A multi-objective model predictive control method for a three-phase four-wire three-level inverter

Abstract

The application discloses a kind of three-phase four-wire three-level inverter multi-objective model predictive control method, belong to the field of power electronic converter.This method realizes three-phase output voltage quality and midpoint voltage balance multi-objective optimization without weight by using series structure model predictive control;Three-phase output voltage cost function is arranged in series structure, and the voltage reference value of inverter side bridge arm is optimized, in series structure, by the time of O switch state action is partially decomposed into equal P, N switch state action time, without affecting output voltage quality, while reducing midpoint voltage fluctuation, avoid weight coefficient design;The resonant generated by introducing wave trap to suppress filter and load interaction, and the output voltage reference value and predicted value of next two times are used to construct cost function to eliminate digital control delay effect.The application can realize three-phase four-wire three-level inverter with balanced / unbalanced load output voltage quality and midpoint voltage balance without weight multi-objective optimization control.

Description

Technical Field

[0001] This invention belongs to the field of power electronic converter technology, specifically relating to a multi-objective model predictive control method for a three-phase four-wire three-level inverter. Background Technology

[0002] Compared to two-level inverters, three-level inverters offer advantages such as lower output voltage harmonic content, lower device switching stress, and lower power loss, making them widely used in industry. During system operation, unbalanced load conditions are inevitable. Three-phase four-wire three-level inverters connect the midpoint of the DC-side capacitor to the AC-side load neutral point to provide a path for the zero-sequence current generated when an unbalanced load is connected to the inverter output, ensuring output voltage quality. In control schemes, model predictive control (MMDC) offers advantages over traditional linear control methods, including faster dynamic response, simpler structure, and fewer adjustment parameters. Continuous-set MMDC, through deadbeat-free MMDC, reduces the number of calculations to one, significantly reducing the computational load compared to finite-set MMDC. To meet the ever-increasing performance demands of inverter systems, MMDC is continuously optimized to achieve multi-objective predictive control, a process typically implemented through a cost function.

[0003] In existing technologies, the literature "Fang Fan, Li Yuan, Xiao Xianyong, et al. Finite set model predictive control strategy for energy storage quasi-Z source inverters [J]. Proceedings of the CSEE, 2019, 39(07): 2133-2144" utilizes a cost function with weighted coefficients constructed from system state variables to achieve multi-objective model predictive control that improves the output current quality of the inverter system, reduces the average switching frequency of the system, and improves the transient response of the system. The literature "Yang Yong, Xie Menxi, Fan Mingdi. Finite set model predictive control method for T-type three-level single-phase inverters [J]. Journal of Electrical Engineering and Control, 2019, 23(03): 120-126" constructs current tracking and T-type three-level neutral point voltage balance objective terms with weighted coefficients in the objective function to achieve multi-objective model predictive control that enables fast and accurate tracking of the given current and neutral point voltage balance control. However, the above literatures employ a cost function with weighted coefficients to achieve multi-objective model predictive control. In the cost function, the more control objectives there are, the more weighting coefficients are required. Furthermore, since different control objectives have different properties and dimensions, the design of weighting coefficients is quite complex. Additionally, the literature "Tian Yazhuo, Liu Chenwei, Zhang Yongjun, et al. An Improved Sequence Model Predictive Control Strategy for T-Type Three-Level Inverters [J]. Journal of Electrical Engineering, 2024, 39(24):7821-7832" proposes a sequence model predictive control strategy based on a series-parallel hybrid structure. This strategy uses a parallel structure to synchronously optimize objectives with similar priorities, and a series structure to sequentially control objectives with different priorities, thus avoiding the need for weighting coefficient design and effectively achieving multi-objective model predictive control that suppresses common-mode voltage and reduces the average switching frequency. However, the improved sequence model predictive control strategy for multi-objective optimization without weighting coefficients proposed in this literature, while not requiring weighting coefficient design, cannot be used to solve the multi-objective optimization problem of three-level inverters with unbalanced loads.

[0004] Therefore, there is an urgent need for a three-level inverter control method that can achieve unweighted multi-objective optimization control to balance output voltage quality and midpoint voltage under unbalanced load conditions. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a multi-objective model predictive control method for a three-phase four-wire three-level inverter. This invention enables unweighted multi-objective optimization control of output voltage quality and neutral point voltage balance under balanced / unbalanced load conditions, while suppressing resonance caused by filter-load interaction and eliminating the effects of digital control delay.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a multi-objective model predictive control method for a three-phase four-wire three-level inverter, comprising the following steps:

[0007] Step S1: Sample the three-phase output voltage, the three-phase inverter side filter inductor current, and the three-phase load current of the three-phase four-wire three-level inverter, establish the system mathematical model, and discretize it using the forward Euler method to obtain the system discrete model;

[0008] Step S2: Calculate the predicted value of the three-phase output voltage at time k+2 based on the discrete model of the system.

[0009] Step S3: Perform phase correction on the three-phase output voltage reference value at time k+2, and construct the three-phase output voltage cost function based on the corrected three-phase output voltage reference value at time k+2 and the predicted value of the three-phase output voltage at time k+2.

[0010] Step S4: The reference value of the three-phase output voltage at time k+2, together with the sampling signal at time k, is passed through the series structure model prediction control with notch filter to obtain the P switch state action time and N switch state action time of the three-phase inverter side bridge arm; wherein, the sampling signal includes the filter inductor current, three-phase output voltage, three-phase load current and the voltages of the two DC side capacitors.

[0011] Step S5: Based on the operating time of the P switch state and the operating time of the N switch state, and combined with the modulation method, obtain the driving signal for driving the switching transistors of the three-level inverter.

[0012] The series structure model predictive control with notch filter includes arranging the output voltage cost functions of phases a, b, and c in a series structure, controlling phases a, b, and c sequentially, and comparing the duration t of the non-zero switching state when each phase satisfies the minimum output voltage cost function. x The sum of the durations of the N-switch and P-switch states required to achieve midpoint voltage balance, t x_NP The magnitude relationship determines whether to decompose the action time of the O switch state of that phase in order to achieve multi-objective optimization control of output voltage quality and neutral point voltage balance. x represents a, b, and c, which represent the three phases a, b, and c.

[0013] Furthermore, the three-phase four-wire three-level inverter adopts a split capacitor three-level inverter structure, consisting of an input DC source, a DC-side capacitor composed of two equal capacitors connected in series, a three-arm three-level inverter bridge, a three-phase LC filter, and a three-phase load. The midpoint of the DC-side capacitor is connected to the neutral point of the AC-side load of the inverter to form a three-phase four-wire system. Each phase of the three-level inverter has three switching states: O switching state, P switching state, and N switching state.

[0014] Further, in step S4, the series structure model predictive control with notch filter specifically includes:

[0015] Step A1: Use the binary search method to optimize the reference value of the inverter side bridge arm voltage at time k, and obtain the reference value of the inverter side bridge arm voltage at time k+2 that satisfies the minimum cost function of the output voltage of phase a at time k+2.

[0016] Step A2: After processing the reference voltage value of the inverter side bridge arm of phase a obtained in step A1 through a notch filter, the final reference voltage value v of the inverter side bridge arm of phase a at time k is obtained. ao * (k);

[0017] Step A3: Based on the reference value v of the inverter side bridge arm voltage obtained in step A2... ao * (k) Calculate the duration t of the non-O switching state of phase a when the output voltage cost function of phase a is minimized. a And the reference value v of the bridge arm voltage on the inverter side of phase a. ao * The change in midpoint voltage caused by (k) v NP_a (k);

[0018]

[0019]

[0020] Among them, V dc For the input DC source voltage, T s One sampling period; i a (k) represents the current in the filter inductor on the inverter side of phase a at time k, C dc The capacitance values ​​of the two equal capacitors on the DC side are used for input.

[0021] Step A4: Calculate the sum of the action times t of the N-switch and P-switch states required for phase b to achieve neutral voltage balance after the introduction of phase b. b_NP ;

[0022]

[0023] Among them, v NP (k) represents the difference between the upper and lower capacitor voltages on the input DC side at time k, i b (k) represents the filter inductor current on the b-phase inverter side at time k;

[0024] Step A5: Optimize the reference voltage value of the b-phase inverter side bridge arm using the binary search method, and obtain the reference voltage value v of the b-phase inverter side bridge arm at time k after processing with a notch filter. bo * (k);

[0025] Step A6: Based on the b-phase inverter side bridge arm voltage reference value v obtained in step A5bo * (k) Calculate the duration t of the non-O switching state of phase b when the output voltage cost function of phase b is minimized. b ;

[0026]

[0027] Step A7: Compare t b_NP With t b The magnitude of the value determines whether the state action time of phase b O switch should be decomposed, and the state action time t of phase b bridge arm P switch is calculated. b_p and the duration of the N switch state t b_n ;

[0028] Step A8: Based on the operating time of switch P and switch N in phase b obtained in step A7, calculate the change in midpoint voltage v caused by the operation of the phase b switch. NP_b (k);

[0029]

[0030] Step A9: Calculate the sum of the action times t of the N-switch and P-switch states required for phase c to achieve neutral voltage balance after the introduction of phase c. c_NP ;

[0031]

[0032] Among them, i c (k) represents the current of the filter inductor on the inverter side of phase c at time k;

[0033] Step A10: Optimize the reference voltage value of the c-phase inverter side bridge arm using the binary search method, and obtain the reference voltage value v of the c-phase inverter side bridge arm at time k after processing with a notch filter. co * (k);

[0034] Step A11: Based on the reference value v of the c-phase inverter side bridge arm voltage obtained in step A10... co * (k), calculate the duration t of the non-O switching state of phase c when the cost function of phase c output voltage is minimized. c ;

[0035]

[0036] Step A12: Compare t c_NP With t c The magnitude of the value determines whether the state action time of phase C O switch should be decomposed, and the state action time t of phase C bridge arm P switch is calculated. c_p and the duration of the N switch state tc_n ;

[0037] Step A13: Based on the operating time of the P switch and the N switch obtained in step A12, calculate the change in midpoint voltage v caused by the operation of the c-phase switch. NP_c (k);

[0038]

[0039] Step A14: Calculate the sum of the action times t of the N switch state and the P switch state required to achieve the neutral point voltage balance in phase a. a_NP ;

[0040]

[0041] Step A15: Compare t a With t a_NP The magnitude of the value determines whether the action time of the phase a O switch state should be decomposed, and the action time t of the phase a bridge arm P switch state is calculated. a_p and the duration of the N switch state t a_n ;

[0042] Step A16: Convert the P switch state duration and N switch state duration of phases a, b, and c within this sampling period into switching transistor drive control signals; based on the non-O switch state duration t of each phase when the output voltage cost function is minimized... x The sum of the durations of the N-switch and P-switch states required to achieve midpoint voltage balance, t x_NP The magnitude relationship determines the modulation method, when t x ≥t x_NP When using dual-carrier SPWM modulation method, when t x <t x_NP The dual-modulation wave carrier pulse width modulation (DMW-CBPWM) method is used to obtain the drive signal for each phase switch.

[0043] Further, in step A1, the dichotomy method includes:

[0044] Set the initial values ​​V of the two boundary values ​​of the inverter side bridge arm voltage at time k. left =-V dc / 2,V right =V dc / 2, Substitute the boundary values ​​into step S2 to obtain the predicted value of the inverter output voltage at time k+2, and based on the output voltage cost function value at time k+2 obtained in step S3, set the midpoint V. mid For the current V left and V right The average value, i.e., V mid =(Vleft +V right ) / 2, V mid As the boundary value for one side of the next optimization, the boundary value with the smaller current output voltage cost function is selected as the boundary value for the other side of the next optimization. After m optimizations, the final boundary values ​​V on both sides are obtained. left and V right The reference value of the inverter side bridge arm voltage at time k is determined to be the final value V at both boundary values. left and V right The average value.

[0045] Furthermore, in step A2, the transfer function of the notch filter in the continuous domain is:

[0046]

[0047] Where Q is the notch filter quality factor. The notch frequency; Equal to the system's resonant angular frequency ω r The notch width Δω = 0.1ω r At frequency ω r The notch depth A(ω) at ±Δω is... r The amplitude of the system output voltage resonant frequency (±Δω) is less than or equal to the negative value; the notch filter transfer function is discretized using discretization methods such as the bilinear transform method.

[0048] Furthermore, in step A7, determining whether to decompose the action time of the b-phase O switch state specifically includes:

[0049] If t b ≥t b_NP Therefore, the duration of the b-phase O switch state is not decomposed. The condition of minimizing the output voltage cost function of the b-phase is prioritized. The task of suppressing the midpoint voltage fluctuation is transferred to the c-phase through a series structure. The duration t of the b-phase P switch state is determined based on the sign of the reference value of the bridge arm voltage on the b-phase inverter side. b_p and the duration of the N switch state t b_n ;

[0050]

[0051] If t b <t b_NP Then, the duration of the b-phase O switch state is partially decomposed into equal durations of the P and N switches, and allocated to the corresponding P and N switch state durations to obtain the duration t of the b-phase P switch state. b_p and the duration of the N switch state t b_n .

[0052] When v bo* When (k)>0, it corresponds to phase b t b_p t b_n for:

[0053]

[0054] When v bo * When (k) < 0, it corresponds to phase b t. b_p t b_n for:

[0055] .

[0056] Furthermore, in step A12, determining whether to decompose the action time of the c-phase O switch state specifically includes:

[0057] If t c ≥t c_NP Therefore, the duration of the C-phase O switch state is not decomposed. The condition of minimizing the C-phase output voltage cost function is prioritized. The task of suppressing midpoint voltage fluctuations is transferred to phase A through a series structure. The duration t of the C-phase P switch state is determined based on the sign of the reference value of the C-phase inverter side bridge arm voltage. c_p and the duration of the N switch state t c_n ;

[0058]

[0059] If t c <t c_NP Then, the duration of the c-phase O switch state is partially decomposed into equal durations of the P and N switches, and allocated to the corresponding P and N switch state durations to obtain the duration t of the c-phase P switch state. c_p and the duration of the N switch state t c_n .

[0060] When v co * When (k)>0, it corresponds to phase ct. c_p t c_n for:

[0061]

[0062] When v co * When (k) < 0, it corresponds to phase c t. c_p t c_n for:

[0063] .

[0064] Furthermore, in step A15, determining whether to decompose the action time of the phase O switch state specifically includes:

[0065] If t a ≥t a_NP Therefore, the duration of the phase a O switch state is not decomposed; instead, the duration t of the phase a P switch state is determined based on the sign of the reference voltage value of the phase a inverter side bridge arm. a_p and the duration of the N switch state t a_n ;

[0066]

[0067] If t a <t a_NP Then, the duration of the O switch state in phase a is partially decomposed into equal durations of the P and N switches, and allocated to the corresponding P and N switch state durations in phase a, thus obtaining the duration t of the P switch state in phase a. a_p and the duration of the N switch state t a_n .

[0068] When v ao * When (k)>0, it corresponds to phase t of phase a. a_p t a_n for:

[0069]

[0070] When v ao * When (k) < 0, it corresponds to phase t of phase a. a_p t a_n for:

[0071] .

[0072] Further, in step S3, the phase correction of the three-phase output voltage reference value at time k+2 specifically involves: performing phase compensation on the three-phase output voltage reference value at time k+2 based on the angle deviation of the fundamental signal of the output reference voltage after passing through the notch filter; the correction formula for the phase correction is:

[0073]

[0074] Among them, v ca * (k+2), v cb * (k+2), v cc * (k+2) represents the reference values ​​of the output voltages of phases a, b, and c at time k+2. The amplitude of the three-phase output voltage reference value. To output the fundamental angular frequency of the reference voltage, This represents the angular deviation of the fundamental signal after passing through the notch filter. The sampling period is For the fundamental period, .

[0075] This invention also provides a multi-objective model predictive control system for a three-phase four-wire three-level inverter, used to implement the above-mentioned method, characterized in that it includes:

[0076] The sampling module is used to sample the three-phase output voltage, the three-phase inverter-side filter inductor current, the three-phase load current, and the voltages of the two DC-side capacitors of the three-phase four-wire three-level inverter.

[0077] The modeling and discretization module is used to establish the mathematical model of the system and discretize it using the forward Euler method to obtain the discrete model of the system.

[0078] The prediction module is used to calculate the predicted value of the three-phase output voltage at time k+2 based on the discrete model of the system.

[0079] The cost function construction module is used to construct the three-phase output voltage cost function based on the predicted value of the three-phase output voltage at time k+2 and the reference value of the three-phase output voltage at time k+2 after phase correction.

[0080] The series structure model predictive control module is used to take the reference value of the three-phase output voltage at time k+2 and the sampled signal at time k together and pass them through the series structure model predictive control with notch filter to obtain the P switch state action time and N switch state action time of the three-phase inverter side bridge arm.

[0081] The drive signal generation module is used to generate a drive signal for driving the switching transistors of the three-level inverter based on the operating time of the P switch state and the operating time of the N switch state combined with a modulation method.

[0082] The series structure model predictive control with notch filter includes arranging the output voltage cost functions of phases a, b, and c in a series structure, controlling phases a, b, and c sequentially, and comparing the duration t of the non-zero switching state when each phase satisfies the minimum output voltage cost function. x The sum of the durations of the N-switch and P-switch states required to achieve midpoint voltage balance, t x_NP The magnitude of the voltage difference determines whether the duration of the phase O switch state should be decomposed to achieve multi-objective optimization control of output voltage quality and midpoint voltage balance.

[0083] Compared with existing technologies, this invention has the following advantages: The multi-objective model predictive control method for three-phase four-wire three-level inverters proposed in this invention effectively achieves multi-objective optimization control of output voltage quality and neutral point voltage balance when the three-phase four-wire three-level inverter is equipped with balanced / unbalanced loads, avoiding the design of weight coefficients in traditional multi-objective optimization control. Simultaneously, this invention can eliminate resonance caused by filter-load interaction and eliminate the effects of digital control delay, further improving output voltage quality. Attached Figure Description

[0084] Figure 1 This is a schematic diagram of the topology of a three-phase four-wire three-level inverter in an embodiment of the present invention;

[0085] Figure 2 This is a flowchart of the predictive control of the series structure model in an embodiment of the present invention;

[0086] Figure 3 This is a detailed block diagram of the predictive control process for the series structure model in an embodiment of the present invention;

[0087] Figure 4 This is a schematic diagram illustrating the principle of generating a PWM sequence using a dual-carrier SPWM modulation method in the x-phase of this invention.

[0088] Figure 5 This is a schematic diagram illustrating the principle of generating a PWM sequence using the dual-modulation wave carrier pulse width modulation (DMW-CBPWM) method in phase x of this invention. Detailed Implementation

[0089] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0090] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0091] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0092] This embodiment provides a multi-objective model predictive control method applied to a three-phase four-wire three-level inverter. By using series-structure model predictive control, multi-objective optimization of system output voltage quality and neutral point voltage balance under balanced / unbalanced load conditions is achieved, avoiding the need for designing cost function weighting coefficients.

[0093] like Figure 1 As shown, the three-phase four-wire three-level inverter adopts a split capacitor three-level inverter structure, which is powered by an input DC source V. dc The inverter consists of a DC-side capacitor composed of two equal capacitors C1 and C2 connected in series, a three-arm three-level inverter bridge, a three-phase LC filter, and a three-phase load. The midpoint o of the DC-side capacitor is connected to the neutral point of the AC-side load of the inverter to form a three-phase four-wire system, providing a path for the zero-sequence current generated when the inverter output is connected to an unbalanced load.

[0094] A three-level inverter has three switching states per phase: O (zero), P (power), and N (neutral). Each phase arm consists of four switching transistors, S... x1 With S x3 S x2 With S x4 Each component is complementary and conducts (x=a, b, c). When S... x1 S x2 When conducting, S x3 S x4 When turned off, the voltage v of the x-phase output inverter side bridge arm is... xo For V dc / 2, corresponding to the state of switch P; when S x2 S x3 When conducting, S x1 S x4 When turned off, the voltage v of the x-phase output inverter side bridge arm is... xo A value of 0 corresponds to the O switch state; when S... x3 S x4 When conducting, S x1 S x2 When turned off, the voltage v of the x-phase output inverter side bridge arm is... xo -V dc / 2 corresponds to the N switch state.

[0095] The multi-objective model predictive control strategy first establishes a mathematical model based on the working principle of a three-phase four-wire three-level inverter, and then discretizes the mathematical model using the forward Euler method. Figure 2The series structure model with notch filter shown is used to predict and control the duration of the P and N switching states of the three-phase three-level inverter bridge arm. Combined with the dual-carrier SPWM modulation method and the dual-modulation wave carrier pulse width modulation (DMW-CBPWM) method, the driving signal for driving the switching transistors of the three-level inverter is obtained, realizing multi-objective optimization control of output voltage quality and midpoint voltage balance.

[0096] Predictive control of a series structure with notch filters involves arranging the output voltage cost functions of phases a, b, and c in a series configuration. First, phase a is controlled. In the series configuration, a binary search method is used to optimize the inverter-side bridge arm voltage reference value for phase a, obtaining the phase a inverter-side bridge arm voltage reference value v that minimizes the output voltage cost function of phase a. ao_s * v ao_s * After passing through the notch filter, the final reference value of the bridge arm voltage on the a-phase inverter side is obtained. ao * and based on v ao * The duration t of the non-O switching state of phase a when the output voltage cost function of phase a is minimized is obtained. a At the same time, it was obtained that when the voltage of the inverter side bridge arm of phase a is v ao * The resulting change in midpoint voltage v NP_a (ii) Controlling phase b. Assuming that the midpoint voltage balance can be achieved after introducing control of phases a and b, calculate the sum of the action times t of the N switch state and the P switch state required for phase b at this time. b_NP The bisection method is used to optimize the reference voltage value of the bridge arm on the b-phase inverter side, thereby obtaining the reference voltage value v of the b-phase inverter side bridge arm that minimizes the cost function of the b-phase output voltage. bo_s * v bo_s * After passing through a notch filter, the final reference value of the b-phase inverter side arm voltage, V, is obtained. bo * and based on v bo * The duration t of the non-O switching state of phase b when the cost function of phase b output voltage is minimized is obtained. b By comparing t b_NP and t b The magnitude is used to determine whether the state time of phase B O switch should be decomposed. (1) If t b ≥t b_NP Therefore, the switching state time of phase b is not decomposed. The condition of minimizing the output voltage cost function of phase b is prioritized. The task of suppressing the midpoint voltage fluctuation is transferred to phase c through the series structure. Based on the reference value v of the bridge arm voltage on the inverter side of phase b... bo* The duration t of the b-phase P switch state is obtained. b_p and the duration of the N switch state t b_n When v bo * When >0, the state action time t of phase b P switch is obtained. b_p At this time, the b-phase N switch state action time t b_n =0; when v bo * When <0, the state action time t of phase b N switch is obtained. b_n At this time, the b-phase P switch state action time t b_p =0. (2) If t b <t b_NP The duration of the O switch state is partially decomposed into equal durations of the P and N switches, and then allocated to the corresponding P and N switch state times to obtain the duration t of the b-phase P switch state. b_p N switch state action time t b_n Calculate the change in midpoint voltage v caused by the operation of phase b switch. NP_b (III) Controlling phase c. Assuming that the introduction of phases a, b, and c can achieve neutral voltage balance, the sum of the action times t of the N and P switch states required for phase c to achieve neutral voltage balance is obtained. c_NP The bisection method is used to optimize the reference voltage value of the C-phase inverter side bridge arm, and the optimized value v of the C-phase inverter side bridge arm voltage value that minimizes the cost function of the C-phase output voltage is obtained. co_s * After passing through the notch filter, it serves as the final reference value V for the bridge arm voltage of the c-phase inverter side. co * and based on v co * The duration t of the non-O switching state of phase c when the output voltage cost function is minimized is obtained. c By comparing t c_NP and t c The magnitude is used to determine whether the state time of phase C O switch should be decomposed. (1) If t c ≥t c_NP Therefore, the switching state time of phase C is not decomposed. The condition of minimizing the output voltage cost function is prioritized. The task of suppressing the midpoint voltage fluctuation is transferred to phase A through the series structure. Based on the reference value v of the bridge arm voltage on the inverter side of phase C... co * The duration t of the c-phase P switch state is obtained. c_p and the duration of the N switch state t c_n When v co * When >0, the state action time t of phase C P switch is obtained. c_pAt this time, the state action time of the c-phase N switch is t. c_n =0; when v co * <0 indicates the duration of the c-phase N switch state t c_n At this time, the duration of the C-phase P switch state is t. c_p =0. (2) If t c <t c_NP The duration of switch O is partially decomposed into equal durations of switch P and N, and then allocated to the corresponding P and N switch states to obtain the duration t of switch C. c_p N switch state action time t c_n Calculate the change in neutral voltage V caused by the operation of the c-phase switch. NP_c (iv) Finally, if the midpoint voltage fluctuation is 0 after considering the change in the three-phase midpoint voltage, then there is no need to decompose the O-switching state of phase a; if the midpoint voltage fluctuation still exists, then it is necessary to decompose the O-switching state of phase a. In this case, in order to satisfy the midpoint voltage balance, the sum of the action times t of the N-switching state and the P-switching state required for phase a is calculated. a_NP (1) If t a >t a_NP Therefore, the switching state time of phase a O is not decomposed, and the voltage reference value v of the inverter side bridge arm of phase a is used. ao * The action time t of the phase P switch state is obtained. a_p N switch state action time t a_n When v ao * >0 yields the action time t of phase P switch state. a_p At this time, the state of switch N in phase a is active for time t. a_n =0; when v ao * <0 indicates the action time t of phase N switch state. a_n At this time, the state of switch P in phase a is active for time t. a_p =0. (2) If t a <t a_NP The duration of switch O is partially decomposed into equal durations of switch P and N, and then allocated to the corresponding P and N switch states to obtain the duration t of switch P in phase a. a_p N switch state action time t a_n Finally, the duration of each phase's P-switch state and N-switch state is converted into a switching control signal to drive the switching on and off of the three bridge arm switches of the inverter using a modulation method. The duration of the non-O switching state of phase x is determined by the condition that phase satisfies the minimum output voltage cost function. x The sum of the durations of the N-switch and P-switch states required for this phase to achieve midpoint voltage balance, tx_NP The magnitude relationship determines the modulation method (x represents a, b, and c, indicating the three phases a, b, and c). When t x ≥t x_NP The x-phase S is obtained by using the dual-carrier SPWM modulation method. x1 ~S x4 The driving signal; when t x <t x_NP The x-phase S is obtained by using the DMW-CBPWM modulation method. x1 ~S x4 The driving signal.

[0097] Specifically, the multi-objective model predictive control method for a three-phase four-wire three-level inverter includes the following steps:

[0098] Step S1: Establish and discretize the system mathematical model.

[0099] Sampling three-level inverter three-phase output voltage v ca v cb v cc The current i of the filter inductor on the three-phase inverter side a i b i c and three-phase load current i oa i ob i oc The system mathematical model is established as follows:

[0100] (1)

[0101] (2)

[0102] Among them, L a L b L c These are the filter inductors for phases a, b, and c of a three-phase LC filter; R and C are also present. f These are the parasitic resistance of the filter inductance and the filter capacitance of a three-phase LC filter, respectively; v ao v bo v co These are the inverter side bridge arm voltages for phases a, b, and c, respectively.

[0103] Discretizing equations (1) and (2) using the forward Euler method, the system discrete model is:

[0104] (3)

[0105] (4)

[0106] Among them, T s One sampling period; ia (k), i b (k), i c (k) represents the inverter-side filter inductor currents of phases a, b, and c at time k; v ao (k), v bo (k), v co (k) represents the inverter-side bridge arm voltages of phases a, b, and c at time k; v ca (k), v cb (k), v cc (k) represents the output voltages of phases a, b, and c at time k; i oa (k), i ob (k), i oc (k) represents the load currents of phases a, b, and c at time k; i a (k+1),i b (k+1),i c (k+1) represents the inverter-side filter inductor currents of phases a, b, and c at time k+1; v ao (k+1), v bo (k+1), v co (k+1) represents the inverter-side bridge arm voltages of phases a, b, and c at time k+1; v ca (k+1), v cb (k+1), v cb (k+1) represents the output voltages of phases a, b, and c at time k+1; i oa (k+1),i ob (k+1),i oc (k+1) represents the load currents of phases a, b, and c at time k+1.

[0107] Step S2: Calculate the predicted value of the three-phase output voltage at time k+2.

[0108] From step S1, we obtain the predicted values ​​v of the output voltages of phases a, b, and c at time k+2. ca (k+2), v cb (k+2), v cc (k+2) is:

[0109] (5)

[0110] Since the fundamental period is much larger than the sampling period, it is assumed that i oa (k+1)≈i oa (k), i ob (k+1)≈i ob (k), i oc (k+1)≈i oc (k), then the prediction model for the inverter output voltage at time k+2 can be obtained as follows:

[0111] (6)

[0112] Step S3: Construct the three-phase output voltage cost function

[0113] The cost function g of the output voltage of the prediction model for phases a, b, and c at time k+2 ca (k+2), g cb (k+2), g cc (k+2) is:

[0114] (7)

[0115] Among them, v ca * (k+2), v cb * (k+2), v cc * (k+2) represents the reference values ​​of the output voltages of phases a, b, and c at time k+2.

[0116] Because the use of a notch filter results in phase lag, it is necessary to adjust the reference values ​​v of the three-phase output voltages (a, b, and c). ca * (k+2), v cb * (k+2), v cc * (k+2) is used for phase correction.

[0117] Reference value of output voltage of phases a, b, and c at time k+2: v ca * (k+2), v cb * (k+2), v cc * (k+2) is

[0118] (8)

[0119] in, The amplitude of the three-phase output voltage reference value. To output the fundamental angular frequency of the reference voltage, This represents the angular deviation of the fundamental signal after passing through the notch filter. The sampling period is For the fundamental period, .

[0120] Step S4: Predictive Control of Series Structure with Notch Filter

[0121] The reference values ​​v of the output voltages of phases a, b, and c at time k+2 generated in step S3 ca* (k+2), v cb * (k+2), v cc * (k+2), together with the sampled signal at time k, the current i of the filter inductor on the three-phase inverter side. a (k), i b (k), i c (k), Three-phase output voltage v ca (k), v cb (k), v cc (k), Three-phase load current i oa (k), i ob (k), i oc (k) and the voltage v of DC side capacitors C1 and C2. c1 (k) and v c2 (k) together Figure 2 The predictive control model of the series structure with notch filter shown obtains the P-switching state action time t of the bridge arms of the three-phase inverters a, b, and c. a_p t b_p t c_p and the duration t of the N switch state a_n t b_n t c_n By combining the dual-carrier SPWM modulation method and the dual-modulated wave carrier modulation (DMW-CBPWM) method, the drive signal S for driving the switching transistors of the three-level inverter is obtained. a1 ~S a4 S b1 ~S b4 and S c1 ~S c4 .

[0122] like Figure 3 As shown, the specific implementation steps of predictive control for a series structure with a notch filter are as follows:

[0123] Step A1: In the series structure, the binary search method is used to optimize the reference voltage value of the inverter side bridge arm of phase a at time k, so as to obtain the reference voltage value v of the inverter side bridge arm of phase a at time k+2 that minimizes the cost function of the output voltage of phase a. ao_s * (k).

[0124] The specific process of the dichotomy method is as follows:

[0125] When the three-level inverter is in the P-switch state, v xo =V dc / 2, x represents a, b, and c, signifying the three phases a, b, and c; In the O switch state, v xo =0; When N is in switch state, vxo =-V dc / 2. At time k, the initial values ​​of the boundary values ​​on both sides are V. left =v xo (k)=-V dc / 2,V right =v xo (k)= V dc / 2, Substitute the boundary values ​​into step S2 to obtain the predicted value v of the inverter output voltage at time k+2. ca (k+2), v cb (k+2), v cc (k+2), and according to step S3, obtain the cost function value g of the inverter output voltage of the prediction model at time k+2. ca (k+2)=g ca (v ao (k)), g cb (k+2)= g cb (v bo (k)), g cc (k+2)=g cc (v co (k)). And according to equation (9), the optimization judgment is performed, and the midpoint V is set. mid For the current V left and V right The average value, i.e., V mid =(V left +V right ) / 2, V mid If the boundary value of one side is chosen as the boundary value of the next optimization, and the current output voltage cost function is smaller, then the boundary value of the other side is chosen as the boundary value of the next optimization. Then:

[0126] (9)

[0127] After m optimizations, the final boundary values ​​V on both sides are obtained. left and V right Finally, the reference value v of the inverter side bridge arm voltage at time k was determined. xo_s * (k) is:

[0128] (10)

[0129] Step A2: The reference value v of the inverter side bridge arm voltage at time k obtained in step A1. ao_s * (k) After passing through the notch filter, the final reference value v of the inverter side bridge arm voltage at time k is obtained. ao * (k).

[0130] The transfer function of the notch filter in the continuous domain is:

[0131] (11)

[0132] Where Q is the notch filter quality factor; This is the notch filter frequency.

[0133] Amplitude and frequency of notch filter and phase frequency The characteristics are:

[0134] (12)

[0135] To ensure effective resonance suppression; Equal to the system's resonant angular frequency ω r The notch width Δω = 0.1ω r At frequency ω r The notch depth A(ω) is ±Δω. r ±Δω) is the amplitude of the system output voltage resonant frequency that is less than or equal to the negative value.

[0136] Discretizing the notch filter transfer function using the bilinear transform method yields the discretized notch filter transfer function expression as follows:

[0137] (13)

[0138] Among them, T s The sampling period.

[0139] After passing through a notch filter, the reference value of the inverter side bridge arm voltage at time k is obtained. ao * The expression for (k) is:

[0140] (14)

[0141] in, ; ; ;v ao_s * (k-1) 、v ao_s * (k-2) represent the reference voltages of the inverter side bridge arm of phase a at times k-1 and k-2, respectively, obtained through the binary search method; v ao * (k-1), v ao * (k-2) represent v respectively ao_s * (k-1) 、v ao_s *(k-2) The reference voltage of the inverter side bridge arm of phase a at time k-1 and k-2 after passing through the notch filter.

[0142] Step A3: Based on the reference value v of the inverter side bridge arm voltage at time k obtained in step A2... ao * (k) Calculate the duration t of the non-O switching state of phase a when the output voltage cost function is minimized. a for:

[0143] (15)

[0144] Calculate the reference value v of the bridge arm voltage on the inverter side of phase a. ao * The change in midpoint voltage caused by (k) v NP_a (k):

[0145] (16)

[0146] Step A4: For the midpoint voltage control of phase b, calculate the sum of the action times t of the N switch state and the P switch state required for phase b to achieve midpoint voltage balance, assuming that the introduction of phase b can achieve midpoint voltage balance. b_NP for:

[0147] (17)

[0148] Among them, v NP (k)=v C1 (k)-v C2 (k), C dc For the input DC side capacitor values, C1=C2=C dc .

[0149] For t b_NP Limit the amplitude: when t b_NP When < 0, t b_NP =0; To prevent direct transitions between P and N switch states, a certain time t0 needs to be allowed between the P and N switch states. b_NP >T s When -2t0, t b_NP =T s -2t0.

[0150] Step A5: For the b-phase output voltage control, the bisection method as in Step A1 is used to optimize the b-phase inverter side bridge arm voltage reference value. The optimized b-phase inverter side bridge arm voltage reference value v that minimizes the b-phase output voltage cost function is obtained. bo_s * (k), and then after passing through the notch filter as in step A2, the reference value v of the bridge arm voltage on the b-phase inverter side at time k is obtained.bo * (k):

[0151] (18)

[0152] Among them, v bo_s * (k-1) 、v bo_s * (k-2) represent the reference voltages of the b-phase inverter side bridge arm obtained by binary search at times k-1 and k-2, respectively; v bo * (k-1), v bo * (k-2) represent v respectively bo_s * (k-1) 、v bo_s * (k-2) Reference voltage of the b-phase inverter side bridge arm after passing through the notch filter.

[0153] Step A6: Based on the reference value v of the bridge arm voltage on the b-phase inverter side at time k obtained in step A5. bo * (k) Calculate the duration t of the non-O switching state of phase b when the output voltage cost function of phase b is minimized. b :

[0154] (19)

[0155] Step A7: By comparing t b_NP With t b The magnitude of the value is used to determine whether the state time of phase b O switch needs to be decomposed, and the state time t of phase b bridge arm P switch is calculated. b_p and N switch state time t b_n The details are as follows:

[0156] (1) If t b ≥t b_NP Therefore, the duration of the b-phase O switch state is not decomposed; the condition of minimizing the output voltage cost function is prioritized, and the task of suppressing midpoint voltage fluctuations is transferred to phase c through a series structure. Based on the b-phase inverter side bridge arm voltage reference value v... bo * (k) and formula (20) yield the b-phase P switch state action time t b_p and the duration of the N switch state t b_n :

[0157] (20)

[0158] (2) If t b <tb_NP The duration of the O switch state is partially decomposed into equal P and N switch state times, and then allocated to the corresponding P and N switch state durations to obtain the duration t of the b-phase P switch state. b_p N switch state action time t b_n .

[0159] When v bo * When (k)>0, it corresponds to phase b t b_p t b_n for:

[0160] (twenty one)

[0161] When v bo * When (k) < 0, it corresponds to phase b t. b_p t b_n for:

[0162] (twenty two)

[0163] Step A8: Based on the b-phase P switch state action time t obtained in step A7 b_p and the duration of the N switch state t b_n Calculate the change in midpoint voltage v caused by the operation of phase b switch. NP_b (k) is:

[0164] (twenty three)

[0165] Step A9: For the midpoint voltage control of phase c, assuming that the midpoint voltage can be balanced after introducing phases a, b, and c, obtain the sum of the action times t of the N switch state and the P switch state required for phase c to achieve midpoint voltage balance. c_NP for:

[0166] (twenty four)

[0167] For t c_NP Limit the amplitude: when t c_NP When < 0, t c_NP =0; To prevent direct transitions between P and N switch states, a certain time t0 needs to be allowed between the P and N switch states. c_NP >T s When -2t0, t c_NP =T s -2t0.

[0168] Step A10: For c-phase output voltage control, the bisection method as in step A1 is used to optimize the c-phase inverter side bridge arm voltage reference value. The optimized c-phase inverter side bridge arm voltage reference value v that minimizes the c-phase output voltage cost function is obtained. co_s * (k), and then after passing through the notch filter as in step A2, the reference value v of the c-phase inverter side bridge arm voltage at time k is obtained. co * (k):

[0169] (25)

[0170] Among them, v co_s * (k-1) 、v co_s * (k-2) represent the reference voltages of the c-phase inverter side bridge arm obtained by the binary search method at times k-1 and k-2, respectively; v co * (k-1), v co * (k-2) represent v respectively co_s * (k-1) 、v co_s * (k-2) Reference voltage of the c-phase inverter side bridge arm after passing through the notch filter.

[0171] Step A11: Based on the reference value v of the inverter side bridge arm voltage at time k obtained in step A10... co * (k), calculate the duration t of the non-O switching state of phase c when the output voltage cost function is minimized. c .

[0172] (26)

[0173] Step A12: By comparing t c_NP With t c The magnitude of the value is used to determine whether the state time of phase C O switch needs to be decomposed, and the state action time t of phase C bridge arm P switch is calculated. c_p and the duration of the N switch state t c_n The details are as follows:

[0174] (1) If t c ≥t c_NP Therefore, the duration of the c-phase O switch state is not decomposed; the condition of minimizing the output voltage cost function is prioritized, and the task of suppressing midpoint voltage fluctuations is transferred to phase a through a series structure. Based on the reference value v of the c-phase inverter side bridge arm voltage... co *(k) and formula (27) yield the duration of the P-phase switch state and the duration of the N-phase switch state:

[0175] (27)

[0176] (2) If t c <t c_NP The duration of the O switch state is partially decomposed into equal durations of the P and N switches, and then allocated to the corresponding P and N switch state times to obtain the duration t of the c-phase P switch state. c_p N switch state action time t c_n .

[0177] When v co * When (k)>0, it corresponds to phase ct. c_p t c_n for:

[0178] (28)

[0179] When v co * When (k) < 0, it corresponds to phase c t. c_p t c_n for:

[0180] (29)

[0181] Step A13: Based on the c-phase P switch state action time t obtained in step A12 c_p and the duration of the N switch state t c_n Calculate the change in neutral voltage V caused by the operation of phase c switch. NP_c (k) is:

[0182] (30)

[0183] Step A14: In order to satisfy the midpoint voltage balance within one sampling period, the sum of the action times of the N switch state and the P switch state required for phase a is t. a_NP for:

[0184] (31)

[0185] For t a_NP Limit the amplitude: when t a_NP When < 0, t a_NP =0; To prevent direct transitions between P and N switch states, a certain time t0 needs to be allowed between the P and N switch states. a_NP >T s When -2t0, t a_NP =T s-2t0.

[0186] Step A15: Compare t obtained in step A3 a With t a_NP The magnitude of the value is used to determine whether the action time of the phase a O switch state should be decomposed, and the action time t of the phase a bridge arm P switch state is calculated. a_p and the duration of the N switch state t a_n The details are as follows:

[0187] (1) If t a ≥t a_NP Therefore, the duration of the phase a O switch state is not decomposed. Based on the reference value v of the phase a inverter side bridge arm voltage obtained in step A2... ao * (k) and formula (32) yield the action time t of the P-phase switch state. a_p and the duration of the N switch state t a_n :

[0188] (32)

[0189] (2) If t a <t a_NP The duration of the O switch state in phase a is partially decomposed into equal durations of the P and N switches, and then allocated to the corresponding P and N switch state times in phase a, thus obtaining the duration t of the P switch state in phase a. a_p N switch state action time t a_n .

[0190] When v ao * When (k)>0, it corresponds to phase t of phase a. a_p t a_n for:

[0191] (33)

[0192] When v ao * When (k) < 0, it corresponds to phase t of phase a. a_p t a_n for:

[0193] (34)

[0194] Step A16: Determine the duration (t) of the corresponding P and N switch states of a, b, and c within the sampling period. a_p , t a_n ;t b_p , t b_n ;t c_p , t c_nThis is converted into a switching control signal to drive the switching of the three bridge arm switches of the inverter. The non-zero switching state of phase x is determined by the time t when the output voltage cost function is minimized. x The sum of the durations of the N-switch and P-switch states required for this phase to achieve midpoint voltage balance, t x_NP The magnitude relationship determines the modulation method (x represents a, b, and c, indicating the three phases a, b, and c). When t x ≥t x_NP ,pass Figure 4 The dual-carrier SPWM modulation method shown in Table 1 and the correspondence between the on / off state and the switching state of the switch transistors are used to obtain the x-phase S x1 ~S x4 The driving signal; when t x <t x_NP ,pass Figure 5 The DMW-CBPWM modulation method shown and the correspondence between the on / off state and the switching state of the switch transistor in Table 1 are used to obtain the x-phase S x1 ~S x4 The driving signal is then received. The process continues until the signal enters the next sampling period, at which point the above steps are repeated.

[0195] Table 1. Correspondence between switch on / off state and switch status of three-phase four-wire T-type three-level inverter

[0196]

[0197] This embodiment also provides a controller, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the multi-objective model predictive control method for the three-phase four-wire three-level inverter described above.

[0198] This embodiment also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the multi-objective model predictive control method for the three-phase four-wire three-level inverter described above.

[0199] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0200] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0201] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0202] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0203] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A multi-objective model predictive control method for a three-phase four-wire three-level inverter, characterized in that, Includes the following steps: Step S1: Sample the three-phase output voltage, the three-phase inverter side filter inductor current, and the three-phase load current of the three-phase four-wire three-level inverter, establish the system mathematical model, and discretize it using the forward Euler method to obtain the system discrete model; Step S2: Calculate the predicted value of the three-phase output voltage at time k+2 based on the discrete model of the system. Step S3: Perform phase correction on the three-phase output voltage reference value at time k+2, and construct the three-phase output voltage cost function based on the corrected three-phase output voltage reference value at time k+2 and the predicted value of the three-phase output voltage at time k+2. Step S4: The reference value of the three-phase output voltage at time k+2, together with the sampling signal at time k, is passed through the series structure model prediction control with notch filter to obtain the P switch state action time and N switch state action time of the three-phase inverter side bridge arm; wherein, the sampling signal includes the filter inductor current, three-phase output voltage, three-phase load current and the voltages of the two DC side capacitors. Step S5: Based on the operating time of the P switch state and the operating time of the N switch state, and combined with the modulation method, obtain the driving signal for driving the switching transistors of the three-level inverter. The series structure model predictive control with notch filter includes arranging the output voltage cost functions of phases a, b, and c in a series structure, controlling phases a, b, and c sequentially, and comparing the duration t of the non-zero switching state when each phase satisfies the minimum output voltage cost function. x The sum of the durations of the N-switch and P-switch states required to achieve midpoint voltage balance, t x_NP The magnitude relationship determines whether to decompose the action time of the O switch state of that phase in order to achieve multi-objective optimization control of output voltage quality and neutral point voltage balance. x represents a, b, and c, which represent the three phases a, b, and c.

2. The multi-objective model predictive control method for a three-phase four-wire three-level inverter according to claim 1, characterized in that, The three-phase four-wire three-level inverter adopts a split capacitor three-level inverter structure, consisting of an input DC source, a DC-side capacitor composed of two equal capacitors connected in series, a three-arm three-level inverter bridge, a three-phase LC filter, and a three-phase load. The midpoint of the DC-side capacitor is connected to the neutral point of the AC-side load of the inverter to form a three-phase four-wire system. Each phase of the three-level inverter has three switching states: O switching state, P switching state, and N switching state.

3. The multi-objective model predictive control method for a three-phase four-wire three-level inverter according to claim 1, characterized in that, In step S4, the model predictive control of the series structure with notch filter specifically includes: Step A1: Use the binary search method to optimize the reference value of the inverter side bridge arm voltage at time k, and obtain the reference value of the inverter side bridge arm voltage at time k+2 that satisfies the minimum cost function of the output voltage of phase a at time k+2. Step A2: After processing the reference voltage value of the inverter side bridge arm of phase a obtained in step A1 through a notch filter, the final reference voltage value v of the inverter side bridge arm of phase a at time k is obtained. ao * (k); Step A3: Based on the reference value v of the inverter side bridge arm voltage obtained in step A2... ao * (k) Calculate the duration t of the non-O switching state of phase a when the output voltage cost function of phase a is minimized. a And the reference value v of the bridge arm voltage on the inverter side of phase a. ao * The change in midpoint voltage caused by (k) v NP_a (k); Among them, V dc For the input DC source voltage, T s One sampling period; i a (k) represents the current in the filter inductor on the inverter side of phase a at time k, C dc The capacitance values ​​of the two equal capacitors on the DC side are used for input. Step A4: Calculate the sum of the action times t of the N-switch and P-switch states required for phase b to achieve neutral voltage balance after the introduction of phase b. b_NP ; Among them, v NP (k) represents the difference between the upper and lower capacitor voltages on the input DC side at time k, i b (k) represents the filter inductor current on the b-phase inverter side at time k; Step A5: Optimize the reference voltage value of the b-phase inverter side bridge arm using the binary search method, and obtain the reference voltage value v of the b-phase inverter side bridge arm at time k after processing with a notch filter. bo * (k); Step A6: Based on the b-phase inverter side bridge arm voltage reference value v obtained in step A5 bo * (k) Calculate the duration t of the non-O switching state of phase b when the output voltage cost function of phase b is minimized. b ; Step A7: Compare t b_NP With t b The magnitude of the value determines whether the state action time of phase b O switch should be decomposed, and the state action time t of phase b bridge arm P switch is calculated. b_p and the duration of the N switch state t b_n ; Step A8: Based on the operating time of switch P and switch N in phase b obtained in step A7, calculate the change in midpoint voltage v caused by the operation of the phase b switch. NP_b (k); Step A9: Calculate the sum of the action times t of the N-switch and P-switch states required for phase c to achieve neutral voltage balance after the introduction of phase c. c_NP ; Among them, i c (k) represents the current of the filter inductor on the inverter side of phase c at time k; Step A10: Optimize the reference voltage value of the c-phase inverter side bridge arm using the binary search method, and obtain the reference voltage value v of the c-phase inverter side bridge arm at time k after processing with a notch filter. co * (k); Step A11: Based on the reference value v of the c-phase inverter side bridge arm voltage obtained in step A10... co * (k), calculate the duration t of the non-O switching state of phase c when the cost function of phase c output voltage is minimized. c ; Step A12: Compare t c_NP With t c The magnitude of the value determines whether the state action time of phase C O switch should be decomposed, and the state action time t of phase C bridge arm P switch is calculated. c_p and the duration of the N switch state t c_n ; Step A13: Based on the operating time of the P switch and the N switch obtained in step A12, calculate the change in midpoint voltage v caused by the operation of the c-phase switch. NP_c (k); Step A14: Calculate the sum of the action times t of the N switch state and the P switch state required to achieve the neutral point voltage balance in phase a. a_NP ; Step A15: Compare t a With t a_NP The magnitude of the value determines whether the action time of the phase a O switch state should be decomposed, and the action time t of the phase a bridge arm P switch state is calculated. a_p and the duration of the N switch state t a_n ; Step A16: Convert the P switch state duration and N switch state duration of phases a, b, and c within this sampling period into switching transistor drive control signals; based on the non-O switch state duration t of each phase when the output voltage cost function is minimized... x The sum of the durations of the N-switch and P-switch states required to achieve midpoint voltage balance, t x_NP The magnitude relationship determines the modulation method, when t x ≥t x_NP When using dual-carrier SPWM modulation method, when t x <t x_NP The dual-modulation wave carrier pulse width modulation method is used to obtain the driving signal of each phase switch transistor.

4. The multi-objective model predictive control method for a three-phase four-wire three-level inverter according to claim 3, characterized in that, In step A1, the dichotomy method includes: Set the initial values ​​V of the two boundary values ​​of the inverter side bridge arm voltage at time k. left =-V dc / 2,V right =V dc / 2, Substitute the boundary values ​​into step S2 to obtain the predicted value of the inverter output voltage at time k+2, and based on the output voltage cost function value at time k+2 obtained in step S3, set the midpoint V. mid For the current V left and V right The average value, i.e., V mid =(V left +V right ) / 2, V mid As the boundary value for one side of the next optimization, the boundary value with the smaller current output voltage cost function is selected as the boundary value for the other side of the next optimization. After m optimizations, the final boundary values ​​V on both sides are obtained. left and V right The reference value of the inverter side bridge arm voltage at time k is determined to be the final value V at both boundary values. left and V right The average value.

5. The multi-objective model predictive control method for a three-phase four-wire three-level inverter according to claim 3, characterized in that, In step A2, the transfer function of the notch filter in the continuous domain is: Where Q is the notch filter quality factor. The notch frequency; Equal to the system's resonant angular frequency ω r The notch width Δω = 0.1ω r At frequency ω r The notch depth A(ω) at ±Δω is... r The amplitude of the system output voltage resonant frequency (±Δω) is less than or equal to the negative value; the notch filter transfer function is discretized using a discretization method.

6. The multi-objective model predictive control method for a three-phase four-wire three-level inverter according to claim 3, characterized in that, In step A7, determining whether to decompose the action time of the b-phase O switch state specifically includes: If t b ≥t b_NP Therefore, the duration of the b-phase O switch state is not decomposed. The condition of minimizing the output voltage cost function of the b-phase is prioritized. The task of suppressing the midpoint voltage fluctuation is transferred to the c-phase through a series structure. The duration t of the b-phase P switch state is determined based on the sign of the reference value of the bridge arm voltage on the b-phase inverter side. b_p and the duration of the N switch state t b_n ; If t b <t b_NP Then, the duration of the b-phase O switch state is partially decomposed into equal durations of the P and N switches, and allocated to the corresponding P and N switch state durations to obtain the duration t of the b-phase P switch state. b_p and the duration of the N switch state t b_n ; When v bo * When (k)>0, it corresponds to phase b t b_p t b_n for: When v bo * When (k) < 0, it corresponds to phase b t. b_p t b_n for: 。 7. The multi-objective model predictive control method for a three-phase four-wire three-level inverter according to claim 3, characterized in that, In step A12, determining whether to decompose the action time of the c-phase O switch state specifically includes: If t c ≥t c_NP Therefore, the duration of the C-phase O switch state is not decomposed. The condition of minimizing the C-phase output voltage cost function is prioritized. The task of suppressing midpoint voltage fluctuations is transferred to phase A through a series structure. The duration t of the C-phase P switch state is determined based on the sign of the reference value of the C-phase inverter side bridge arm voltage. c_p and the duration of the N switch state t c_n ; If t c <t c_NP Then, the duration of the c-phase O switch state is partially decomposed into equal durations of the P and N switches, and allocated to the corresponding P and N switch state durations to obtain the duration t of the c-phase P switch state. c_p and the duration of the N switch state t c_n ; When v co * When (k)>0, it corresponds to phase ct. c_p t c_n for: When v co * When (k) < 0, it corresponds to phase c t. c_p t c_n for: 。 8. The multi-objective model predictive control method for a three-phase four-wire three-level inverter according to claim 3, characterized in that, In step A15, determining whether to decompose the action time of the phase a O switch state specifically includes: If t a ≥t a_NP Therefore, the duration of the phase a O switch state is not decomposed; instead, the duration t of the phase a P switch state is determined based on the sign of the reference voltage value of the phase a inverter side bridge arm. a_p and the duration of the N switch state t a_n ; If t a <t a_NP Then, the duration of the O switch state in phase a is partially decomposed into equal durations of the P and N switches, and allocated to the corresponding P and N switch state durations in phase a, thus obtaining the duration t of the P switch state in phase a. a_p and the duration of the N switch state t a_n ; When v ao * When (k)>0, it corresponds to phase t of phase a. a_p t a_n for: When v ao * When (k) < 0, it corresponds to phase t of phase a. a_p t a_n for: 。 9. The multi-objective model predictive control method for a three-phase four-wire three-level inverter according to claim 1, characterized in that, In step S3, the phase correction of the three-phase output voltage reference value at time k+2 specifically involves: performing phase compensation on the three-phase output voltage reference value at time k+2 based on the angle deviation of the fundamental signal of the output reference voltage after passing through the notch filter; the correction formula for the phase correction is: Among them, v ca * (k+2), v cb * (k+2), v cc * (k+2) represents the reference values ​​of the output voltages of phases a, b, and c at time k+2. The amplitude of the three-phase output voltage reference value. To output the fundamental angular frequency of the reference voltage, This represents the angular deviation of the fundamental signal after passing through the notch filter. The sampling period is For the fundamental period, .

10. A multi-objective model predictive control system for a three-phase four-wire three-level inverter, used to implement the method as described in any one of claims 1-9, characterized in that, include: The sampling module is used to sample the three-phase output voltage, the three-phase inverter-side filter inductor current, the three-phase load current, and the voltages of the two DC-side capacitors of the three-phase four-wire three-level inverter. The modeling and discretization module is used to establish the mathematical model of the system and discretize it using the forward Euler method to obtain the discrete model of the system. The prediction module is used to calculate the predicted value of the three-phase output voltage at time k+2 based on the discrete model of the system. The cost function construction module is used to construct the three-phase output voltage cost function based on the predicted value of the three-phase output voltage at time k+2 and the reference value of the three-phase output voltage at time k+2 after phase correction. The series structure model predictive control module is used to take the reference value of the three-phase output voltage at time k+2 and the sampled signal at time k together and pass them through the series structure model predictive control with notch filter to obtain the P switch state action time and N switch state action time of the three-phase inverter side bridge arm. The drive signal generation module is used to generate a drive signal for driving the switching transistors of the three-level inverter based on the operating time of the P switch state and the operating time of the N switch state combined with a modulation method. The series structure model predictive control with notch filter includes arranging the output voltage cost functions of phases a, b, and c in a series structure, controlling phases a, b, and c sequentially, and comparing the duration t of the non-zero switching state when each phase satisfies the minimum output voltage cost function. x The sum of the durations of the N-switch and P-switch states required to achieve midpoint voltage balance, t x_NP The magnitude of the voltage difference determines whether the duration of the phase O switch state should be decomposed to achieve multi-objective optimization control of output voltage quality and midpoint voltage balance.

Images