Predictive Flux Control Method, System and Medium for LC Filter-Type Permanent Magnet Motor System

Abstract

The present invention discloses a predictive flux control method, system and medium for an LC-filtered permanent magnet motor system, belonging to the field of power electronics and electric drive. By sampling the state variables of the LC-filtered permanent magnet motor system at time k, a discrete state space equation of the system is established using the zero-order hold method, and a stator flux vector prediction model is established; the stator flux vector reference is obtained by using the electromagnetic torque reference and the parameters of the permanent magnet motor; a loss function is constructed, and the optimal control vector reference of the inverter considering the system voltage constraint is solved; the optimal control vector reference of the inverter is Park inverse-transformed, and control pulses are generated through space vector modulation to act on the inverter switching tubes, completing the predictive flux control of the LC-filtered permanent magnet motor system. This method can achieve the joint control of torque and flux only by controlling the stator flux vector, and excellent torque and flux control performance can be obtained without any design parameters.

Description

Technical Field

[0001] The present invention discloses a method, a system and a medium for predicting magnetic flux control of an LC-filtered permanent magnet motor system, belonging to the field of power electronics and electric drive. Background Art

[0002] Due to its characteristics such as high efficiency, high power density and compact structure, permanent magnet motors play an important role in the field of industrial drive. However, in the applications of permanent magnet motor drives with long-distance power supply and wide-bandgap power converter feeding, the inverter output voltage has a higher dv / dt, which in turn causes serious overvoltage problems in the permanent magnet motor, greatly reducing the motor life. For this reason, an LC filter is generally added to the output side of the frequency converter in industry to filter out the high dv / dt in the inverter output voltage, forming an LC-filtered permanent magnet motor system. Although this solution can improve electromagnetic compatibility, the introduced LC filter network and the motor stator inductance form a high-order coupling system, increasing the control complexity of the system.

[0003] The prior art with the publication number CN111431460A discloses a sensorless model predictive magnetic flux control method for a permanent magnet motor. First, the motor speed ω and the rotor position angle θ are observed through a sliding mode observer and a phase-locked loop based on SOGI. e ; Next, the given speed ω * and the speed ω pass through a speed loop SMC controller to obtain the given torque T e* ; Then, the load disturbance value is observed from the speed ω and the d / q-axis currents i d / i q and the load disturbance value is fed forward and compensated to the given torque T e* ; Finally, the observed speed ω, rotor position angle θ e , given torque T e* , load disturbance value, and the sampled three-phase voltages u a / u b / u c , three-phase currents i a / i b / i c etc. are substituted into the model predictive magnetic flux control module for calculation. This method has cumbersome prediction steps and a large amount of calculation due to the lack of effective information acquisition means.

[0004] The existing control method for the LC filter type permanent magnet motor system is proportional-integral control. However, the high-order coupling characteristics of the LC filter type permanent magnet motor system increase the cascade loop and control parameters of the proportional-integral control significantly, exacerbate the phase lag, and make it difficult to meet the requirements of dynamic response. Model predictive control has the characteristics of intuitive physical meaning and multi-objective optimization, and is regarded as an effective solution to solve the control problems of high-order systems. The traditional multi-objective predictive control methods for the LC filter type permanent magnet motor system all control the voltage and current loops, without directly controlling the torque and flux linkage, and cannot meet the requirements of high dynamic response of torque and flux linkage in practical engineering. Secondly, most of the existing predictive control methods are based on the finite control set predictive control architecture, which has the disadvantage of unfixed switching frequency, will exacerbate the torque and flux linkage ripple of the motor, and is not conducive to the design of the LC filter. In addition, the existing predictive control methods often need to introduce multiple weight factors to balance multiple control objectives, resulting in problems such as large parameter tuning workload and difficult to guarantee the optimal performance. Therefore, it is urgent to design a predictive control method for the LC filter type permanent magnet motor system that is simple to implement and can directly optimize the flux linkage and torque. Summary of the Invention

[0005] Aiming at the deficiencies of the existing technology, a predictive flux linkage control method for the LC filter type permanent magnet motor system is provided. By establishing a stator flux linkage prediction model of the LC filter type permanent magnet motor system, designing a quadratic cost function based on the stator flux linkage vector reference and the tracking error of the prediction model, and calculating the optimal control command that can minimize the quadratic cost function and meet the inverter voltage constraint, the unified control of torque and flux linkage is realized. The present invention is simple to implement, without any design parameters, and can effectively improve the dynamic and steady-state tracking performance of torque and flux linkage.

[0006] To achieve the above technical objectives, the present invention discloses a predictive flux linkage control method for the LC filter type permanent magnet motor system, and the steps are as follows:

[0007] Use sensors to sample the state variables of the LC filter type permanent magnet motor system at time k, and perform Park transformation on the state variables;

[0008] Adopt the zero-order hold discretization method, and use the state complex vector after Park transformation to construct the discrete state space equation of the LC filter type permanent magnet motor system;

[0009] Based on the discrete state space equation, establish a stator flux linkage vector prediction model of the LC filter type permanent magnet motor system;

[0010] Calculate the electromagnetic torque reference by using the proportional-integral control of the error between the electrical angular velocity reference and the feedback, and obtain the stator flux linkage vector reference by combining the permanent magnet motor information: permanent magnet flux linkage, stator inductance, and number of pole pairs parameters;

[0011] Construct a loss function based on the square of the two-norm of the stator flux vector prediction error by using the error between the stator flux vector reference and the stator flux vector prediction model;

[0012] Calculate the optimal control vector reference of the inverter considering the maximum voltage constraint of the inverter in the LC-filtered permanent magnet motor system by solving the loss function;

[0013] Perform Park inverse transformation on the optimal control vector reference of the inverter, and generate pulses to control the switching tubes of the inverter through space vector modulation to achieve direct control of the torque and flux of the LC-filtered permanent magnet motor.

[0014] Furthermore, the LC-filtered permanent magnet motor system includes a three-phase inverter, an output LC filter, and a permanent magnet motor connected in series. The state variables of the LC-filtered permanent magnet motor system sampled by sensors at time k include: the three-phase filter inductor current i fabc,k , the three-phase filter capacitor voltage v sabc,k , and the three-phase stator current i sabc,k . The electrical angle θ e and the electrical angular velocity ω e of the motor rotor are also detected in real time; the speed loop uses a PI controller, and the proportional coefficient and integral coefficient involved in the PI controller are calculated and tuned through a typical type-II system; the speed loop inputs the speed command ω e * and the error between the actual speed ω e into the internal PI regulator of the speed loop through closed-loop feedback to generate the electromagnetic torque reference value T e * for adjusting the electromagnetic torque; use Park transformation to transform the sampled three-phase state variables into dq-axis state complex vectors: the filter inductor current i f = i fd + ji fq , the filter capacitor voltage v s = v sd + jv sq , and the stator current i s = i sd + ji sq , where j is the imaginary unit of the complex vector, i fd , i fq are the d-axis and q-axis components of the filter inductor current, v sd , v sq are the d-axis and q-axis components of the filter capacitor voltage, i sd , i sq are the d-axis and q-axis components of the stator current.

[0015] Furthermore, Park transformation transforms the abc three-phase state variables into the dq-axis state complex vectors; Park inverse transformation converts the inverter optimal control vector reference in the dq-axis into the inverter optimal control vector reference in the αβ-axis.

[0016] Furthermore, the method for constructing the discrete state space equation of the LC-filtered permanent magnet motor system is as follows:

[0017] ;

[0018] where, , , , ;

[0019] In the formula, is the state matrix composed of the filter inductor current, filter capacitor voltage, and stator current at the k+1 moment, A d , B d , D d are the coefficient matrices of the discrete state space equation calculated by the zero-order hold method, is the state matrix composed of the filter inductor current, filter capacitor voltage, and stator current sampled at the k moment, v i,k is the three-phase inverter output voltage at the k moment, A, B, and D are the coefficient matrices of the LC-filtered permanent magnet motor system in the continuous time domain, e is the base of the natural logarithm, T s is the sampling period, represents the evolution process of the LC-filtered permanent magnet motor system from the current k moment state x k to the k+1 moment state x k+1 , which is also called the state transition matrix; I is the third-order identity matrix, ω e is the electrical angular velocity of the motor, L f represents the filter inductor, C f represents the filter capacitor, L s represents the motor stator inductor, R s represents the motor stator resistance, ψ f represents the permanent magnet flux linkage of the motor, jω e ψ f represents the back electromotive force complex vector of the motor.

[0020] Furthermore, the stator flux linkage vector prediction model established based on the discrete state space equation is as follows:

[0021] ;

[0022] In the formula, represents the predicted value of the stator flux linkage vector composed of the d-axis and q-axis stator flux linkage components and at the k+1 moment, is the coefficient matrix, and j represents the imaginary unit of the complex vector.

[0023] Furthermore, the method for obtaining the stator flux vector reference is as follows:

[0024] ;

[0025] In the formula, represents the stator flux vector reference, and are the real and imaginary parts of the stator flux vector reference, i.e., the d-axis and q-axis stator flux reference components respectively; and are the amplitude and argument of the stator flux vector reference calculated using the permanent magnet flux ψ f , the stator inductance L s , the electromagnetic torque reference T e * and the number of pole pairs p.

[0026] Furthermore, based on the error between the stator flux vector reference and the stator flux vector prediction model, a loss function J based on the square of the two-norm of the stator flux vector prediction error is constructed:

[0027] ;

[0028] In the formula, ψ s * is the stator flux vector reference, represents the predicted value of the stator flux vector at time k + 1, represents the two-norm operation, which is obtained by first calculating the sum of the squares of the real and imaginary parts of the complex vector inside the operator and then taking the square root, equivalent to the magnitude of the complex vector.

[0029] Furthermore, the specific process of obtaining the optimal control vector reference of the inverter in the dq-axis considering the maximum voltage constraint of the LC-filtered permanent magnet motor system by solving the loss function is as follows:

[0030] ;

[0031] where, v i * represents the optimal control vector reference of the inverter that can minimize the loss function J without considering the system voltage constraint, which is obtained by solving the partial derivative of the loss function J with respect to the inverter output voltage and setting the partial derivative equal to 0:

[0032] ;

[0033] In the formula, and is the real and imaginary parts of v i * , and is the two-norm of the reference of the optimal control vector of the inverter, i.e., the amplitude of is the maximum voltage constraint of the system, represents that when the amplitude of v i * is greater than the maximum voltage constraint of the inverter , v will be scaled down proportionally according to i * to obtain the reference of the optimal control vector of the inverter under the dq axis considering the maximum voltage constraint of the inverter . .

[0034] A predictive flux linkage control system for an LC-filtered permanent magnet motor system, comprising a sensor detection unit, a Park transformation unit, and a predictive flux linkage control unit connected in sequence. The predictive flux linkage control unit includes a proportional-integral unit, a stator flux linkage vector prediction model establishment unit, a stator flux linkage vector reference calculation unit, and a loss function calculation unit, an optimal control vector reference unit, a Park inverse transformation unit, and a space vector modulation unit connected in sequence; wherein the proportional-integral unit is connected to the stator flux linkage vector reference calculation unit, and both the stator flux linkage vector reference calculation unit and the stator flux linkage vector prediction model establishment unit are connected to the loss function calculation unit;

[0035] The sensor detection unit is used to obtain the three-phase filter inductor current, the three-phase filter capacitor voltage, and the three-phase stator current, and to detect the electrical angle and electrical angular velocity of the motor rotor in real time;

[0036] The Park transformation unit is used to transform the sampled abc three-phase state variables into dq-axis state complex vectors;

[0037] The proportional-integral unit controls the speed of the permanent magnet motor by proportional-integral of the error between the electrical angular velocity reference and the feedback and calculates the electromagnetic torque reference;

[0038] The stator flux linkage vector prediction model establishment unit is used to establish a stator flux linkage vector prediction model for the LC-filtered permanent magnet motor system;

[0039] The stator flux linkage vector reference calculation unit calculates the stator flux linkage vector reference by using the permanent magnet flux linkage, stator inductance, and number of pole pairs information of the permanent magnet motor collected by the sensor detection unit and combining with the electromagnetic torque reference;

[0040] The loss function calculation unit constructs a loss function based on the square of the two-norm of the stator flux linkage vector prediction error by using the error between the stator flux linkage vector reference and the stator flux linkage vector prediction model;

[0041] An optimal control vector reference unit that obtains the optimal control vector reference of the dq-axis inverter considering the maximum voltage constraint of the inverter in the LC-filtered permanent magnet motor system;

[0042] A Park inverse transformation unit that transforms the optimal control vector reference of the dq-axis inverter into the optimal control vector reference of the αβ-axis inverter;

[0043] A space vector modulation unit that generates pulses to control the switching tubes of the inverter.

[0044] A computer-readable storage medium, characterized in that a computer program is stored in the computer-readable storage medium, and the computer program is adapted to be loaded and executed by a processor to perform the predictive flux control method for the LC-filtered permanent magnet motor system.

[0045] Compared with the prior art, the beneficial effects of the present invention are as follows: For the prior art of the LC-filtered permanent magnet motor system, most are multi-objective voltage and current predictive control, which requires introducing multiple weight coefficients to balance control objectives, resulting in cumbersome weight tuning and poor dynamic and steady-state performance due to the inability to directly control torque and flux; In contrast, the method provided by the present invention can achieve direct control of torque and flux by controlling the stator flux vector, and no design parameters are introduced, the implementation is simple, and the performance is not limited by parameters, so better torque and flux control performance can be obtained. In addition, the method provided by the present invention generates pulses based on space vector modulation. Compared with the existing predictive control method based on a finite control set, it can effectively reduce the steady-state ripple of torque and flux, and the switching frequency is constant, which is more convenient for the design of the LC filter. Description of the Drawings

[0046] Figure 1 It is a structural block diagram of the predictive control method for the LC-filtered permanent magnet motor system of the present invention.

[0047] Figure 2 It is the steady-state response diagram of the stator flux, electromagnetic torque, and stator current of the LC-filtered permanent magnet motor under the method of the present invention.

[0048] Figure 3 It is the steady-state response diagram of the stator flux, electromagnetic torque, and stator current of the LC-filtered permanent magnet motor under the traditional multi-objective predictive control method.

[0049] Figure 4 It is the dynamic response diagram of the stator flux, electromagnetic torque, and stator current of the LC-filtered permanent magnet motor under sudden load under the method of the invention.

[0050] Figure 5 It is the dynamic response diagram of the stator flux, electromagnetic torque, and stator current of the LC-filtered permanent magnet motor under sudden load under the traditional multi-objective predictive control method. Detailed implementation manners

[0051] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention.

[0052] Figure 1 For the LC filter type permanent magnet motor system of the present invention and the corresponding flowchart of the predictive control method for the LC filter type permanent magnet motor system. Among them, the DC bus voltage is converted into an AC voltage square wave signal through a three-phase voltage source inverter, and then connected to the permanent magnet motor through an output LC filter. Sampling is performed on the three-phase filter inductor current i fabc of the LC filter type permanent magnet motor system, the three-phase filter capacitor voltage u sabc and the three-phase motor stator current i sabc is performed. Figure 1 In, V dc represents the DC bus voltage; L f represents the filter inductor on the output side of the inverter, and C f represents the filter capacitor. θ e and ω e are respectively the electrical angle and electrical angular velocity of the motor rotor; the speed loop adopts a PI control method, and the proportional coefficient and integral coefficient involved in the PI controller are calculated and tuned through a typical II system; the speed loop inputs the speed command (ω e * ) and the error of the actual speed (ω e ) into the internal PI regulator of the speed loop through closed-loop feedback to generate an electromagnetic torque reference value (T e * ) for adjusting the electromagnetic torque. The speed loop dynamically converts the speed command into a torque reference command to ensure high-precision speed regulation of the motor; ψ sd * is the set reference value of the stator flux d-axis, and T e * is the electromagnetic torque reference output by the outer speed loop; i f,k , u s,k , i s,k respectively represent the inverter current, capacitor voltage, and stator current sampled in the dq coordinate system at the kth moment; v*ic is the optimal control reference command.

[0053] The predictive flux control method for the LC filter type permanent magnet motor system includes the following steps:

[0054] Step 1: Establish the continuous-time domain state space model x of the LC filter type permanent magnet motor system:

[0055] ;

[0056] In the formula, d represents the differential operator, is the state variable in the continuous-time domain of the LC filter type permanent magnet motor system, is the output voltage of the inverter. The specific expressions of matrices A, B, and D are:

[0057] , , ;

[0058] In the formula, j represents the imaginary unit, L f , C f , L s , R s and ψ f respectively represent: the value of the filter inductor, the value of the filter capacitor, the value of the stator inductor of the motor, the value of the stator resistance of the motor, and the value of the rotor magnetic flux of the motor. jω e ψ f represents the back electromotive force of the permanent magnet motor.

[0059] Step 2: Use the zero-order hold to discretize the continuous-time domain state space model to establish the stator magnetic flux prediction model ψ s,k+1 , which is specifically implemented through the following formula:

[0060] ;

[0061] In the formula, represents the predicted value of the stator magnetic flux at the future k + 1 moment, and the output matrix ; is the state variable matrix composed of the filtered inductor current, the filtered capacitor voltage, and the stator current value of the motor sampled at the k moment, is the predicted value at the k + 1 moment obtained after discretization by the zero-order hold, , are the discrete state matrix and the discrete input matrix respectively, is the constant perturbation matrix, I is the third-order unit diagonal matrix, T s is the sampling period.

[0062] Step 3: Design the quadratic cost function J based on the stator magnetic flux reference tracking, which is implemented through the following formula:

[0063] ;

[0064] In the formula, ψ s * is the stator magnetic flux vector reference, and the specific expression is as follows:

[0065] ;

[0066] In the formula, represents the amplitude of the stator flux vector reference, where ψ sd * is the set d-axis reference value of the stator flux, T e * is the electromagnetic torque reference output by the outer-loop speed loop, and p is the number of pole pairs of the motor.

[0067] Step 4: Solve for the optimal control command v that minimizes the quadratic cost function J and satisfies the inverter voltage constraint ic * designed through the following mathematical model:

[0068] First, let the partial derivative of the quadratic cost function J with respect to the inverter output voltage be equal to 0, i.e.: , then the unconstrained optimal control command v that can minimize J can be obtained i * , v i * The specific expression of is:

[0069] ;

[0070] Secondly, considering the maximum value constraint of the inverter voltage in the actual hardware, the optimal control command v that minimizes the quadratic cost function J and satisfies the inverter voltage constraint ic * is calculated by the following formula:

[0071] ;

[0072] In the formula, is the amplitude of the optimal control command at time k; the maximum inverter voltage constraint V max is one-third of the square root of the inverter DC bus voltage value V dc .

[0073] To verify the predictive flux control method for the LC-filtered permanent magnet motor system provided by the present invention, the method provided by the present invention is applied to the LC-filtered permanent magnet motor system, and the system parameters are given in Table 1:

[0074] Table 1

[0075] .

[0076] The DC bus voltage in Table 1 is a range value, the lower limit voltage corresponding to the speed is 150V, and the highest voltage is freely set within the range allowed by the hardware.

[0077] Figure 2 , Figure 3The figures respectively show the comparison diagrams of the steady-state responses of the stator flux linkage, electromagnetic torque, and stator current of the LC-filtered permanent magnet motor under the method provided by the present invention and the traditional multi-objective predictive control method. By comparing and observing Figure 2 and Figure 3 it can be seen that compared with the traditional multi-objective model predictive control method, the electromagnetic torque and stator flux linkage ripples under the method of the present invention are smaller, the sinusoidality of the stator current is higher, and it has better steady-state performance.

[0078] Figure 4 、 Figure 5 The figures respectively show the comparison diagrams of the dynamic responses of the stator flux linkage, electromagnetic torque, and stator current of the LC-filtered permanent magnet motor under sudden load conditions under the method provided by the invention and the traditional multi-objective predictive control method. By comparing and observing Figure 4 and Figure 5 the waveforms, it can be seen that the dynamic process of the traditional multi-objective model predictive control method has large oscillations and obvious oscillation phenomena; in contrast, the method provided by the present invention has a fast dynamic response. Therefore, the dynamic performance of the method of the present invention is superior to that of the traditional multi-objective model predictive control method.

Claims

1. The predictive flux control method for a permanent magnet motor system with LC filter is characterized in that, The steps are as follows: The state variables of the LC-filtered permanent magnet motor system at time k are sampled by sensors, and Park transformation is performed on the state variables. The LC-filtered permanent magnet motor system includes a three-phase inverter, an output LC filter, and a permanent magnet motor connected in series. The state variables of the LC-filtered permanent magnet motor system sampled by sensors at time k include: the three-phase filter inductor currents i fabc,k 、the three-phase filter capacitor voltages v sabc,k and the three-phase stator currents i sabc,k , and the electrical angle θ e of the motor rotor and the electrical angular velocity ω e are detected in real time. The speed loop uses a PI controller, and the proportional coefficient and integral coefficient involved in the PI controller are calculated and tuned through a typical type II system. The speed loop feeds back the error between the input speed command ω e * and the actual speed ω e into the internal PI regulator of the speed loop to generate a reference value T e * of the electromagnetic torque for adjusting the electromagnetic torque. The sampled three-phase state variables are transformed into dq-axis state complex vectors by Park transformation: the filter inductor current i f = i fd + ji fq , the filter capacitor voltage v s = v sd + jv sq and the stator current i s =i sd + ji sq , where j is the imaginary unit of the complex vector, i fd and i fq are the d-axis and q-axis components of the filter inductor current, v sd and v sq are the d-axis and q-axis components of the filter capacitor voltage, and i sd and i sq are the d-axis and q-axis components of the stator current; Adopt the zero-order hold discretization method, and use the state complex vector after Park transformation to construct the discrete state space equation of the LC-filtered permanent magnet motor system; the method for constructing the discrete state space equation of the LC-filtered permanent magnet motor system is as follows: ； Wherein, ， ， ， ； Wherein, is the state matrix composed of the filter inductor current, filter capacitor voltage, and stator current at the (k + 1)-th moment, A d , B d , D d are the coefficient matrices of the discrete state space equation calculated by the zero-order hold method, is the state matrix composed of the filter inductor current, filter capacitor voltage, and stator current sampled at the k-th moment, v i,k is the three-phase inverter output voltage at the k-th moment, A, B, and D are the coefficient matrices of the LC filter type permanent magnet motor system in the continuous time domain, e is the base of the natural logarithm, T s is the sampling period, represents the evolution process of the LC filter type permanent magnet motor system from the current state x k at the k-th moment to the state x k+1 at the (k + 1)-th moment, which is also called the state transition matrix; I is the third-order identity matrix, ω e is the electrical angular velocity of the motor, L f represents the filter inductor, C f represents the filter capacitor, L s represents the motor stator inductor, R s represents the motor stator resistance, ψ f represents the permanent magnet flux linkage of the motor, jω e ψ f represents the complex back electromotive force vector of the motor; Based on the discrete state space equation, establish the stator flux vector prediction model of the LC-filtered permanent magnet motor system; Calculate the electromagnetic torque reference by using the proportional-integral control of the error between the electrical angular velocity reference and the feedback, and combine the permanent magnet motor information: permanent magnet flux linkage, stator inductance, and number of pole pairs parameters to obtain the stator flux vector reference; Use the error between the stator flux vector reference and the stator flux vector prediction model to construct a loss function based on the square of the two-norm of the stator flux vector prediction error; Calculate the optimal control vector reference of the inverter considering the maximum voltage constraint of the inverter in the LC-filtered permanent magnet motor system through the loss function; Perform the Park inverse transformation on the optimal control vector reference of the inverter, and generate pulse control for the switching tubes of the inverter through space vector modulation to realize the direct control of the torque and flux of the LC-filtered permanent magnet motor.

2. The predictive flux linkage control method for the LC filter type permanent magnet motor system according to claim 1, wherein The Park transformation transforms the abc three-phase state variables into the dq-axis state complex vector; the Park inverse transformation converts the optimal control vector reference of the inverter on the dq-axis into the optimal control vector reference of the inverter on the αβ-axis.

3. The predictive flux linkage control method for the LC filter type permanent magnet motor system according to claim 1, wherein The stator flux vector prediction model established based on the discrete state space equation is as follows: ； In the formula, represents the predicted value of the stator flux vector formed by the d-axis and q-axis stator flux components and at the (k + 1)-th moment, is the coefficient matrix, and j represents the imaginary unit of the complex vector.

4. The predictive flux linkage control method for the LC filter type permanent magnet motor system according to claim 3, characterized in that The method for obtaining the stator flux vector reference is as follows: ； In the formula, represents the stator flux vector reference, and are the real and imaginary parts of the stator flux vector reference, namely the d-axis and q-axis stator flux reference components respectively; and are the amplitude and argument of the stator flux vector reference calculated using the permanent magnet flux ψ f , stator inductance L s , electromagnetic torque reference T e * and the number of pole pairs p respectively.

5. The predictive flux linkage control method for the LC filter type permanent magnet motor system according to claim 4, wherein Use the error between the stator flux vector reference and the stator flux vector prediction model to construct a loss function J based on the square of the two-norm of the stator flux vector prediction error: ； where ψ s * is the reference of the stator flux vector, represents the predicted value of the stator flux vector at the (k + 1)-th moment, represents the two-norm operation, which is obtained by first calculating the sum of the squares of the real and imaginary parts of the complex vector within the operator and then taking the square root, equivalent to the magnitude of the complex vector.

6. The predictive flux linkage control method for an LC-filter type permanent magnet motor system according to claim 5, wherein The optimal control vector reference of the inverter under the dq axis considering the maximum voltage constraint of the LC filter type permanent magnet motor system is calculated by solving the loss function The specific process is as follows: ； Among them, v i * represents the optimal control vector reference of the inverter that can minimize the loss function J without considering the system voltage constraint. It is obtained by solving the partial derivative of the loss function J with respect to the inverter output voltage and setting the partial derivative equal to 0: ； In the formula, and are the real and imaginary parts of v i * , is the two-norm of the reference of the optimal control vector of the inverter, that is, is the amplitude of is the maximum voltage constraint of the system, represents that when the amplitude of v i * is greater than the maximum voltage constraint of the inverter , v is scaled down proportionally according to i * to obtain the reference of the optimal control vector of the inverter under the dq axis considering the maximum voltage constraint of the inverter . .

7. An LC filter type permanent magnet motor system predictive flux control system used in the LC filter type permanent magnet motor system predictive flux control method according to claim 1, characterized in that: It includes a sensor detection unit, a Park transformation unit, and a predicted flux control unit connected in sequence. The predicted flux control unit includes a proportional-integral unit, a stator flux vector prediction model establishment unit, a stator flux vector reference calculation unit, and a loss function calculation unit, an optimal control vector reference unit, a Park inverse transformation unit, and a space vector modulation unit connected in sequence; Among them, the proportional-integral unit is connected to the stator flux vector reference calculation unit, and both the stator flux vector reference calculation unit and the stator flux vector prediction model establishment unit are connected to the loss function calculation unit; The sensor detection unit is used to obtain the three-phase filter inductor current, the three-phase filter capacitor voltage, and the three-phase stator current, and to detect the electrical angle and electrical angular velocity of the motor rotor in real time; The Park transformation unit is used to transform the sampled abc three-phase state variables into the dq-axis state complex vector; The proportional-integral unit controls the speed of the permanent magnet motor by the proportional-integral of the error between the electrical angular velocity reference and the feedback and calculates the electromagnetic torque reference; The stator flux vector prediction model establishment unit is used to establish the stator flux vector prediction model of the LC-filtered permanent magnet motor system; The stator flux vector reference calculation unit calculates the stator flux vector reference by collecting the permanent magnet flux linkage, stator inductance, and number of pole pairs information of the permanent magnet motor through the sensor detection unit and combining the electromagnetic torque reference; The loss function calculation unit uses the error between the stator flux vector reference and the stator flux vector prediction model to construct a loss function based on the square of the two-norm of the stator flux vector prediction error; Optimal control vector reference unit, which obtains the optimal control vector reference of the dq-axis inverter considering the maximum voltage constraint of the inverter of the LC-filtered permanent magnet motor system; Park inverse transformation unit, which is used to transform the optimal control vector reference of the dq-axis inverter into the optimal control vector reference of the αβ-axis inverter; Space vector modulation unit, which is used to generate pulses to control the switching tubes of the inverter.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which is suitable for being loaded and executed by a processor to perform the predictive flux control method of the LC-filtered permanent magnet motor system according to any one of claims 1 to 6.

Images