A fair sequence model predictive control multi-objective optimization method for parallel t-type three-level rectifiers

Abstract

The application discloses a fair and sequential model predictive control multi-objective optimization method for a parallel T-type three-level rectifier, and aims at solving the problems of power distribution, midpoint voltage balance, circulating current suppression, DC voltage stability and grid current tracking in a parallel T-type three-level rectifier system in a high-power rectifier occasion. The application adopts double closed-loop control, the outer ring adopts adaptive droop control with voltage feedforward, realizes the stability of DC voltage and load power distribution, the inner ring adopts fair and sequential model predictive control, through the fair priority order, not only the advantage that the sequential model predictive control does not need to select a weight factor is reserved, but also the problem that the optimal solution cannot be selected due to the fixed priority order of the sequential model predictive control is solved, the balance of the midpoint voltage, the circulating current suppression and the tracking of the grid current are realized, and the overall scheme realizes the multi-objective optimization control of the parallel system.

Description

Technical Field

[0001] This invention belongs to the field of power electronic converter control technology, specifically relating to a fair sequence model predictive control multi-objective optimization method for parallel T-type three-level rectifiers. Background Technology

[0002] Currently, with the widespread application of new energy vehicles, large-scale DC charging technology is increasingly becoming the development trend of new energy vehicle charging. Furthermore, high-power rectifiers are required in railway locomotive traction, mine hoisting systems, and wind power generation systems. The superior power density and current harmonics of T-type three-level rectifiers have led to their wider application in high-power applications. When a single T-type three-level rectifier cannot meet the load power demand, a parallel structure is usually used to further increase the power rating.

[0003] However, parallel T-type three-level rectifiers face challenges in power distribution, DC voltage stability, midpoint voltage balance, and circulating current suppression. To address the multi-objective optimization problem of parallel T-type three-level rectifiers, a finite set model predictive control method with multi-objective optimization capabilities is employed.

[0004] This invention employs a dual closed-loop control strategy. The outer loop uses adaptive droop control with voltage feedforward to solve DC voltage stability and power distribution problems, while the inner loop uses fair sequential model predictive control as a multi-objective optimized current controller.

[0005] Droop control is a common method for controlling DC voltage and solving power distribution problems. Its basic principle is to linearly reduce the voltage as the output current increases, but this inevitably leads to DC voltage droop. To address this, adaptive droop control is employed, where the droop coefficient continuously changes with the DC current to reduce DC voltage droop. Simultaneously, to compensate for DC voltage droop, a feedforward loop is used to balance the DC voltage with a reference value.

[0006] Finite set model predictive control (FMCC) is a control method with multi-objective optimization capabilities. Compared to traditional control methods, it does not require PWM modulation; instead, it directly outputs the optimal switching sequence by designing an optimization problem, i.e., a cost function. However, traditional FMCC faces the challenge of selecting the weighting factors for the multi-objective cost function. To address this, some studies have employed sequential model predictive control (SMC) to avoid selecting weighting factors for the cost function. However, this introduces the problem of prioritizing SMCC. Therefore, we propose fair sequential model predictive control, which not only avoids the weighting factor selection problem but also solves the priority selection problem in SMCC. Summary of the Invention

[0007] To achieve the above objectives, the present invention adopts the following technical solution.

[0008] A fair sequence model predictive control multi-objective optimization method for parallel T-type three-level rectifiers includes the following steps:

[0009] Step S1: Acquire DC voltage U using voltage and current sensors. dc DC current I of the two rectifiers dcx The voltage V of the two capacitors on the DC side px and V nx Three-phase power grid voltage;

[0010] Step S2: Estimate the three-phase rectifier-side current i of the two rectifiers using a Kalman filter. x2 and the three-phase filter capacitor voltage U cx ;

[0011] Step S3: Obtain the phase information of the grid voltage e through a phase-locked loop. The adaptive droop control with voltage feedforward calculates the reference value of the grid-side current based on the set DC voltage reference value, the DC current of each of the two rectifiers, the rated power of each of the two rectifiers, the DC voltage, and the initial droop coefficient. The reference value of the rectifier-side current is then derived from the circuit relationship of the LCL-type grid-connected inverter.

[0012] Step S4: The reference value of the rectifier side current is delayed by one step using the Lagrange extrapolation method and then sent to the fair sequence model predictive controller for cost function tracking.

[0013] Step S5: Based on the relationship between common-mode voltage and switching state, select the seven switching states with a common-mode voltage of 0 as the initial candidate vectors;

[0014] Step S6: Based on the collected rectifier side current, midpoint voltage difference, LCL capacitor voltage, DC voltage, and rectifier side reference current after Lagrange extrapolation delay, calculate the cost functions G1 and G2 values ​​for each of the seven switching states, and store the calculation results and the corresponding switching states.

[0015] Step S7: Calculate the average cost function of the seven switch states. Switch states with a cost function less than the average value are those to be cross-calculated.

[0016] Step S8: Normalize and sort the seven stored switch states and their cost function values. After sorting, each normalized value still has a one-to-one correspondence with the switch state.

[0017] Step S9: The portion of the cost function S1 that is less than the average value Combination 1, the portion where the cost function G2 is less than the average. Combination 2 is formed by selecting the cross vectors based on the average value and extracting the cost function of the cross portion from another cost function. and The values ​​of these combinations are added together to form a new combination Y. The minimum cost function in combination Y is the optimal solution S. opt (k),

[0018] Step S10: Place S opt (k) is converted into a switching vector, which is then converted into the on / off signal of the switching device of the T-type three-level grid-connected inverter and sent in at the next sampling time. Step S1 is executed again at the next sampling time.

[0019] Furthermore, the DC voltage reference value in step S3 is 500V.

[0020] Furthermore, the cost function described in step S4 is:

[0021]

[0022] In the formula, i nx2 (n = A, B, C), These represent the rectifier-side current of the xth rectifier and its reference value, respectively.

[0023] Furthermore, the expression for the number of cross switch states mentioned in step S7 is:

[0024]

[0025] In the formula, L and N represent the number of vectors less than the average value in the cost functions G1 and G2, respectively, and are also the number of cross vectors.

[0026] Furthermore, in step S9, both cost function values ​​in the combination Y are normalized, indicating that they have the same priority.

[0027] Furthermore, the expression for the optimal solution described in step S9 is: S opt (k)=argmin s() {Y(S i (k),(k))|i=1,…,2},where, s opt (k) represents the optimal switching vector calculated at time k, n2 represents the sum of the number of cross vectors, and Y represents the new cost function after combination.

[0028] Furthermore, the optimal solution does not use a fixed priority; it prioritizes either the cost function G1 or the cost function G2.

[0029] The technical principle of this invention is as follows: In high-power applications, a single T-type three-level rectifier cannot meet the power requirements. Parallel rectifiers can improve the system's power rating. However, parallel T-type three-level rectifiers face challenges such as power distribution, midpoint voltage balance, circulating current suppression, DC voltage stabilization, and grid current tracking. By employing dual closed-loop control, the outer loop uses adaptive droop control with voltage feedforward to achieve power distribution and DC voltage stabilization, while the inner loop uses fair sequence model predictive control to achieve midpoint voltage balance, circulating current suppression, and grid current tracking. The overall scheme achieves multi-objective optimized control of parallel T-type three-level rectifiers.

[0030] The outer loop employs adaptive droop control with voltage feedforward to achieve power distribution of the parallel system load. To address the droop problem caused by DC voltage, an adaptive droop coefficient that continuously changes with the DC current is used to reduce DC voltage droop. Simultaneously, the difference between the DC voltage and the DC voltage reference value is used as a feedforward element to compensate for the DC voltage drop. The inner loop employs fair-sequence model predictive control with a non-fixed priority order. This not only retains the advantage of sequential model predictive control that does not require selecting weighting factors, but also solves the problem of sequential model predictive control's inability to select the optimal solution due to a fixed priority order.

[0031] The method of this invention can solve the problems of parallel rectifiers in high-power rectifier applications, such as power distribution, midpoint voltage balance, circulating current suppression, DC voltage stability, grid current tracking, complex control system, and heavy processing load.

[0032] The dual closed-loop rectifier control method of the present invention includes:

[0033] The outer loop employs adaptive droop control with voltage feedforward. The sampled DC voltage and DC current are used to obtain the grid-side current i through the adaptive droop controller and a proportional-integral (PI) regulator. x1 The d-axis reference value is used to ensure a high power factor for the rectifier, and the grid-side current i... x1 The q-axis reference value is set to 0, and then the three-phase current reference value on the rectifier side is obtained by modeling and calculating the LCL filter. Meanwhile, in order to reduce the impact of droop control on DC voltage, voltage feedforward is used to compensate for the reduction in DC voltage.

[0034] The inner loop employs fair-sequence model predictive control, with two cost functions representing two levels. The two cost function values ​​for the initial candidate vectors are calculated separately, and then the average cost function is calculated. Candidate vectors with a cost function value less than the average are selected for cross-calculation. The two cost function values ​​for all cross-vectors are determined, and the normalized cost function values ​​are summed to form a new combination Y. The optimal switching vector S is then selected from combination Y.opt (k) serves as the drive signal for the rectifier switching devices.

[0035] Compared with the prior art, the present invention has the following technical advantages:

[0036] 1. This invention adopts a dual closed-loop control strategy. The outer loop uses an adaptive droop controller with voltage feedforward to control the DC voltage and distribute the power. The inner loop uses fair sequential model predictive control with multi-objective optimization function to suppress circulating current from the control level and achieve balance of the midpoint voltage.

[0037] 2. This invention optimizes the initial candidate vectors, reducing the number of initial candidate vectors from 27 to 7, thereby solving the problem of circulating current suppression from the control level without increasing any hardware cost of the system. At the same time, it reduces the number of loop operations and alleviates the burden on the processing unit.

[0038] 3. This invention adopts fair sequential model predictive control technology, which does not require PWM modulation. At the same time, it not only retains the advantage of sequential model predictive control that it does not need to select weight factors, but also solves the problem that sequential model predictive control cannot select the optimal solution due to priority order, thus ensuring the balance of the midpoint voltage and the tracking of the grid-side current. Attached Figure Description

[0039] Figure 1 : Topology diagram of two parallel T-type three-level rectifiers in this invention;

[0040] Figure 2 : Schematic diagram of the droop control power distribution principle in this invention;

[0041] Figure 3 : Control block diagram of the two parallel T-type three-level rectifier system in this invention;

[0042] Figure 4 The grid-side current and AC-side power waveforms of rectifiers with different power ratings in this invention are shown in the diagrams. The left diagram shows the grid-side current at different power ratings, and the right diagram shows the AC-side power at different power ratings.

[0043] Figure 5 The waveform diagrams comparing the circulating current suppression effect and the midpoint voltage balance effect of different control methods under the power grid imbalance condition in this invention are shown. The left figure is the circulating current suppression effect under the power grid imbalance condition, and the right figure is the midpoint voltage balance effect under the power grid imbalance condition.

[0044] Figure 6 The waveforms of DC voltage and AC power changes under varying voltage reference values ​​in this invention are shown in the diagrams. The left diagram shows the DC voltage recovery effect after the voltage reference value changes, and the right diagram shows the power change after the voltage reference value changes. Detailed Implementation

[0045] The technical solution will be clearly and completely described below with reference to preferred embodiments and accompanying drawings. It should be understood that the preferred embodiments are merely illustrative of the invention and not intended to limit the scope of protection of the invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of the present invention.

[0046] according to Figure 1 The topology diagram shown uses voltage and current sensors to collect DC voltage U. dc DC current I of the two rectifiers dcx The voltage V of the two capacitors on the DC side px and V nx The three-phase grid voltage e is used to estimate the three-phase rectifier-side current i of the two rectifiers using a Kalman filter. x2 and the three-phase filter capacitor voltage U cx .

[0047] This invention employs a dual closed-loop control strategy, with identical control strategies for both rectifiers. The outer loop voltage control utilizes adaptive droop control with voltage feedforward. The sampled DC voltage and DC current are used to obtain the grid-side current i through the adaptive droop controller and PI regulator. x1 The d-axis reference value is used to ensure a high power factor for the rectifier, and the grid-side current i... x1 The q-axis reference value is set to 0, and the three-phase current reference value on the rectifier side is obtained through modeling and calculation of the LCL filter. To mitigate the impact of droop control on DC voltage, voltage feedforward is employed to compensate for the DC voltage reduction. The inner-loop current control utilizes fair-sequence model predictive control (FSM). Analysis of the circulating current loop of the T-type three-level LCL rectifier reveals that the main influencing factors of zero-sequence circulating current are the midpoint voltage difference and common-mode voltage difference between the two rectifiers. Based on the common-mode voltage calculation formula, the initial candidate vectors are determined to be seven switching states with zero common-mode voltage. This significantly reduces the computational load of model predictive control, lowers the common-mode voltage difference between the two rectifiers, thus suppressing zero-sequence circulating current, and reduces one control objective of sequential model predictive control, simplifying its structure. FSM uses current tracking and midpoint voltage stability as two control objectives. After fair priority selection, the optimal switching vector is output to the power devices of the T-type three-level rectifier.

[0048] The outer ring employs adaptive droop control, as detailed below:

[0049] DC voltage droop control controls the active power input to the DC network by detecting the difference between the DC voltage and a reference voltage, thereby achieving power balance and voltage stability. This invention employs VI characteristic droop control, which can be expressed as:

[0050]

[0051] In the formula, U dc , These represent the DC voltage of the rectifier and its reference value, respectively. dc , K represents the DC current of the rectifier and its reference value, respectively. d This represents the droop coefficient.

[0052] The droop coefficient K d The droop factor is changed to an adaptive one, typically setting the system operating reference point to: when the transmission power is 0, the DC bus voltage is the system rated voltage; that is, setting... The rated voltage is used. After adding an adaptive droop factor, the VI characteristic droop control is changed to:

[0053]

[0054] In the formula, P N Representing the rated power of the rectifier, the droop factor changes with the DC current, achieving adaptive droop control. When two rectifiers are connected in parallel, the principle of droop control for power distribution is as follows: Figure 2 As shown.

[0055] Due to inherent flaws in the droop control principle, the DC voltage will exhibit a certain degree of droop. To improve this droop characteristic, a feedforward is added based on the difference between the DC voltage and the reference value to compensate for the impact of droop control on the DC voltage.

[0056] The inner loop adopts fair-sequence model predictive control, the details of which are as follows:

[0057] First, based on the topology of two parallel T-type three-level rectifiers, the flow path of the zero-sequence circulating current is analyzed. It can be determined that the influencing factors of the zero-sequence circulating current are the midpoint voltage difference and the common-mode voltage difference between the two rectifiers. The midpoint voltage difference between the two rectifiers is generally 0–10V, while the common-mode voltage difference is related to the DC voltage reference value. When the DC voltage reference value is 500V, the common-mode voltage difference between the two rectifiers can reach hundreds of volts. Clearly, the common-mode voltage difference is the main influencing factor of the zero-sequence circulating current. Based on the relationship between the common-mode voltage difference and the switching state, 7 out of 27 switching states generate a group of switching states with 0 common-mode voltage. This invention uses these 7 switching states as the initial candidate switching states. Thus, there are only 7 initial candidate vectors, greatly reducing the computational load of the system. Furthermore, this method not only achieves the desired effect at the control level without increasing any hardware cost, but also reduces one control objective in sequential model predictive control, simplifying the design of sequential model predictive control.

[0058] After suppressing the zero-sequence circulating current, the control objectives of the inner-loop current control are to balance the midpoint voltage and track the current. To achieve current tracking, a cost function is designed, which is expressed as:

[0059]

[0060] In the formula, i nx2 (n = A, B, C), These represent the rectifier-side current of the xth rectifier and its reference value, respectively.

[0061] Regarding the balance problem of midpoint voltage, analyzing the midpoint voltage yields the following predictive formula for the midpoint voltage difference:

[0062] ΔV1(k+1)=-S(k) T i 12 (k)T s / C 11 +ΔV1(k) Equation 4

[0063] In the formula, ΔV1 represents the midpoint voltage difference of the first rectifier, and C 11 T s i 12 S() represents the upper capacitor on the DC side of the rectifier, the sampling time, the rectifier-side current of the first rectifier, and the switching vector at time k.

[0064] The cost function for the corresponding midpoint voltage is:

[0065] G2=ΔV x (k+2) Equation 5

[0066] To compensate for the computational delay, the cost functions for both midpoint voltage balancing and current tracking are extrapolated by a time interval. In the formula, ΔV x (+2) represents the midpoint voltage deviation after extrapolation of the xth rectifier.

[0067] This invention employs fair model predictive control in the inner loop to select the optimal vector. After determining the cost functions of midpoint voltage balance and current tracking as the two-stage control objectives, traditional sequential model predictive control uses a process of sequentially selecting the optimal vector from the first stage to the second stage. Regardless of whether the number of switching vectors selected from the first to the second stage is fixed or variable based on the cost function value, it cannot avoid the problem of difficulty in determining the priority of the two control objectives due to the sequential structure, thus failing to select the optimal solution. Therefore, a fair sequential model predictive control is proposed to solve the problem of difficulty in determining the priority. The specific details are as follows:

[0068] First, based on the sampled system state and the prediction model, the values ​​of the two cost functions for the initial seven vectors are calculated and the results are saved. Then, the average of the two cost functions is calculated. This average is used to determine the number of vectors that can be exchanged between the two cost functions; vectors with values ​​less than the average are moved to another level, while those with values ​​greater than the average are eliminated. The number of crossover vectors is calculated. Next, the values ​​of the two cost functions are normalized and sorted. Finally, based on the calculation results, the selected vectors are sent to another level. The cost function values ​​of the selected vectors are read from the previously saved results, and the normalized values ​​of the two cost functions for the vectors selected from both levels are added to form a new cost function group. The optimal solution is selected from this cost function group. This selection of the optimal solution is not based on a fixed priority order, but rather on a fair priority order, ensuring the reliability of the optimal solution.

[0069] The control block diagram of the two parallel T-type three-level rectifier system of the present invention is as follows: Figure 3 As shown, the required current and voltage signals are first sampled, and the phase of the grid voltage is obtained using a phase-locked loop (PLL). The DC voltage, DC voltage reference value, and DC current are then fed into an adaptive droop controller with voltage feedforward. Based on the droop control, a grid-side current reference value in phase with the grid voltage is generated. Then, according to the circuit equation of the LCL filter, the grid-side current reference value is converted into a rectifier-side current reference value and fed into a fair sequence model predictive controller. The fair sequence model predictive controller selects the optimal switching vector and sends it to the rectifier based on the sampled signal and the rectifier-side current reference value.

[0070] The control block diagram of the two parallel T-type three-level rectifier system in this embodiment is as follows: Figure 3As shown, in this embodiment, the execution steps of the two rectifiers are the same; here, the main steps of one rectifier are described. The fair sequence model predictive control multi-objective optimization method for parallel T-type three-level rectifiers in this embodiment includes the following steps:

[0071] Step S1: Acquire DC voltage U using voltage and current sensors. dc DC current I of the two rectifiers dcx The voltage V of the two capacitors on the DC side px and V nx Three-phase power grid voltage;

[0072] Step S2: Estimate the three-phase rectifier-side current i of the two rectifiers using a Kalman filter. x2 and the three-phase filter capacitor voltage U cx ;

[0073] Step S3: Obtain the phase information of the grid voltage e through a phase-locked loop. The adaptive droop control with voltage feedforward calculates the reference value of the grid-side current based on the set DC voltage reference value, the DC current of each of the two rectifiers, the rated power of each of the two rectifiers, the DC voltage, and the initial droop coefficient. The reference value of the rectifier-side current is then derived from the circuit relationship of the LCL-type grid-connected inverter.

[0074] Step S4: The reference value of the rectifier side current is delayed by one step using the Lagrange extrapolation method and then sent to the fair sequence model predictive controller for cost function tracking.

[0075] Step S5: Based on the relationship between common-mode voltage and switching state in Table 1, select the seven switching states with a common-mode voltage of 0 as the initial candidate vectors;

[0076] Table 1 Relationship between Switching State and Common Mode Voltage

[0077]

[0078] Step S6: Based on the collected rectifier side current, midpoint voltage difference, LCL capacitor voltage, DC voltage, and rectifier side reference current after Lagrange extrapolation delay, calculate the cost functions G1 and G2 values ​​for each of the seven switching states, and store the calculation results and the corresponding switching states.

[0079] Step S7: Calculate the average cost function of the seven switch states. Switch states with a cost function less than the average value are to be cross-calculated. The expression for the number of cross-calculated switch states is as follows:

[0080]

[0081] In the formula, L and N represent the number of vectors less than the average value in the cost functions G1 and G2, respectively, and are also the number of cross vectors.

[0082] Step S8: Normalize and sort the seven stored switch states and their cost function values. After sorting, each normalized value still has a one-to-one correspondence with the switch state.

[0083] Step S9: The portion of the cost function G1 that is less than the average value Combination 1, the portion where the cost function G2 is less than the average. Combination 2 is formed by selecting the cross vectors based on the average value and extracting the cost function of the cross portion from another cost function. and Add their values ​​to form a new combination Y. In combination Y, both cost function values ​​are normalized, indicating that they have the same priority. The minimum cost function value in combination Y is the optimal solution S. opt (k), the expression for selecting the optimal solution is:

[0084] S opt (k)=argmin s(k) {Y(S i Equation 7

[0085] In the formula, S opt (k) represents the optimal switching vector calculated at time k, n2 represents the sum of the number of cross vectors, and Y represents the new cost function after combination.

[0086] The optimal solution selected does not use a fixed priority; it can be either cost function G1 priority or cost function G2 priority, which ensures the reliability of the optimal solution selection.

[0087] Step S10: Place S opt (k) is converted into a switching vector, which is then converted into the on / off signal of the switching device of the T-type three-level grid-connected inverter and sent in at the next sampling time. Step S1 is executed again at the next sampling time.

[0088] Effect Experiment Example

[0089] To verify the effectiveness of the proposed fair-sequence model predictive control and the multi-objective optimization of the parallel T-type three-level rectifier system, two units employing... Figure 3 Taking a parallel T-type three-level rectifier system as an example, the grid-side current and AC-side power after two rectifiers of different power are connected in parallel are as follows: Figure 4As shown. When grid imbalance occurs, the circulating current suppression effect and the midpoint voltage balance effect of the proposed method compared with the two existing control methods are as follows: Figure 5 As shown. When the DC-side voltage reference value and load power change, the DC voltage stabilization effect and AC-side power distribution are as follows. Figure 6 As shown in Table 2, the parameters of the two parallel T-type three-level rectifiers and their control circuit are shown in Table 2.

[0090] Table 2. Parameters of the two parallel T-type three-level rectifiers and their control circuitry.

[0091]

[0092] from Figure 4 It can be seen that the two rectifiers with rated power of 5kW and 10kW respectively achieved good power distribution, verifying the effectiveness of adaptive droop control with voltage feedforward in power distribution. Figure 5 It can be seen that under the condition of grid voltage imbalance, the three control methods all exhibit good circulating current suppression effects because they use seven switching states that reduce the common-mode voltage to 0 as initial candidate vectors. Specifically, the circulating current of the fair-sequence model predictive control is 19.54 mA, the circulating current of the tolerance-based sequence model predictive control is 65.38 mA, and the circulating current of the zero common-mode voltage model predictive control is 74.91 mA. The fair-sequence model predictive control shows a significant advantage in circulating current suppression compared to the other two existing methods. Regarding midpoint voltage balance, the midpoint voltage difference of the fair-sequence model predictive control is 1.2V, the midpoint voltage difference of the tolerance-based sequence model predictive control is 4.5V, and the midpoint voltage difference of the zero common-mode voltage model predictive control is 4.4V. The fair-sequence model predictive control also demonstrates a significant advantage in midpoint voltage balance, proving the effectiveness of the non-fixed priority order of the fair-sequence model predictive control in selecting the optimal solution. Figure 6The diagram shows the DC voltage recovery effect and AC power distribution when the DC voltage reference value changes. When the DC voltage reference value is 500V, the rated power of the rectifiers is 3472W and 6944W respectively. When the DC voltage reference value changes to 600V, the rated power changes to 5kW and 10kW. The AC power before and after the reference value change meets the power distribution target of the two rectifiers with different power ratings, proving the effectiveness of adaptive droop control with voltage feedforward in power distribution. The DC voltage recovery diagram before and after the reference value change shows that when the DC voltage reference value increases, the difference between the DC voltage and the reference value becomes -100V for all three control methods. Then, within a short time, the DC voltage recovers to the reference value of 600V, and the difference between the DC voltage and the reference value shrinks to near 0V. All three control methods achieve DC voltage recovery, proving that adaptive droop control with voltage feedforward solves the voltage droop problem of droop control and has a good control effect in DC voltage stability.

[0093] The above embodiments are merely illustrative of the technical concept and features of the present invention, intended to enable those skilled in the art to understand the content of the present invention and implement it accordingly. They should not be construed as limiting the scope of protection of the present invention. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein. All equivalent changes or modifications made according to the spirit and essence of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A fair sequence model predictive control multi-objective optimization method for parallel T-type three-level rectifiers, characterized in that, Includes the following steps: Step S1: Acquire DC voltage U using voltage and current sensors. dc DC current I of the two rectifiers dcx The voltage V of the two capacitors on the DC side px and V nx Three-phase power grid voltage; Step S2: Estimate the three-phase rectifier-side current i of the two rectifiers using a Kalman filter. x2 and the three-phase filter capacitor voltage U cx ; Step S3: Obtain the phase information of the grid voltage e through a phase-locked loop. The adaptive droop control with voltage feedforward calculates the reference value of the grid-side current based on the set DC voltage reference value, the DC current of each of the two rectifiers, the rated power of each of the two rectifiers, the DC voltage, and the initial droop coefficient. The reference value of the rectifier-side current is then derived from the circuit relationship of the LCL-type grid-connected inverter. Step S4: The reference value of the rectifier side current is delayed by one step using the Lagrange extrapolation method and then sent to the fair sequence model predictive controller for cost function tracking. Step S5: Based on the relationship between common-mode voltage and switching state, select the seven switching states with a common-mode voltage of 0 as the initial candidate vectors; Step S6: Based on the collected rectifier side current, midpoint voltage difference, LCL capacitor voltage, DC voltage, and rectifier side reference current after Lagrange extrapolation delay, calculate the cost functions G1 and G2 values ​​for each of the seven switching states, and store the calculation results and the corresponding switching states. Step S7: Calculate the average cost function of the seven switch states. Switch states with a cost function less than the average value are those to be cross-calculated. Step S8: Normalize and sort the seven stored switch states and their cost function values. After sorting, each normalized value still has a one-to-one correspondence with the switch state. Step S9: The portion of the cost function G1 that is less than the average value Combination 1, the portion where the cost function G2 is less than the average. Combination 2 is formed by selecting the cross vectors based on the average value and extracting the cost function of the cross portion from another cost function. and The values ​​of these combinations are added together to form a new combination Y. The minimum cost function in combination Y is the optimal solution S. opt (k); Step S10: Place S opt (k) is converted into a switching vector, which is then converted into the on / off signal of the switching device of the T-type three-level grid-connected inverter and sent in at the next sampling time. Step S1 is executed again at the next sampling time.

2. The fair sequence model predictive control multi-objective optimization method for parallel T-type three-level rectifiers according to claim 1, characterized in that, The reference value for DC voltage in step S3 is 500V.

3. The fair sequence model predictive control multi-objective optimization method for parallel T-type three-level rectifiers according to claim 1, characterized in that, The cost function described in step S4 is: In the formula, i nx2 (n = A, B, C), These represent the rectifier-side current of the xth rectifier and its reference value, respectively.

4. The fair sequence model predictive control multi-objective optimization method for parallel T-type three-level rectifiers according to claim 1, characterized in that, The expression for the number of cross switch states in step S7 is: In the formula, L and N represent the number of vectors less than the average value in the cost functions G1 and G2, respectively, and are also the number of cross vectors.

5. The fair sequence model predictive control multi-objective optimization method for parallel T-type three-level rectifiers according to claim 1, characterized in that, In step S9, both cost function values ​​in the combination Y are normalized, indicating that they have the same priority.

6. The fair sequence model predictive control multi-objective optimization method for parallel T-type three-level rectifiers according to claim 1, characterized in that, The expression for the optimal solution mentioned in step S9 is: In the formula, S opt (k) represents the optimal switching vector calculated at time k, n2 represents the sum of the number of cross vectors, and Y represents the new cost function after combination.

7. The fair sequence model predictive control multi-objective optimization method for parallel T-type three-level rectifiers according to claim 6, characterized in that, The optimal solution mentioned does not use a fixed priority; it prioritizes either the cost function G1 or the cost function G2.

Images