Permanent magnet synchronous motor speed regulation control method based on improved cerebellar model neural network controller, storage medium and equipment

Abstract

The invention belongs to the technical field of motor speed regulation control, and designs a permanent magnet synchronous motor speed regulation control method based on an improved cerebellum model neural network controller, a storage medium and equipment. Secondly, designing an improved cerebellum model neural network controller to be applied to a speed ring in a vector control structure of the permanent magnet synchronous motor, improving a weight adjustment mode and a learning rate of an original network by the controller, and updating the weight of the cerebellum model neural network by adopting a gradient descent method in combination with a reciprocal relationship of learning times; a calculation mode of dividing by average distribution of a network generalization parameter C is replaced, and the learning efficiency of the network is improved; meanwhile, the value of the network learning rate is dynamically adjusted according to the running state of the motor, and the rapidity and stability of the system are improved. The result shows that the improved cerebellum model neural network controller effectively improves the response speed of the system and greatly improves the dynamic stability of the permanent magnet synchronous motor.

Description

Technical Field

[0001] The present invention relates to the technical field of permanent magnet synchronous motor control, and in particular to a permanent magnet synchronous motor speed regulation control method, storage medium and device based on an improved cerebellar model neural network controller. Background Art

[0002] Permanent magnet synchronous motors (PMSMs) offer numerous advantages, including high efficiency, compact size, and light weight, and are widely used in electric vehicles, household appliances, medical devices, and other fields. Currently, traditional PID control is still widely used in PMS speed regulation systems. However, due to the complex characteristics of PMSs, such as time-varying, nonlinear, and strongly coupled characteristics, PID control parameters are difficult to automatically adjust, making it difficult to achieve ideal control results. This has led to the application of intelligent control methods such as neural network control in PMSMs.

[0003] The Cerebellar Model Articulation Controller (CMAC) is a neural network based on local approximation. It possesses strong generalization and learning capabilities, and its learning speed is faster than other neural networks, making it well-suited for real-time control. CMACs are often used in parallel with PID control in permanent magnet synchronous motor control systems. To a certain extent, they can alleviate the effects of nonlinearity and parameter uncertainty on the control system, thereby improving its stability. However, in conventional CMACs, the error correction value is evenly distributed across all activated address units, potentially causing the network to overlearn, slowing convergence. Furthermore, setting the learning rate too high or too low can affect the controller's learning speed and the motor's operational stability. Summary of the Invention

[0004] In response to the problems existing in the above-mentioned prior art, the present invention provides a permanent magnet synchronous motor speed control method with an improved cerebellum model neural network controller to effectively improve the dynamic performance of the permanent magnet synchronous motor control and the learning speed of the network used.

[0005] The technical solutions adopted by the present invention to solve the technical problems are as follows:

[0006] A permanent magnet synchronous motor speed control method based on an improved cerebellum model neural network controller comprises the following steps:

[0007] S1: Establish the state equation of the permanent magnet synchronous motor under the dq axis of the two-phase rotating coordinate system; set the direct axis current to 0, i d *= 0 vector control strategy, the dq axis current passes through the two PI controllers of the current loop to obtain the dq axis voltage components of the two-phase rotating coordinate system After inverse Park transformation, vector pulse width modulation and inverter, the three-phase current i is obtained a 、i b 、i c Then, the actual output dq axis current i of the motor is obtained through Clark transformation and Park transformation. d 、i q ; According to the actual motor speed n rpm and given speed n ref The difference between the permanent magnet synchronous motor speed tracking error is defined as e(k) = n ref -n rpm , which is fed back to the improved cerebellum model neural network controller of the speed loop to obtain the control output u(k) of the speed loop, and let i q * =u(k), realizing dual closed-loop control of permanent magnet synchronous motor; the improved cerebellar model neural network controller includes two parts: cerebellar model neural network controller and PD controller; in the cerebellar model neural network controller part, the given speed n ref Actual motor speed n rpm As the input of the cerebellum model neural network controller, the output u of the cerebellum model neural network control is obtained through input linear quantization, concept mapping algorithm, cerebellum model neural network and output calculation. nn (k), in the PD controller part, the position PD is used to obtain the output u of the PD control PD (k),u nn (k) and u PD (k) The sum of the two parts of the output is the total output u(k) of the improved cerebellum model neural network controller;

[0008] S2: Update the weights ω of the cerebellum model neural network i and number of learning times f i , the gradient descent method is combined with the inverse relationship of the number of learning times to replace the calculation method of dividing by the average distribution of the network generalization parameter C; at the same time, the value of the network learning rate is dynamically adjusted according to the operating status of the motor to improve the speed and stability of the system;

[0009] S3: Set an acceptable error range Δ, when ||u(k)-u nn When (k)||≥Δ, the weights are continuously updated until the set accuracy value is reached;

[0010] Furthermore, in step S1:

[0011] Without considering the influence of magnetic circuit saturation and core loss of the rotor in the motor, the state equation of the permanent magnet synchronous motor under the dq axis of the two-phase rotating coordinate system is established:

[0012]

[0013] Among them, u d 、u q are the dq axis components of the stator voltage, i d 、i q are the dq axis components of the stator current, R is the stator resistance, ω e is the electrical angular velocity, L d , L q are the dq axis components of the stator inductance, ψ f is the permanent magnet flux, For the derivation operation, T e is the electromagnetic torque, P n is the number of pole pairs of the permanent magnet synchronous motor, J is the moment of inertia, ω m is the rotor mechanical angular velocity, T L is the load torque, B is the damping coefficient;

[0014] Furthermore, the improved cerebellum model neural network controller is specifically as follows:

[0015] The input is linearly quantized as:

[0016]

[0017] Among them, X i 、S i is the i-th dimension input, input quantization value, Q i is the quantization level of the input, X imax 、X imin are the maximum and minimum values ​​of the input of the i-th dimension respectively, and floor() is the rounding down function;

[0018] The concept mapping algorithm is expressed as:

[0019]

[0020] Where r is the address of the input vector in space M, C is the network generalization parameter, ceil(·) is the upward rounding function; k and l represent the mapping dimension; Q l Indicates the quantization level of the l-th dimension input; i indicates the input dimension; S k Represents the k-th dimension input quantization value; S n Represents the quantized value of the n-th dimension input;

[0021] The output u of the cerebellum model neural network control nn (k) is:

[0022]

[0023] Among them, α i is a binary selection vector, C is the network generalization parameter, that is, the number of address units activated by the cerebellum model neural network in one learning cycle, ω i (k) is the weight of the cerebellum model neural network of the i-th address unit at time k;

[0024] Output u of position PD control PD (k) is:

[0025] u PD (k) = K p e(k)+K d [e(k)-e(k-1)](7)

[0026] Among them, K p , K d are the proportional coefficient and differential coefficient of position PD respectively;

[0027] The PD algorithm is used in controller design;

[0028] By adding equation (4) and equation (5), the total output u(k) of the cerebellum model neural network controller is obtained as:

[0029] u(k)=u nn (k)+u PD (k)(8)

[0030] Take i q * =u(k);

[0031] The adjustment index of the cerebellum model neural network is:

[0032]

[0033] Where E(k) represents the network error at time k;

[0034] Furthermore, the process of step S2 is as follows:

[0035] The gradient descent method is combined with the inverse relationship of the number of learning times to update the weights ω of the cerebellum model neural network. i (k) The specific method is:

[0036]

[0037] ω i (k)=ω i (k-1)+Δω i (k) (11)

[0038] Among them, Δω i (k) is the weight correction of the cerebellum model neural network of the i-th address unit at time k; β(k) is the network learning rate at time k, and the learning rate at time zero is β(0), β(k)∈(0,1); σ is the balance learning coefficient, which is used to improve the learning efficiency of the network;

[0039] The value of the network learning rate β(k) is dynamically adjusted according to the motor operating status. The specific implementation method is as follows:

[0040]

[0041] Where ρ is the adaptive gain adjustment coefficient, ρ>0, ω e (k) is the electrical angular velocity of the motor rotor at time k, T L (k), T L (k-1) represents the load torque of the motor at time k and time k-1 respectively;

[0042] The weight ω in the cerebellum model neural network i (k) and number of learning times f i The initial values ​​are all set to 0; the maximum number of learning times is set to f max , when f i <f max When f i Add 1;

[0043] Furthermore, the process of step S3 is as follows:

[0044] Set an acceptable error range Δ, specifically:

[0045]

[0046] Among them, i qmax is the maximum expected current of the q-axis;

[0047] A computer-readable storage medium, wherein the storage medium stores at least one instruction, at least one program, code set, or instruction set, wherein the at least one instruction, at least one program, code set, or instruction set is loaded and executed by the processor to implement the above-mentioned permanent magnet synchronous motor speed control method based on the improved cerebellar model neural network controller;

[0048] A server includes a processor and a memory, wherein the memory stores at least one instruction, and the instruction is loaded and executed by the processor to implement the above-mentioned permanent magnet synchronous motor speed control method based on the improved cerebellum model neural network controller.

[0049] Beneficial effects of the present invention:

[0050] This method rationally distributes weight correction errors based on the number of learning times, or confidence levels, of each activated address unit, avoiding an even distribution of weight corrections and improving network learning efficiency. Furthermore, a balanced learning coefficient is introduced to further optimize the network structure. Finally, the network learning rate is adaptively processed, enabling the motor to maintain good dynamic performance under various complex operating conditions. This method combines a classic PID controller with a cerebellar model neural network controller. The cerebellar model neural network controller implements feedforward control, creating an inverse dynamic model of the controlled object, while the PID controller implements feedback control, ensuring system stability while suppressing disturbances. The design process is simple and efficient. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 is the speed-current dual closed-loop vector control framework of the permanent magnet synchronous motor, where u α 、u β is the voltage component of the α-β axis of the stationary coordinate system, i α 、i β is the α-β axis current component of the stationary coordinate system, θ e is the electrical angle of the motor rotor, and 1 / S is the integral term.

[0052] Figure 2 It is a schematic diagram of the structure of the improved cerebellum model neural network controller of the present invention.

[0053] Figure 3 The figure shows the comparison curves of the motor speed under sudden load increase or decrease when the speed loop adopts PI control, conventional CMAC control, model-free sliding mode control (MFSMC) and improved cerebellar model neural network control (MC-CMAC).

[0054] Figure 4 These are the motor sudden load increase or decrease torque response curves of the speed loop under PI control, conventional CMAC control, model-free sliding mode control (MFSMC) and improved cerebellar model neural network (MC-CMAC). DETAILED DESCRIPTION

[0055] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0056] like Figure 1 As shown in FIG, the permanent magnet synchronous motor control realizes high performance control of the permanent magnet synchronous motor by improving the design of the speed loop controller.

[0057] Figure 2 It is a schematic diagram of the structure of the improved cerebellar model neural network controller (MC-CMAC) of the present invention, which includes two parts: cerebellar model neural network control and PD control.

[0058] In this embodiment, a MATLAB simulation model is used to simulate a permanent magnet synchronous motor speed control method based on an improved cerebellar model neural network of the present invention.

[0059] The MATLAB simulation model parameters are shown in Table 1 below:

[0060] Table 1 Simulation model parameters

[0061]

[0062] The speed loop adopts PI control, conventional CMAC control, model-free sliding mode control (MFSMC) and the improved cerebellar model neural network control (MC-CMAC) of the present invention, and the motor sudden increase or decrease load speed comparison curve is shown as follows: Figure 3 As shown, the torque response curve is as follows Figure 4 shown.

[0063] from Figure 3 As can be seen in the figure, the PI control motor speed overshoots by 141 r / min at startup. After a sudden load increase, the motor speed drops to 148 r / min, with a speed recovery time of 0.22 s. After a sudden load reduction, the speed changes by 73 r / min, with a speed recovery time of 0.18 s. Under MFSMC control, the motor speed does not overshoot at startup. After sudden load increases and decreases, the speed changes by 63 r / min and 20 r / min, respectively, with speed recovery times of 0.06 s and 0.05 s, respectively. Under CMAC control, the motor speed does not overshoot at startup. After sudden load increases and decreases, the speed changes by 58 r / min and 30 r / min, respectively, with speed recovery times of 0.06 s and 0.05 s, respectively. Under the MC-CMAC control method, the system speed does not overshoot after no-load startup and reaches the set speed faster than CMAC and MFSMC. After sudden load increases and decreases, the speed changes by 38 r / min and 20 r / min, respectively, with speed recovery times of 0.03 s and 0.03 s, respectively, showing the shortest speed recovery time.

[0064] from Figure 4 It can be seen from the figure that when the load is suddenly increased or decreased, the torque recovery time of the motor controlled by the PI method is 0.2s and 0.16s respectively, the torque recovery time under MFSMC control is 0.06s and 0.04s, the torque recovery time under CMAC control is 0.07s and 0.06s, and the torque recovery time under MC-CMAC control is 0.03s and 0.028s. The MC-CMAC method recovers to the pre-disturbance state faster and shows better ability to resist load disturbances.

Claims

1. A permanent magnet synchronous motor speed control method based on an improved cerebellum model neural network controller, characterized in that: The following steps are involved: S1: Establish the state equation of the permanent magnet synchronous motor under the dq axis of the two-phase rotating coordinate system; set the direct axis current to 0, i d * = 0 vector control strategy, the dq axis current passes through the two PI controllers of the current loop to obtain the dq axis voltage components of the two-phase rotating coordinate system After inverse Park transformation, vector pulse width modulation and inverter, the three-phase current i is obtained a 、i b 、i c Then, the actual output dq axis current i of the motor is obtained through Clark transformation and Park transformation. d 、i q ; According to the actual motor speed n rpm and given speed n ref The difference between the permanent magnet synchronous motor speed tracking error is defined as e(k) = n ref -n rpm , which is fed back to the improved cerebellum model neural network controller of the speed loop to obtain the control output u(k) of the speed loop, and let i q * =u(k), realizing dual closed-loop control of permanent magnet synchronous motor; the improved cerebellar model neural network controller includes two parts: cerebellar model neural network controller and PD controller; in the cerebellar model neural network controller part, the given speed n ref Actual motor speed n rpm As the input of the cerebellum model neural network controller, the output u of the cerebellum model neural network control is obtained through input linear quantization, concept mapping algorithm, cerebellum model neural network and output calculation. nn (k), in the PD controller part, the position PD is used to obtain the output u of the PD control PD (k),u nn (k) and u PD (k) The sum of the two parts of the output is the total output u(k) of the improved cerebellum model neural network controller; S2: Update the weights ω of the cerebellum model neural network i and number of learning times f i , the gradient descent method is combined with the inverse relationship of the number of learning times to replace the calculation method of dividing by the average distribution of the network generalization parameter C; at the same time, the value of the network learning rate is dynamically adjusted according to the operating status of the motor to improve the speed and stability of the system; S3: Set an acceptable error range Δ, when ||u(k)-u nn When (k)||≥Δ, the weights are continuously updated until the set accuracy value is reached.

2. The permanent magnet synchronous motor speed control method based on the improved cerebellum model neural network controller according to claim 1 is characterized in that: In the step S1: Without considering the influence of magnetic circuit saturation and core loss of the rotor in the motor, the state equation of the permanent magnet synchronous motor under the dq axis of the two-phase rotating coordinate system is established: Among them, u d 、u q are the dq axis components of the stator voltage, i d 、i q are the dq axis components of the stator current, R is the stator resistance, ω e is the electrical angular velocity, L d , L q are the dq axis components of the stator inductance, ψ f is the permanent magnet flux, For the derivation operation, T e is the electromagnetic torque, P n is the number of pole pairs of the permanent magnet synchronous motor, J is the moment of inertia, ω m is the rotor mechanical angular velocity, T L is the load torque, and B is the damping coefficient.

3. The permanent magnet synchronous motor speed control method based on the improved cerebellum model neural network controller according to claim 1 is characterized in that: The improved cerebellum model neural network controller is specifically as follows: The input is linearly quantized as: Among them, X i 、S i is the i-th dimension input, input quantization value, Q i is the quantization level of the input, X imax 、X imin are the maximum and minimum values ​​of the input of the i-th dimension respectively, and floor() is the rounding down function; The concept mapping algorithm is expressed as: Where r is the address of the input vector in space M, C is the network generalization parameter, ceil(·) is the upward rounding function; k and l represent the mapping dimension; Q l Indicates the quantization level of the l-th dimension input; i indicates the input dimension; S k Represents the k-th dimension input quantization value; S n Represents the quantized value of the n-th dimension input; The output u of the cerebellum model neural network control nn (k) is: Among them, α i is a binary selection vector, C is the network generalization parameter, that is, the number of address units activated by the cerebellum model neural network in one learning cycle, ω i (k) is the weight of the cerebellum model neural network of the i-th address unit at time k; Output u of position PD control PD (k) is: u PD (k)=K p e(k)+K d [e(k)-e(k-1)](7) Among them, K p , K d are the proportional coefficient and differential coefficient of position PD respectively; The PD algorithm is used in controller design; By adding equation (4) and equation (5), the total output u(k) of the cerebellum model neural network controller is obtained as: u(k)=u nn (k)+u PD (k) (8) Take q * = u(k); The adjustment index of the cerebellum model neural network is: Among them, E(k) represents the network error at time k.

4. The permanent magnet synchronous motor speed control method based on the improved cerebellum model neural network controller according to claim 1 is characterized in that: The process of step S2 is as follows: The gradient descent method is combined with the inverse relationship of the number of learning times to update the weights ω of the cerebellum model neural network. i (k) The specific method is: oh i (k)=ω i (k-1)+Do i (k) (11) Among them, Δω i (k) is the weight correction of the cerebellum model neural network of the i-th address unit at time k; β(k) is the network learning rate at time k, and the learning rate at time zero is β(0), β(k)∈(0,1); σ is the balance learning coefficient, which is used to improve the learning efficiency of the network; The value of the network learning rate β(k) is dynamically adjusted according to the motor operating status. The specific implementation method is as follows: Where ρ is the adaptive gain adjustment coefficient, ρ>0, ω e (k) is the electrical angular velocity of the motor rotor at time k, T L (k), T L (k-1) represents the load torque of the motor at time k and time k-1 respectively; The weight ω in the cerebellum model neural network i (k) and number of learning times f i The initial values ​​are all set to 0; the maximum number of learning times is set to f max , when f i <f max When f i Add 1.

5. The permanent magnet synchronous motor speed control method based on the improved cerebellar model neural network according to claim 1 is characterized in that: The process of step S3 is as follows: Set an acceptable error range Δ, specifically: Among them, i qmax is the maximum expected current of the q-axis.

6. A computer-readable storage medium, characterized in that The storage medium stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set are loaded and executed by the processor to implement the permanent magnet synchronous motor speed control method based on the improved cerebellar model neural network controller described in any one of claims 1-5.

7. A server, characterized in that: The server includes a processor and a memory, wherein the memory stores at least one instruction, and the instruction is loaded and executed by the processor to implement the permanent magnet synchronous motor speed control method based on the improved cerebellum model neural network controller described in any one of claims 1-5.

Images