conduct a startup experiment and calculate the loss function Θ. 2 ; Increment the optimized parameters Conduct a startup experiment and calculate the loss function Θ. 3 ; Calculate the gradient of the objective function with respect to the controller parameters. Where, Θ i = [Θ 1 Θ 2 Θ 3 , The initial parameter values ​​are updated as follows: The iteration continues until the result converges or the set maximum number of iterations is reached, and then the automatic calibration of the controller parameters to be optimized is completed. Step 3: Establish a radial basis function (RBF) neural network model, and train the neural network using the correspondence between different operating conditions and their corresponding calibrated controller parameters, so that the trained neural network can automatically adjust to obtain the optimal controller parameters according to the operating conditions. Step 4: Combine the trained RBF neural network with the adaptive integral terminal sliding mode speed loop control model. The RBF neural network calculates the optimal controller parameters using the load torque and target speed under different operating conditions, and then obtains the control voltage reference value by passing through the speed loop and current loop in sequence.

2. The method as described in claim 1, characterized in that: Step 1 specifically establishes an equivalent mathematical model for the surface-mounted permanent magnet synchronous motor, including: establishing the following dq-axis voltage equations: In the formula, u d and u q These are the d-axis and q-axis stator voltage components, respectively; i d and i q These are the d-axis and q-axis stator current components, respectively. R s Stator resistance; L s For stator inductance; ψ f For rotor permanent magnet flux linkage, ω e The rotor's electric angular velocity is given by t, where t is time. The following torque equations for a permanent magnet synchronous motor are established: In the formula, T e The electromagnetic torque is p; the number of pole pairs is p. And establish the following dynamic equations for the permanent magnet synchronous motor: In the formula, J is the moment of inertia of the rotor; ω m This refers to the rotor's mechanical angular velocity; T l is the load torque; B is the viscous friction coefficient.

3. The method as described in claim 2, characterized in that: Step 3, the construction and training process of the RBF neural network, includes the following steps: a. Define input data The output data Y = [η1η2η3] is normalized, and training and test sets are constructed. b. Construct an RBF neural network model, set network parameters including the number of training iterations, learning rate, and minimum error of the training objective, and set initial values ​​for weights and biases; c. Train the RBF neural network model using the training set; d. Test the RBF neural network model using the test set; e. Determine whether the training results meet expectations by inverse normalizing the prediction results and comparing them with the true values; f. If the prediction result matches the expected error, end the training, output the weights and biases of the neural network, and obtain the trained optimal controller parameter prediction model; otherwise, update the weights and biases, and repeat the be step until the training is complete.

4. The method as described in claim 3, characterized in that: The process of calculating the voltage reference value in step 4 includes the following steps: a. Input the load torque and target speed into the trained RBF neural network to calculate the corresponding optimal controller parameters η1, η2, and η3; b. Input the optimal controller parameters η1, η2, and η3 into the adaptive integral terminal speed slip mode controller model to calculate the current reference value i. q ref ; c. Set the current reference value i s ref Input a current loop based on deadbeat predictive control to calculate the voltage reference value u. d ref and u q ref .

Images