Multi-step model predictive current control method with variable prediction step size

Abstract

This disclosure provides a multi-step model predictive current control method with variable prediction step size, including: Step 1, obtaining torque reference value. T e ref Step 2, obtain the current based on the principle of no deadbeat ( k + N Reference voltage vector of the control cycle u s Nref Step 3: Calculate the voltage vector application time corresponding to the two non-zero voltage vectors and one zero voltage vector required to synthesize the reference voltage vector; Step 4: Construct a candidate voltage vector sequence set; Step 5: Select... N Step 6: Construct the optimal application vector within each control cycle; Step 7: Construct the drive signal for each switch of the inverter in the next control cycle; Step 8: Update the average switching frequency of the switches in the next cycle, construct a proportional-integral regulator using the difference between the reference switching frequency and the actual switching frequency as input, and update the prediction step size for the next cycle. N The method disclosed herein reduces the computational burden, enhances the steady-state performance of the system, and enables online adjustment of the prediction step size.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the field of induction motor speed control, and more particularly, to a variable step length multi-step model predictive current control method. BACKGROUND

[0002] Multi-step model predictive control (MPC) is an effective means to improve system control accuracy and reduce switching frequency of traditional model predictive control. Multi-step MPC predicts the trajectory of the controlled variable in a longer time scale, optimizes the switching state sequence applied to the inverter, and realizes the goal of limiting the controlled variable within the hysteresis control range with the minimum switching state switching times. The calculation amount increases exponentially with the prediction step length.

[0003] To reduce system complexity and calculation amount, some people have proposed a multi-step MPC method based on spherical decoding, which greatly reduces the online optimization range of the optimal solution, but is still relatively complex and difficult to implement online on existing digital signal processor (DSP) platforms. Some scholars have reduced the range of candidate vectors to reduce the computational burden. For example, a new voltage vector sequence distribution space diagram is established in the finite control set model predictive control (FCS-MPC), in which the vector is unchanged in two steps. On this basis, the reference voltage calculation scheme based on the principle of no beat is extended to two control periods. According to the amplitude and sector of the reference voltage in two control periods, a fast voltage vector sequence positioning scheme is designed to reduce the computational burden. However, longer control periods will result in lower action granularity, hindering control performance. Currently, there is no control strategy that can reduce control complexity while breaking the fixed step length limit in traditional control. Therefore, it is necessary to develop a simple and practical multi-step model predictive method to obtain better control performance. SUMMARY

[0004] In order to solve the problems of large calculation amount and fixed prediction step length in traditional multi-step model predictive control, the present disclosure provides a variable step length multi-step model predictive current control method.

[0005] The present disclosure provides a variable step length multi-step model predictive current control method, comprising the following steps: Step 1: Obtain the torque reference value through the speed outer loop proportional integral (PI) regulator T e ref ; the given rotor flux amplitude reference value Ψ r ref is the rated value; Step 2: Obtain the current reference value at time (t+1) according to the motor mathematical model k + N )i s Nref Therefore, based on the principle of no deadbeat current, ( k + N Reference voltage vector of the control cycle u s Nref ; Step 3: Based on the principle of volt-second balance, calculate the voltage vector action time corresponding to the two non-zero voltage vectors and one zero voltage vector required to synthesize the reference voltage vector; Step 4: Construct a set of candidate voltage vector sequences that can achieve deadbeat control based on the reference voltage vector; Step 5, based on N The evaluation value function is based on the principle of minimizing current harmonics in each control cycle. N The optimal application vector within each control cycle; Step 6: Construct the drive signal for each switch of the inverter in the next control cycle based on the pulse sequence and action time in Step 5; Step 7: Update the average switching frequency of the switching transistor in the next cycle based on the driving signal in Step 6. Use the difference between the reference switching frequency and the actual switching frequency as input to construct a proportional-integral regulator and update the prediction step size for the next cycle. N .

[0006] In some embodiments, in step 2 N The current reference value and reference voltage vector for each control cycle are obtained as follows: Considering the steady-state condition, they are calculated using equation (1). k + N The reference value of the current at time ( ) is obtained; according to the mathematical model of the induction motor, the stator current prediction value at the next time moment is obtained by discretization, which is expressed as Equation (2); finally, the current is calculated by Equation (3). k + N Reference voltage vector for the control cycle: (1); (2); (3); in , , , T e ref This is a torque reference value. Ψ r ref This is a reference value for the rotor flux linkage amplitude. Ψ r For rotor flux linkage, is is a stator current, u s is a stator voltage, ω r is a rotational speed, θ r is a motor rotor flux angle; N p is a number of pole pairs of an asynchronous motor, j is an imaginary unit, R s is a stator resistance, L s is a stator inductance of an asynchronous motor, L r is a rotor inductance of an asynchronous motor, L m is a mutual inductance of an asynchronous motor, superscript k +1 represents parameter information of a next time, T sc is a control period.

[0007] In some embodiments, the application time calculation method of two non-zero voltage vectors and one zero voltage vector in step 3 is: based on the volt-second balance principle, the reference voltage vector is synthesized by two non-zero vectors u 1, u 2 and one zero vector u 0, the application time of the corresponding vector is calculated by formula (4): (4); wherein T N = NT sc , the modulation ratio M is , the reference voltage angle θ is .

[0008] In some embodiments, the construction method of the candidate voltage vector sequence set in step 4 is: in order to realize the shoot-through-free control, the candidate voltage vector sequence set contains two non-zero vectors and a zero vector in the sector where the reference voltage vector u s Nref is located, and the principle of only one switching jump between the front and rear vectors is satisfied; and the first vector in the candidate voltage vector sequence set is the basic voltage vector in the sector where the voltage vector u s Nref at the end of the first control period is located. k u end k ​​The basic voltage vector with a constant or only one switching frequency; To further reduce the root mean square value of harmonic current, four voltage vectors are applied in one control cycle, and the application time of the repeatedly applied voltage vectors is also equally divided accordingly.

[0009] In some embodiments, in step 5 N The optimal application vector selection method within each control cycle is as follows: Calculate the current error of each vector sequence at the end of the action of each basic vector according to equation (5). i error Then, the effective value of the current harmonics is calculated based on the segmented current error vector. Ih rms Finally, based on minimizing the value function J As shown in equation (6), select to apply to N Optimal vector sequence for each control cycle u opt and the corresponding vector action time t opt : (5); (6); in n =1,2,3…… g , g The number of voltage vectors in a candidate vector sequence. T N ( q )express N The first control cycle q The duration of action of each basic voltage vector u s k+N ( q )express N The first control cycle q A voltage vector that acts.

[0010] In some embodiments, the principle for applying the voltage vector in the next control cycle in step 6 is as follows: based on The cumulative sum shown t Cut off u opt and t opt to T N / N The part that is obtained , ; Then, according to equation (7), the truncated part... t optThe application time of the last vector in the middle is adjusted to T N / N Subtract the sum of all previous times; finally, T opt Vector normalization, so that the sum of its elements is T N / N The application vector for the next control cycle is obtained. U opt Corresponding vector action time t opt : (7); in n yes T opt The length.

[0011] In some embodiments, the prediction step size for the next cycle in step 7 N The update method is as follows: the difference between the reference switching frequency and the actual switching frequency shown in equation (8) is... e As input, the proportional-integral regulator shown in equations (9) and (10) is constructed to update the prediction step size for the next cycle. N : (8); (9); (10); in, K p For proportional gain, K i It is integral gain. f sw ref and f sw av They represent k The given average switching frequency at any given time and the actual average switching frequency.

[0012] This disclosure has the following characteristics and advantages: (1) The reference voltage calculation scheme based on the time-free principle is extended to N control cycles, and the initial vector action time is calculated based on the volt-second balance principle to reduce the amount of calculation.

[0013] (2) Based on the principle of minimizing current harmonics in N control cycles, the optimal application vector within a long control cycle is selected, thereby determining the application vector for the next control cycle. This method of quickly determining the vector sequence avoids the 7 n Rolling calculation of powers.

[0014] (3) Using the difference between the reference switching frequency and the actual switching frequency as input, a proportional-integral regulator is constructed to update the prediction step size for the next cycle, thereby achieving online adaptive adjustment of the prediction step size. Attached Figure Description

[0015] Figure 1 This is a hardware structure diagram of an induction motor speed control system; Figure 2 This is a block diagram of a multi-step model predictive current control method with variable prediction step size; Figure 3 shows the experimental results of stator current fast Fourier transform (FFT) analysis when the motor is running at 150 r / min, 750 r / min, and 1500 r / min with rated load, using traditional deadbeat current control at a sampling frequency of 5 kHz and a switching frequency of 5 kHz. Figure 4 shows the FFT analysis results of the stator current when the motor is running at 150 r / min, 750 r / min, and 1500 r / min with rated load, using the multi-step model with variable prediction step size of this disclosure, with the current prediction controlled at a sampling rate of 10 kHz and a given switching frequency of 5 kHz. Figure 5 The results are dynamic experimental results of a motor under sudden loading, using a multi-step model with variable prediction step size as disclosed in this paper to predict current control at a sampling rate of 10kHz. Figure 6 The results are dynamic experimental findings of the motor in forward and reverse rotation at rated speed, using the multi-step model with variable prediction step size as disclosed in this paper to predict current control at a sampling rate of 10kHz. Detailed Implementation

[0016] The following embodiments are intended to enable those skilled in the art to fully understand this disclosure, but do not limit this disclosure in any way.

[0017] This disclosure provides a multi-step model predictive current control method with variable prediction step size, including: Step 1: Obtain the torque reference value through the speed outer loop proportional-integral (PI) regulator. T e ref Given a reference value for rotor flux amplitude. Ψ r ref This is the rated value; Step 2: Obtain (based on the mathematical model of the motor) k + N Current reference value at time ) i s Nref Therefore, based on the principle of no deadbeat current, ( k + N Reference voltage vector of the control cycleu s Nref ; Step 3: Based on the principle of volt-second balance, calculate the voltage vector action time corresponding to the two non-zero voltage vectors and one zero vector required to synthesize the reference voltage vector; Step 4: Construct a set of candidate vector sequences that can achieve deadbeat control based on the reference voltage vector; Step 5: Based on N The evaluation value function is based on the principle of minimizing current harmonics in each control cycle. N The optimal application vector within each control cycle; Step 6: Construct the drive signal for each switch of the inverter in the next control cycle based on the pulse sequence and action time in Step 5; Step 7: Update the average switching frequency of the switching transistor in the next cycle based on the drive signal in Step 6. Use the difference between the reference switching frequency and the actual switching frequency as input to construct a proportional-integral regulator and update the prediction step size for the next cycle. N .

[0018] Figure 1 The hardware circuit structure diagram of this disclosure includes a three-phase voltage source, an asynchronous motor, a three-phase diode rectifier bridge, a DC-side capacitor, an asynchronous motor, a voltage and current sampling circuit, a DSP controller, and a drive circuit. The voltage and current sampling circuit uses voltage Hall sensors and current Hall sensors to collect the DC-side voltage and the a-phase and b-phase currents of the asynchronous motor, respectively. The sampled signals are then processed by a signal conditioning circuit and converted into digital signals by the DSP controller. The DSP controller performs the calculations proposed in this disclosure, outputting six switching pulses, which are then processed by the drive circuit to obtain the final drive signals for the six switching transistors of the inverter.

[0019] Figure 2 This is a block diagram of the control principle of this disclosure. The control method is in Figure 1 Implement the following steps sequentially on the DSP controller: Step 1: Based on the torque command obtained from the outer loop speed PI regulator T e ref Specifically, it is expressed as: ; in and These are the proportional gain and integral gain in the PI controller, respectively. Indicates a given rotational speed. This represents the actual rotational speed, and S is the Laplace operator.

[0020] Then, a reference value for the rotor flux linkage amplitude is given. Ψ rref This is the rated value.

[0021] Step 2: Calculate ( k + N The current reference value at time ) is specifically expressed as follows: ; in i d ref , i q ref They are respectively d-q Shaft stator current setpoint, , , T e ref This is a torque reference value. Ψ r ref This is a reference value for the rotor flux linkage amplitude. Ψ r For rotor flux linkage, N p This represents the number of pole pairs of the asynchronous motor. L r For the rotor inductance of the asynchronous motor, L m For the mutual inductance of the asynchronous motor, j is the imaginary unit. θ r The rotor flux linkage angle of the motor. T sc To control the cycle, N For predicting the step size. Superscript k This indicates the parameter information at the current moment.

[0022] Based on the mathematical model of the induction motor, the stator current prediction value at the next moment is obtained by discretization, specifically expressed as: ; The leakage coefficient Rotor time constant . L s For the stator inductance of the asynchronous motor, R s For stator resistance, R r For rotor resistance, u s This is the stator voltage.

[0023] at last( k + N Reference voltage vector of the control cycle u s Nref It can be calculated as: ; The superscript k+1 indicates the parameter information at the next time step (time k+1). Let T N =N T sc .

[0024] Step 3: Based on the volt-second balance principle, the reference voltage vector can be synthesized from two non-zero vectors and one zero vector. The application time of the corresponding vector can be calculated as follows: ; The modulation ratio M for Reference voltage angle θ for .

[0025] Step 4: To achieve deadbeat control, the candidate vector sequence must include the reference voltage vector. u s Nref The two non-zero vectors and the zero vector in the sector must satisfy the principle of only one switching transition between the preceding and following vectors. The first vector in the candidate vector sequence is... u s Nref The basic voltage vector of the sector is related to the first k Voltage vector at the end of each control cycle u end ( k The basic voltage vector that has a constant number of switching cycles or only one switching cycle.

[0026] To further reduce the root mean square value of harmonic current, four voltage vectors are applied within one control cycle. The application time of the repeatedly applied voltage vectors is then equally divided. Taking the first sector as an example, the candidate pulse sequences and their application times are summarized in Table 1.

[0027] Table 1: u end ( k )]]> ​ candidate vector sequence corresponding vector action time u 0 / u 5]]> ​ u 0, u 1, u 2) ( u 0, u 1, u 2, u 1) ( u 0, u 1, u 2, u 7)]]> ​ <![CDATA[( t 0, t 1, t 2) ( t 0,0.5 t 1, t 2,0.5 t 1) (0.5 t 0, t 1, t 2,0.5 t 0)]]> u 1]]> ​ <![CDATA[( u 1, u 2, u 7) ( u 1, u 2, u 7, u 2)( u 1, u 2, u 1, u 0)( u 1, u 0, u 1, u 2)]]> <![CDATA[( t 1, t 2, t 0) ( t 1,0.5 t 2, t 0,0.5 t 2)(0.5 t 1, t 2,0.5 t 1, t 0) (0.5 t 1, t 0,0.5 t 1, t 2)]]> u 2]]> ​ u 2, u 1, u 0) ( u 2, u 1, u 0, u 1)( u 2, u 1, u 2, u 7)( u 2, u 7, u 2, u 1)]]> ​ <![CDATA[( t 2, t 1, t 0) ( t 2,0.5 t 1, t 0,0.5 t 1)(0.5 t 2, t 1,0.5 t 2, t 0) (0.5 t 2, t 0,0.5 t 2, t 1)]]> u 3]]> ​ u 0, u 1, u 2) ( u 0, u 1, u 2, u 1) ( u 0, u 1, u 2, u 7)( u 2, u 1, u 0) ( u 2, u 1, u 0, u 1)( u 2, u 1, u 2, u 7)( u 2, u 7, u 2, u 1)]]> ​ <![CDATA[( t 0, t 1, t 2) ( t 0,0.5 t 1, t 2,0.5 t 1) (0.5 t 0, t 1, t 2,0.5 t 0) ( t 2, t 1, t 0) ( t 2,0.5 t 1, t 0,0.5 t 1) (0.5 t 2, t 1,0.5 t 2, t 0) (0.5 t 2, t 0,0.5 t 2, t 1)]]> u 6]]> ​ <![CDATA[( u 1, u 2, u 7) ( u 1, u 2, u 7, u 2)( u 1, u 2, u 1, u 0)( u 1, u 0, u 1, u 2)( u 7, u 2, u 1) ( u 7, u 2, u 1, u 0)( u 7, u 2, u 1, u 2)]]> <![CDATA[( t 2, t 1, t 0) ( t 2,0.5 t 1, t 0,0.5 t 1)(0.5 t 2, t 1,0.5 t 2, t 0) (0.5 t 2, t 0,0.5 t 2, t 1)( t 0, t 2, t 1)(0.5 t 0, t 2, t 1,0.5 t 0) ( t 0,0.5 t 2, t 1,0.5 t 2)]]> u 7 / u 4]]> ​ u 7, u 2, u 1) ( u 7, u 2, u 1, u 0) ( u 7, u 2, u 1, u 2)]]> ​ <![CDATA[( t 0, t 2, t 1) (0.5 t 0, t 2, t 1,0.5 t 0) ( t 0,0.5 t 2, t 1,0.5 t 2)]]> Step 5: Current error of various vector sequences at the end of each fundamental vector's action time. i error It can be represented as: ; Then calculate the effective value of current harmonics based on the segmented current error vector. Ih rms Finally, based on minimizing the value function J The selection shown applies to N Optimal vector sequence for each control cycle u opt and the corresponding vector action time topt .

[0028] ; in n =1,2,3…… g , g The number of voltage vectors in a candidate vector sequence. T N ( q )express N The first control cycle q The duration of action of each basic voltage vector u s k+N ( q )express N The first control cycle q A voltage vector that acts.

[0029] Step 6: According to The cumulative sum shown t Cut off u opt and t opt To T N The part of / N is obtained , Then cut off t opt The application time of the last vector in the middle is adjusted to T. N / N minus the sum of all previous times, specifically expressed as: ; in, n yes T opt The length.

[0030] Finally T opt Vector normalization so that the sum of its elements is T N / N, to obtain the application vector for the next control cycle. U opt Corresponding vector action time t opt .

[0031] Step 7: Divide the reference switching frequency and the actual switching frequency. e As input, it is specifically represented as: ; Reconstruct the proportional-integral regulator to update the prediction step size for the next cycle. N Specifically, it is expressed as: ; ; in, K p For proportional gain, K i It is integral gain. f sw ref and f sw av They represent k The given average switching frequency at any given time and the actual average switching frequency.

[0032] The effectiveness of the method proposed in this disclosure can be demonstrated by comparing the experimental results shown in Figures 3 and 4. Figures 3(a)-(c) show the stator current total harmonic distortion (THD) analysis results when the motor operates at 150 r / min, 750 r / min, and 1500 r / min with rated load using conventional deadbeat current control. Figures 4(a)-(c) show the experimental results of the method in this disclosure under the same operating conditions and switching frequency. The comparison between Figures 3 and 4 reveals that, at low, medium, and high speeds, the method in this disclosure exhibits lower current harmonics and better steady-state performance than the comparative method. At 1500 r / min, the total harmonic distortion (THD) of the method in this disclosure is reduced by 36.2% compared to the comparative method. Figure 5 and Figure 6 This is a dynamic experimental waveform diagram of the multi-step model predicting current control with variable prediction step size under sudden loading and forward / reverse rotation at rated speed. From top to bottom, the experimental results are for speed, q-axis current, d-axis current, and A-phase stator current. It can be seen that the system has satisfactory dynamic performance.

[0033] This disclosure solves the problem that existing multi-step predictive control strategies, where the prediction step size is fixed as an integer, result in high computational complexity and cannot guarantee controllable switching frequency. The method described in this disclosure extends the reference voltage calculation scheme based on the current deadbeat control principle to N control cycles, and calculates the three adjacent vectors and their action times to synthesize the reference voltage vector according to the volt-second balance principle, thereby establishing a candidate voltage vector sequence containing three or four vectors, avoiding the problem of 7... n The calculation of the secondary value function reduces the computational burden. Simultaneously, based on minimizing current harmonics over N control cycles, the optimal application vector within those N control cycles is selected, thus determining the application vector for the next control cycle and enhancing the system's steady-state performance. Furthermore, the difference between the reference switching frequency and the actual switching frequency is used as input to construct a proportional-integral controller, enabling online adjustment of the prediction step size.

[0034] Those skilled in the art should understand that the above embodiments are merely exemplary embodiments, and various changes, substitutions, and modifications can be made without departing from the spirit and scope of this disclosure.

Claims

1. A multi-step model predictive current control method with variable prediction step size, characterized in that, Includes the following steps: Step 1: Obtain the torque reference value through the speed outer loop proportional-integral (PI) regulator. T e ref ; Given a reference value for rotor flux amplitude ψ r ref This is the rated value; Step 2, obtain the mathematical model of the motor ( k + N Current reference value at time ) i s Nref Therefore, based on the principle of no deadbeat current, ( k + N Reference voltage vector of the control cycle u s Nref ; Step 3: Based on the principle of volt-second balance, calculate the voltage vector action time corresponding to the two non-zero voltage vectors and one zero voltage vector required to synthesize the reference voltage vector; Step 4: Construct a set of candidate voltage vector sequences that can achieve deadbeat control based on the reference voltage vector; Step 5, based on N The evaluation value function is based on the principle of minimizing current harmonics in each control cycle. N The optimal application vector within each control cycle; Step 6: Construct the drive signal for each switch of the inverter in the next control cycle based on the pulse sequence and action time in Step 5; Step 7: Update the average switching frequency of the switching transistor in the next cycle based on the driving signal in Step 6. Use the difference between the reference switching frequency and the actual switching frequency as input to construct a proportional-integral regulator and update the prediction step size for the next cycle. N .

2. The multi-step model predictive current control method with variable prediction step size according to claim 1, characterized in that, In step 2 N The current reference value and reference voltage vector for each control cycle are obtained as follows: Considering the steady-state condition, they are calculated using equation (1). k + N The reference value of the current at time ( ) is obtained; according to the mathematical model of the induction motor, the stator current prediction value at the next time moment is obtained by discretization, which is expressed as Equation (2); finally, the current is calculated by Equation (3). k + N Reference voltage vector for the control cycle: （1）； （2）； （3）； in , , , T e ref This is a torque reference value. ψ r ref This is a reference value for the rotor flux linkage amplitude. ψ r For rotor flux linkage, i s For stator current, u s Stator voltage, ω r For rotational speed, θ r The rotor flux linkage angle of the motor; N p This represents the number of pole pairs of the asynchronous motor. j The imaginary unit, R s For stator resistance, L s For the stator inductance of the asynchronous motor, L r For the rotor inductance of the asynchronous motor, L m For mutual inductance in asynchronous motors, superscript k +1 indicates the parameter information for the next time step. T sc To control the cycle.

3. The multi-step model predictive current control method with variable prediction step size according to claim 1, characterized in that, The method for calculating the application time of the two non-zero voltage vectors and one zero voltage vector in step 3 is as follows: Based on the volt-second balance principle, the reference voltage vector consists of two non-zero vectors. u 1, u 2 and a zero vector u 0 synthesis, the application time of the corresponding vector is calculated by equation (4): （4）； in T N = NT sc modulation ratio M for Reference voltage angle θ for .

4. The multi-step model predictive current control method with variable prediction step size according to claim 1, characterized in that, The method for constructing the candidate voltage vector sequence set in step 4 is as follows: In order to achieve deadbeat-free control, the candidate voltage vector sequence set includes a reference voltage vector. u s Nref The two non-zero vectors and the zero vector in the sector, and the principle that there is only one switching transition between the preceding and following vectors; while the first vector in the candidate voltage vector sequence set is... u s Nref The basic voltage vector of the sector is related to the first k Voltage vector at the end of each control cycle u end ( k The basic voltage vector with a constant or only one switching frequency; To further reduce the root mean square value of harmonic current, four voltage vectors are applied in one control cycle, and the application time of the repeatedly applied voltage vectors is also equally divided accordingly.

5. The multi-step model predictive current control method with variable prediction step size according to claim 1, characterized in that, In step 5 N The method for selecting the optimal application vector within a control cycle is as follows: Calculate the current error of each vector sequence at the end of each basic vector action time according to equation (5). i error Then, the effective value of the current harmonics is calculated based on the segmented current error vector. Ih rms Finally, based on minimizing the value function J As shown in equation (6), select to apply to N Optimal vector sequence for each control cycle u opt and corresponding vector action time t opt : （5）； （6）； in n =1,2,3…… g , g The number of voltage vectors in a candidate vector sequence. T N ( q )express N The first control cycle q The duration of action of each basic voltage vector u s k+N ( q )express N The first control cycle q A voltage vector that acts.

6. The multi-step model predictive current control method with variable prediction step size according to claim 1, characterized in that, The principle for applying the voltage vector in the next control cycle in step 6 is as follows: based on The cumulative sum shown t Cut off u opt and t opt to T N / N The part that is obtained , ; Then, according to equation (7), the truncated part... t opt The application time of the last vector in the middle is adjusted to T N / N Subtract the sum of all previous times; finally, T opt Vector normalization, so that the sum of its elements is T N / N The application vector for the next control cycle is obtained. U opt Corresponding vector action time t opt : （7）； in n yes T opt The length.

7. The multi-step model predictive current control method with variable prediction step size according to claim 1, characterized in that, The prediction step size for the next cycle in step 7 N The update method is as follows: the difference between the reference switching frequency and the actual switching frequency shown in equation (8) is... e As input, the proportional-integral regulator shown in equations (9) and (10) is constructed to update the prediction step size for the next cycle. N : （8）； （9）； （10）； in, K p For proportional gain, K i It is integral gain. f sw ref and f sw av They represent k The given average switching frequency at any given time and the actual average switching frequency.

Images