Permanent magnet synchronous motor current prediction control method based on parameter identification

Abstract

The invention is suitable for the technical field of permanent magnet synchronous motors, and provides a permanent magnet synchronous motor current prediction control method based on parameter identification, which comprises the following steps: constructing a permanent magnet synchronous motor current prediction control model; performing parameter sensitivity analysis based on the constructed current prediction control model; and according to a parameter sensitivity analysis result of the current prediction control model, carrying out online identification on parameters of the permanent magnet synchronous motor by adopting improved extended Kalman filtering, sending the identified parameters to the current prediction controller in real time, and updating the parameters in the current prediction controller. And the current error of the current prediction control of the permanent magnet synchronous motor during parameter mismatch is reduced. According to the permanent magnet synchronous motor current prediction control method based on improved extended Kalman filtering parameter identification provided by the invention, the identification precision of extended Kalman filtering on motor parameters is improved, so that the current prediction control performance of the permanent magnet synchronous motor is effectively improved under the condition that the motor parameters are mismatched.

Description

Technical Field

[0001] The present invention belongs to the technical field of permanent magnet synchronous motors, and particularly relates to a current predictive control method for permanent magnet synchronous motors based on parameter identification. Background Technique

[0002] Given the large scale of large-capacity thermal power units in China, energy conservation and emission reduction of existing units have become a key transitional strategy. The coal grinding systems of many thermal power units have been upgraded from traditional asynchronous motors to permanent magnet synchronous motors with low speed and large torque as the drive. In order to more effectively apply permanent magnet synchronous motors to coal mills, it is necessary to improve the requirements for the control accuracy of permanent magnet synchronous motors. Among them, the control performance of the inner current loop of the motor is particularly critical, which directly affects the dynamic response speed and steady-state control accuracy of the permanent magnet synchronous motor. Therefore, selecting an efficient current loop control method is crucial for achieving precise control of permanent magnet synchronous motors. Current predictive control has been increasingly widely used in the field of motor control due to its small computational load and excellent dynamic and steady-state performance. However, as a typical model-based method, current predictive control inevitably has the problem of dependence on model parameters. When the temperature of the coal mill in a thermal power plant rises or falls during operation, the model parameters of the motor do not match the actual motor parameters, which in turn leads to a decline in the control performance of the entire system.

[0003] Aiming at the problem that the current predictive control of permanent magnet synchronous motors affects the control performance under parameter mismatch, the present invention proposes a current predictive control method for permanent magnet synchronous motors based on parameter identification. Summary of the Invention

[0004] The purpose of the present invention is to provide a current predictive control method for permanent magnet synchronous motors based on parameter identification, aiming to solve the problems raised in the above background technique.

[0005] The purpose of the present invention is achieved through the following technical solutions: A current predictive control method for permanent magnet synchronous motors based on parameter identification includes the following steps: Step 1: Construct a current predictive control model for the permanent magnet synchronous motor; Step 2: Based on the constructed current predictive control model, perform parameter sensitivity analysis; Step 3: According to the results of the parameter sensitivity analysis of the current predictive control model, use an improved extended Kalman filter to perform online identification of the parameters of the permanent magnet synchronous motor, and send the identified parameters to the current predictive controller in real time and update the parameters in the current predictive controller.

[0006] Further, the specific process of Step 1 is as follows: Step 1.1: The current state equation of the permanent magnet synchronous motor with the stator current as the state variable is: ; In the formula: is the voltage vector component of the d-axis; is the voltage vector component of the q-axis; R is the resistance; is the current vector component of the d-axis; is the current vector component of the q-axis; is the inductance; is the electrical angular velocity; is the magnetic flux linkage; Step 1.2: Predict the current at the next moment according to the forward Euler formula and , and the state equation is specifically discretized into the following equation: ; In the formula: is the sampled value of the d-axis current at moment; is the sampled value of the q-axis current at moment; is the sampled value of the d-axis current at moment; is the sampled value of the q-axis current at moment; is the electrical angular velocity at the th moment; is the sampling period; The principle of current predictive control is to make the actual current value at moment track the current given value at moment within a switching period, that is , where the given current is zero, is the output of the speed loop; Thus, the output voltage vector equation of current predictive control is expressed as follows: ; In the formula: is the given value of the d-axis voltage at moment; is the given value of the q-axis voltage at moment.

[0007] Furthermore, the specific process of step 2 is as follows: Step 2.1: Considering the parameter perturbation in the model, the current prediction model is expressed as: ; In the formula: is the sampled value of the d-axis current considering the parameter perturbation at moment; is the sampled value of the q-axis current considering the parameter perturbation Sampling value of q-axis current at a moment Error between actual resistance and model resistance Error between actual inductance and model inductance Error between actual flux linkage and model flux linkage Step 2.2: Subtract the ideal current prediction model in Step 1.2 from the actual current prediction model in Step 2.1 to calculate the error of the predicted current: ; In the formula, Represents the error of the d-axis predicted current in current predictive control; Represents the error of the q-axis predicted current in current predictive control; Finally, the three motor parameters are obtained, and the control effects from high to low are > > .

[0008] Furthermore, the specific process of Step 3 is as follows: Step 3.1: Construct an improved extended Kalman filter parameter identification system: First, regard the resistance as a constant and set: ; In the formula: Is the state variable matrix; Is the output variable matrix; Obtain the improved extended Kalman filter parameter identification system: ; In the formula: Is a function of the state variable ; Is the state noise of the system model, Is the measurement noise of the system; Step 3.2: Design the state prediction step of the improved extended Kalman filter parameter identification to obtain the prior estimate value: ; In the formula: Is the Prior estimate value at time; Is the Optimal estimate value at time; Is the State process from time to Time; Step 3.3: Design the covariance prediction value of the improved extended Kalman filter parameter identification: ; Wherein: is the prior error covariance prediction matrix; is the state transition matrix of the parameter identification system; is the error covariance update matrix at time is the measurement noise covariance; Step 3.4: Calculate the Kalman gain for the improved extended Kalman filter parameter identification: ; where is the Kalman gain; is the Jacobian matrix, that is, the coefficient matrix of the state variables of the function ; Step 3.5: Calculate the optimal estimated value of the improved extended Kalman filter parameter identification: ; Wherein: is the optimal estimated value; i is the number of matrix summations from 1 to p ; the matrix length ; is the extended observation deviation model; is the extended state gain matrix; is the weight matrix; Obtain the optimal estimated value of the identified parameter: ; Wherein: is the finally obtained estimated value of the state variable; is the estimated value of the d-axis current; is the estimated value of the q-axis current, is the estimated value of the reciprocal of the inductance; is the estimated value of the magnetic flux linkage; According to the current parameter sensitivity analysis in the foregoing, it is pointed out that the inductance has the greatest influence on the current. Therefore, the present invention only needs the estimated value of the inductance , that is, take the reciprocal of; ; Step 3.6: Calculate the covariance update value of the improved extended Kalman filter parameter identification: ; Wherein: is the error covariance update matrix, which is used for the prediction of the prior error covariance in the next time.

[0009] Compared with the prior art, the beneficial effects of the present invention are: The present invention proposes a current predictive control method for a permanent magnet synchronous motor based on improved extended Kalman filter parameter identification. This method improves the identification accuracy of the extended Kalman filter for motor parameters, enabling effective improvement of the current predictive control performance of the permanent magnet synchronous motor in the case of motor parameter mismatch. The method proposed by the present invention can effectively suppress the problem of unstable response current of the permanent magnet synchronous motor caused by control parameter mismatch. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 It is a flowchart of the method of the present invention.

[0011] Figure 2 It is the result of inductance based on traditional extended Kalman filter parameter identification of.

[0012] Figure 3 It is the result of inductance based on improved extended Kalman filter parameter identification of.

[0013] Figure 4 It is the result of current predictive control when the parameters are accurate of.

[0014] Figure 5 It is the result of current predictive control when the model inductance parameter is twice the actual inductance parameter of.

[0015] Figure 6 It is the result of current predictive control of a permanent magnet synchronous motor based on improved extended Kalman filter parameter identification when the model inductance parameter is twice the actual inductance parameter of. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0016] In order to have a clearer understanding of the technical features, objectives, and beneficial effects of the present invention, the technical solutions of the present invention are described in detail below, but it should not be construed as a limitation on the implementable scope of the present invention.

[0017] The following describes the specific implementation of the present invention in detail with reference to specific embodiments.

[0018] Embodiment 1: This embodiment provides a current predictive control method for a permanent magnet synchronous motor based on parameter identification. The flowchart is as Figure 1 shown, and the method includes the following steps: Step 1: Construct a current predictive control model for the permanent magnet synchronous motor; Step 2: Based on the constructed current predictive control model, perform parameter sensitivity analysis; Step 3: According to the results of the parameter sensitivity analysis of the current prediction control model, the improved extended Kalman filter is used to online identify the parameters of the permanent magnet synchronous motor, and the identified parameters are sent to the current prediction controller in real time to update the parameters in the current prediction controller. Thereby, the current error of the permanent magnet synchronous motor current prediction control when there is parameter mismatch is reduced.

[0019] Specifically as follows: 1) Step 1 includes: Step 1.1: In the synchronous rotating coordinate system, the voltage equation of the permanent magnet synchronous motor is: ; In the formula: is the voltage vector component of the d-axis; is the voltage vector component of the q-axis; R is the resistance; is the current vector component of the d-axis; is the current vector component of the q-axis; is the inductance; is the electrical angular velocity, which is the product of the mechanical angular velocity and the number of pole pairs; is the magnetic flux linkage; Selecting the stator current as the state variable, the state equation of the permanent magnet synchronous motor can be obtained as follows: .

[0020] Step 1.2: According to the forward Euler formula, the current and at the next moment can be predicted, and thus the state equation is specifically discretized into the following equation: ; In the formula: is the sampling value of the d-axis current at moment; is the sampling value of the q-axis current at moment; is the sampling value of the d-axis current at moment; is the sampling value of the q-axis current at moment; is the electrical angular velocity at the th moment; is the sampling period. The principle of current prediction control is to make the actual current value at moment track the current given value at moment within a switching period, that is, , where the given current is zero, is the output of the speed loop; thus, the output voltage vector equation of current prediction control is expressed as follows: ; where: is the d-axis voltage reference value at time is the q-axis voltage reference value at time

[0021] 2) According to the theory of current predictive control, the prediction model of the current predictive controller covers three motor parameters: resistance , inductance and flux linkage . When there are deviations in the model parameters, it may lead to inaccurate predicted current. If there is an error in the predicted current, this will further affect the accuracy of the voltage vector, thus having an adverse impact on the control performance of the system. Next, we will conduct an in-depth sensitivity analysis of current predictive control on the premise of parameter mismatch. Considering that the focus of the research is the current loop, only the dynamic response and stability of the d-q axis current are deeply explored. Step 2 includes: Step 2.1: If parameter perturbation existing in the model is considered, the current prediction model can be expressed as: ; where: is the sampled value of the d-axis current at time considering parameter perturbation is the sampled value of the q-axis current at time considering parameter perturbation is the error between the actual resistance value and the model resistance value; is the error between the actual inductance and the model inductance; is the error between the actual flux linkage and the model flux linkage.

[0022] Step 2.2: Subtract the ideal current prediction model in Step 1.2 from the actual current prediction model in Step 2.1 to calculate the error in the predicted current: ; where, represents the error in the d-axis predicted current in current predictive control; represents the error in the q-axis predicted current in current predictive control; under the control strategy of =0, the d-axis current error is not affected by the resistance, and the q-axis current error is less affected by the resistance. And the inductance error has the greatest impact on the predicted current errors and . On the other hand, the flux linkage only appears in the q-axis current prediction model, so it only affects the prediction of the q-axis current. To sum up, the order of the three motor parameters on the control effect from high to low is > > 。

[0023] 3) Step 3 includes: Step 3.1: Construct an improved extended Kalman filter parameter identification system: According to Step 1.1, the stator voltage state equation of the surface-mounted permanent magnet synchronous motor in the synchronous rotating coordinate system: ; To more conveniently construct the state transition matrix and avoid the increase in computational complexity caused by the increase in matrix dimension, the resistance is temporarily regarded as a constant here, and it is set that: ; In the formula: is the state variable matrix; is the output variable matrix.

[0024] The improved extended Kalman filter parameter identification system is obtained: ; In the formula: is a function of the state variable ; is the state noise of the system model, is the measurement noise of the system.

[0025] Step 3.2: Design the state prediction step of the improved extended Kalman filter parameter identification to obtain the prior estimate value: ; In the formula: is the prior estimate value at time; is the optimal estimate value at time; is the state process from time to time. Given the initial values and : ; .

[0026] Step 3.3: Design the covariance predicted value of the improved extended Kalman filter parameter identification: ; In the formula: is the prior error covariance prediction matrix; is the state transition matrix of the parameter identification system; is Error covariance update matrix at a moment; is the measurement noise covariance; select the matrix initial value under the principle of ensuring steady-state tracking and non-divergence of filtering , and the variance matrix , as follows: ; ; In the formula: is the identity matrix; is to take the partial derivative of the state variable function .

[0027] Step 3.4: Calculate the Kalman gain for improved extended Kalman filter parameter identification: ; In the formula: is the Kalman gain; is the Jacobian matrix, that is, the coefficient matrix of the state variable of the function ; According to the improved extended Kalman filter parameter identification system, it can be seen that the Jacobian matrix in the observation equation and given the state noise covariance is: .

[0028] Step 3.5: Calculate the optimal estimated value of improved extended Kalman filter parameter identification: ; In the formula: is the optimal estimated value; is the matrix length; i is the matrix summation times from 1 to p . First, select the matrix length ; design the extended observation deviation model at : ; At the same time, design the improved extended state gain matrix at : ; Similarly, define the weight matrix : ; Through the optimal estimated value of the identified parameter can be obtained: ; In the formula: is the estimated value of the final state variable; is the estimated value of the d-axis current; is the estimated value of the q-axis current, is the estimated value of the reciprocal of the inductance; is the estimated value of the magnetic flux linkage.

[0029] According to the current parameter sensitivity analysis in the previous text, it is pointed out that the inductance has the greatest impact on the current. Therefore, only the inductance is required in this paper , that is, take the reciprocal of; Step 3.6: Calculate the covariance update value for the parameter identification of the improved extended Kalman filter: ; In the formula: is the error covariance update matrix, which is used in the prediction of the prior error covariance for the next time.

[0030] Example 2: This example provides a simulation experiment based on the method of the present invention. According to the parameter sensitivity analysis of the permanent magnet synchronous motor in the previous text, the inductance has the greatest impact on the current error. Therefore, this experiment only considers the comparative experiment on the online parameter identification of the inductance . The specific parameters of this simulation experiment are as follows: Table 1 Simulation experiment parameters

[0031] Set the working condition: the initial value of the rotational speed is 200 / (r min -1 ), and start with a 100N load applied. To verify the accuracy of the parameter identification strategy, the method of the present invention is used for comparative simulation research with the parameter identification method of the traditional extended Kalman filter.

[0032] As Figure 2 and Figure 3 shown, it can be seen that the inductance value identified by the improved extended Kalman filter parameter identification method is 0.004279, while the inductance value identified by the traditional extended Kalman filter parameter identification method is 0.00431, and the actual inductance value is 0.00428. This shows that the parameters of the inductance identified by the method of the present invention are significantly better than the parameters identified by the parameter identification method of the traditional extended Kalman filter.

[0033] Example 3: This example provides another simulation experiment based on the method of the present invention. The specific parameters of this simulation experiment are the same as those in Example 2.

[0034] Set the working condition: when the model inductance parameter is twice the actual inductance parameter, the initial value of the rotational speed is 200 / (r min -1 ), and apply a 100 N load to start with load. To verify the accuracy of the parameter identification strategy, a comparative simulation study is carried out using the method of the present invention and the traditional extended Kalman filter permanent magnet synchronous motor current prediction control method.

[0035] As Figure 4 , Figure 5 and Figure 6 shown, it can be seen that the current prediction control with the improved extended Kalman filter parameter identification reduces the current error by 14%. This shows that the method of the present invention can effectively reduce the current error caused by parameter mismatch, so that the permanent magnet synchronous motor has better control performance.

[0036] The above is only the preferred embodiment of the present invention. It should be pointed out that for those skilled in the art, without departing from the concept of the present invention, several deformations and improvements can be made, which should also be regarded as the protection scope of the present invention, and these will not affect the implementation effect of the present invention and the practicality of the patent.

Claims

1. A method for predictive control of permanent magnet synchronous motor current based on parameter identification, characterized in that: The following steps are involved: Step 1: Construct a current prediction control model for a permanent magnet synchronous motor; Step 2: Perform parameter sensitivity analysis based on the constructed current predictive control model; Step 3: According to the results of parameter sensitivity analysis of the current prediction control model, the improved extended Kalman filter is used to perform online identification of the parameters of the permanent magnet synchronous motor, and the identified parameters are sent to the current prediction controller in real time and the parameters in the current prediction controller are updated.

2. The method for predicting and controlling the current of a permanent magnet synchronous motor based on parameter identification according to claim 1, characterized in that: The specific process of step 1 is as follows: Step 1.1: The current state equation of the permanent magnet synchronous motor with stator current as the state variable is: ； Where: is the voltage vector component of the d-axis; is the voltage vector component of the q axis; R is the resistor; is the current vector component of the d-axis; is the current vector component of the q axis; is the inductor; is the electrical angular velocity; is the magnetic link; Step 1.2: Predict the current at the next moment based on the forward Euler formula and , the state equation is discretized into the following equation: ； Where: for The d-axis current sampling value at the moment; for The q-axis current sampling value at the moment; for The d-axis current sampling value at the moment; for The q-axis current sampling value at the moment; For the The electrical angular velocity at the moment; is the sampling period; the principle of current prediction control is to use The actual current value is tracked at all times The current given value at the moment is , where the given current is zero, is the output of the speed loop; thus, the output voltage vector equation of the current prediction control is expressed as follows: ； Where: for The d-axis voltage given value at the moment; for The q-axis voltage given value at the moment.

3. The method for predicting and controlling current of a permanent magnet synchronous motor based on parameter identification according to claim 2 is characterized in that: The specific process of step 2 is as follows: Step 2.1: Considering the parameter perturbation in the model, the current prediction model is expressed as: ； Where: To consider parameter perturbations The d-axis current sampling value at the moment; To consider parameter perturbations The q-axis current sampling value at the moment; is the error between the actual resistance value and the model resistance value; is the error between the actual inductance and the model inductance; is the error between the actual magnetic flux and the model magnetic flux; Step 2.2: Subtract the ideal current prediction model from step 1.2 from the actual current prediction model from step 2.1 to calculate the error in the predicted current: ； In the formula, Represents the error in the d-axis predicted current in current prediction control; Represents the error in the q-axis predicted current in current predictive control; Finally, the control effects of the three motor parameters from high to low are: > > .

4. The method for predicting and controlling current of a permanent magnet synchronous motor based on parameter identification according to claim 1, characterized in that: The specific process of step 3 is as follows: Step 3.1: Construct an improved extended Kalman filter parameter identification system: First treat the resistance as a constant and set: ； Where: is the state variable matrix; is the output variable matrix; The improved extended Kalman filter parameter identification system is obtained: ； Where: For state variables Function of is the state noise of the system model, is the measurement noise of the system; Step 3.2: Design the state prediction step of the improved extended Kalman filter parameter identification and obtain the prior estimate: ； Where: for A priori estimate of the moment; for The best estimate of the time; for Time has come The state process at each moment; Step 3.3: Design the covariance prediction value for improved extended Kalman filter parameter identification: ； Where: is the prior error covariance estimation matrix; is the state transfer matrix of the parameter identification system; for The error covariance update matrix at time; is the measurement noise covariance; Step 3.4: Calculate the Kalman gain for improved extended Kalman filter parameter identification: ； in is the Kalman gain; is the Jacobian matrix, i.e., the function The coefficient matrix of the state variables; Step 3.5: Calculate the optimal estimate of the improved extended Kalman filter parameter identification: ； in: is the optimal estimate; i From 1 to p The number of matrix summations of ; Matrix length ; To extend the observation bias model; is the extended state gain matrix; is the weight matrix; Get the optimal estimate of the identification parameters: ； Where: is the estimated value of the final state variable; is the estimated value of d-axis current; is the estimated value of q-axis current, is the estimated value of the inverse of the inductance; is the estimated value of magnetic linkage; According to the inductance The biggest impact on current, only inductance is needed Estimated value of , that is, take The reciprocal of Step 3.6: Calculate the covariance update value for the improved extended Kalman filter parameter identification: ； Where: is the error covariance update matrix, which is used to predict the next prior error covariance.

Images