Grid-connected MMC optimization model predictive control method and system for offshore wind power

Abstract

The present invention relates to the technical field of offshore wind power flexible direct current transmission, and specifically discloses a grid-connected MMC optimization model prediction control method and system for offshore wind power, the method comprising: establishing a relationship between the total number N of half-bridge submodules of each phase bridge arm and the number combination of candidate submodules, and determining the number combination of candidate submodules of each phase bridge arm; calculating the output current tracking deviation based on the output current of the next cycle, and determining the number combination of submodules to be tried for each phase bridge arm according to the output current tracking deviation; selecting the first submodule number combination that minimizes the first cost function; calculating the deviation of circulating current tracking based on the circulating current prediction value of the next cycle, and determining the number combination of submodules to be tried for each phase bridge arm according to the deviation of circulating current tracking; and selecting the second submodule number combination that minimizes the second cost function. The present invention realizes the flexible and variable number combination of candidate submodules, and the flexible switching of the number combination of tried submodules with the control target deviation.

Description

Technical Field

[0001] The present invention relates to the technical field of offshore wind power flexible direct current transmission, and in particular to a grid-connected MMC optimization model predictive control method and system for offshore wind power. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] Offshore wind power is currently showing a trend of large-scale wind turbines (above 10MW), large-scale wind farms (above 1GW), and development in deep sea (above 100km). However, the harsh offshore environment leads to poor power quality, and the bidding for offshore wind power on the grid also has stricter requirements on the cost per kilowatt-hour, thus placing higher performance requirements on the entire power conversion system.

[0004] Modular multilevel converter (MMC) has become the preferred topology type for (multi-terminal) flexible direct current transmission converter stations due to its modular design, strong scalability of power and voltage levels, and fault tolerance. In (multi-terminal) flexible direct current transmission systems, MMC is responsible for controlling DC voltage and transmission power. Research on advanced control strategies for MMC on the grid side is crucial to achieving low cost, high energy transmission efficiency, high grid-connected power quality, strong robustness and stability of the entire offshore wind power system.

[0005] The existing predictive control method for MMC is mainly based on the finite set model predictive control (FCS-MPC) method of the optimal candidate level. This method takes the number of submodules in each bridge arm as the control variable, and selects the number of submodules that make the control variable optimal by traversing all candidate submodule arrays and merging them to minimize the cost function. Then, the submodule capacitor voltage sorting algorithm is used to achieve capacitor voltage balancing, and the switch state of each submodule is output. For an MMC with N submodules in each bridge arm, this method only needs to traverse N + 1 possible combinations of submodule input numbers for each phase, and the amount of calculation increases linearly with the increase of the number of submodules, and the amount of calculation is relatively small. Nevertheless, in HVDC applications with hundreds of modules, the amount of calculation of this method is still large and difficult to apply in practice. The existing simplified method is to only traverse three submodule number combinations near the optimal number of submodules in the previous control cycle in each control cycle, so the number of traversals for each phase is reduced from N + 1 to 3, but this simplified scheme will affect the transient performance of the MMC.

[0006] In the transient process of the MMC system, when the reference value of the control variable changes significantly (such as a step change in the power reference command or a power flow reversal), due to the small number of candidate combinations in each cycle, the optimal sub-module input number combination that can minimize the control variable tracking error may be far away from the optimal input number in the previous cycle. Therefore, the controller may need more control steps to find the optimal sub-module input number combination, and the control variable can only approach the reference value more slowly, which greatly increases the transient switching time and loses the inherent dynamic performance advantages of FCS-MPC.

[0007] The existing grid-connected MMC adaptive fast optimization model predictive control algorithm has 5 candidate sub-module combinations in the transient process and 3 candidate sub-module combinations in the steady state process. The candidate sub-module combinations are fixed, resulting in a small number of candidate sub-module combinations compared to all sub-module combinations when the number of sub-modules is large. More control steps are still required to approach the optimal sub-module input number combination. The dynamic performance is poor, and the transient and steady states are only distinguished by the deviation of the control target, which makes the sub-module combinations tried in different situations not flexible enough and have weak adaptability. Summary of the invention

[0008] In order to solve the above problems, the present invention proposes a grid-connected MMC optimization model predictive control method and system for offshore wind power. By associating the number of candidate sub-modules with the total number of sub-modules and associating the combination of the number of attempted sub-modules with the deviation of the control target, the variability of the number of candidate sub-modules and the flexible switching of the combination of the number of attempted sub-modules with the deviation of the control target are realized, thereby balancing the computational burden and dynamic performance.

[0009] In some embodiments, the following technical solutions are adopted:

[0010] A grid-connected MMC optimization model predictive control method for offshore wind power, comprising:

[0011] Establishing a relationship between the total number N of half-bridge submodules in each phase bridge arm and the number of candidate submodule combinations, and determining the number of candidate submodule combinations in each phase bridge arm according to the total number of half-bridge submodules in each phase bridge arm;

[0012] Select three candidate submodule number combinations of the current cycle and bring them into the current prediction equation to obtain the output current of the system in the next cycle; calculate the output current tracking deviation based on the output current of the next cycle, and determine the submodule number combination to be tried in each phase bridge arm according to the output current tracking deviation; select the first submodule number combination that minimizes the first cost function;

[0013] Select three candidate sub-module number combinations for the current cycle and substitute them into the circulation prediction equation to obtain the circulation prediction value of the system in the next cycle; calculate the deviation of the circulation tracking based on the circulation prediction value of the next cycle, and determine the sub-module number combination to be tried in each phase bridge arm according to the deviation of the circulation tracking; select the second sub-module number combination that minimizes the second cost function.

[0014] In other embodiments, the following technical solutions are adopted:

[0015] A grid-connected MMC optimization model predictive control system for offshore wind power, comprising:

[0016] A candidate submodule determination module is used to establish a relationship between the total number N of half-bridge submodules in each phase bridge arm and the number of candidate submodule combinations, and determine the number of candidate submodule combinations in each phase bridge arm according to the total number of half-bridge submodules in each phase bridge arm;

[0017] The prediction current control module is used to select three candidate submodule number combinations in the current cycle and bring them into the current prediction equation to obtain the output current of the system in the next cycle; calculate the output current tracking deviation based on the output current of the next cycle, and determine the submodule number combination to be tried in each phase bridge arm according to the output current tracking deviation; select the first submodule number combination that minimizes the first cost function;

[0018] The predicted circulation control module is used to select three candidate sub-module combinations in the current cycle and substitute them into the circulation prediction equation to obtain the circulation prediction value of the system in the next cycle; calculate the deviation of the circulation tracking based on the circulation prediction value of the next cycle, and determine the sub-module combination to be tried in each phase bridge arm according to the deviation of the circulation tracking; select the second sub-module combination that minimizes the second cost function.

[0019] Compared with the prior art, the present invention has the following beneficial effects:

[0020] (1) The present invention associates the candidate sub-module number combination with the total number of sub-modules and associates the sub-module number combination tried in each control cycle with the deviation amount, thereby achieving flexible variability of the candidate sub-module number combination and flexible switching of the tried sub-module number combination with the control target deviation amount, thereby balancing the relationship between computational burden and dynamic performance.

[0021] (2) The present invention can automatically adjust the number of candidate combinations according to the number of MMC sub-modules, and can effectively improve the dynamic performance of the system when the number of sub-modules is large. In addition, the present invention can automatically adjust the number of attempted combinations according to the deviation of the control target, and ensure a small computational burden when the deviation is small. When the deviation is large, the number of attempted combinations is appropriately increased, thereby exchanging a small increase in computational burden for a significant improvement in dynamic performance.

[0022] Other features and advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a grid-connected MMC topology diagram in the prior art;

[0024] Figure 2 It is a control flow chart of the MMC adaptive fast optimization model prediction in the prior art;

[0025] Figure 3 The present invention is a flow chart of a grid-connected MMC optimization model predictive control method for offshore wind power in an embodiment of the present invention. DETAILED DESCRIPTION

[0026] It should be noted that the following detailed descriptions are illustrative and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used in the present invention have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.

[0027] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0028] Embodiment 1

[0029] Existing grid-connected MMC topologies are as follows: Figure 1 As shown, each phase It consists of two bridge arms (subscripted as Represents the upper bridge arm, the subscript is Each phase bridge arm consists of N half-bridge sub-modules (hereinafter referred to as SM) and a bridge arm reactor. The half-bridge sub-module SM consists of two IGBTs and a voltage Capacitor composition.

[0030] Assuming that the capacitor voltages of each SM in the bridge arm are balanced, the output voltage of the bridge arm ( , ) is completely determined by the number of half-bridge sub-modules SM put into use ( , ). The AC output side of the grid-connected MMC is connected to the AC grid through the RL filter. and are the inductance and resistance of the filter respectively, and are the grid current and voltage respectively. The dynamic equations of the AC and DC equivalent circuits are:

[0031] (1)

[0032] (2)

[0033] In the formula, is the equivalent inductance on the AC output side; It is the circulating current flowing in the x phase; it consists of two parts: DC component and AC component. It represents the AC component, which is an internal circulation current that flows only between the three phases and is mainly composed of double fundamental frequency components; is the bridge arm inductance, is a direct current.

[0034] Assume that the capacitor voltage of SM is balanced. From equations (1) and (2), we can see that the relationship model between the grid current and circulating current and the SM input number is as follows:

[0035] (3)

[0036] (4)

[0037] Where k represents the kth sampling period, , , , and is the average capacitor voltage of the x-phase SMs. SM stands for (Sub modular), and SMs stands for the plural form of submodules; represents the AC grid current output by phase x in the k+1th cycle; represents the circulation of phase x in the k+1th period; Indicates the DC side voltage.

[0038] Figure 2 The existing adaptive fast optimization model predictive control algorithm process is given. The adaptive fast optimization model predictive control algorithm automatically adjusts the candidate SM input number combination according to the system operation mode.

[0039] First, the threshold is set, and then the threshold is used to determine whether the system is in a transient state or a steady state, and the number of candidate SM input combinations is adaptively adjusted. In the transient process, the number of candidate submodule combinations is 5, that is, the current number of SM input combinations 3 nearby combinations and and (0.5 , 0.5 ); In the steady-state process, the number of candidate submodules is 3, which are and . Filter out the SM input number combination with the minimum cost function in the k+1 cycle ).

[0040] Then, the circulation control is predicted. The control principle is the same as above. First, the state is judged according to the threshold. In the transient process, the number of candidate submodules is 5, that is, ), ), ), )and ); In the steady-state process, there are only 3 candidate input combinations ), )and ) to minimize the circulation cost function. In the transient process, two new candidate input combinations +2) and ) is added to the set of candidate input number combinations, and the system uses five candidate input number combinations to minimize the circulation cost function.

[0041] in, ) represents the SM input number combination with the minimum cost function in the k+1 cycle selected by the current tracking link, and the superscript x1 is used to distinguish it from the candidate combination.

[0042] When there are a large number of grid-connected MMC sub-modules, the above method has only 5 candidate combinations in the transient process. When the reference current changes greatly, multiple control cycles are required to achieve the optimal control effect, which affects the transient performance of the grid-connected MMC. In addition, the threshold is set to distinguish the transient steady state, but the threshold selection is difficult and the control flexibility is too low, which can easily lead to poor control effect.

[0043] Based on this, this embodiment discloses a grid-connected MMC optimization model predictive control method for offshore wind power, which associates the candidate sub-module number combination with the total number of sub-modules to achieve flexible variability of the candidate sub-module number combination, and associates the sub-module number combination attempted in each control cycle with the deviation amount to achieve flexible switching of the attempted sub-module number combination with the control target deviation amount.

[0044] Combination Figure 3 The specific process of the control method of this embodiment is as follows:

[0045] (1) Establish a relationship between the total number N of half-bridge sub-modules in each phase bridge arm and the number of candidate sub-module combinations, and determine the number of candidate sub-module combinations in each phase bridge arm according to the total number of half-bridge sub-modules in each phase bridge arm.

[0046] The relationship between the total number N of half-bridge sub-modules per phase bridge arm and the number of candidate sub-modules is obtained by fitting the simulation results. The data obtained by simulation (changing the total number N of sub-modules to see how much λ is, the control effect is better, so as to obtain data with N as the independent variable and λ as the dependent variable) can be input into matlab for curve fitting, and the expression obtained is as follows:

[0047] (5)

[0048] in,

[0049] ;

[0050] in, ... All are fitting coefficients; represents the set of selectable submodules in the K+1th cycle; Indicates the number of optimal submodules selected in the Kth cycle.

[0051] (2) Select three candidate sub-module combinations and substitute them into the current prediction equation to obtain the output current of the system in the next cycle.

[0052] Specifically, the current SM input number is combined and Substituting the three candidate combinations into formula (3), we can get the output current of the system in the next cycle: ,Right now Figure 3 A vector with a direction in . (The delay compensation strategy is an existing technology and will not be described in detail).

[0053] (3) Based on the output current of the next cycle Calculate the output current tracking deviation, according to the output current tracking deviation , determine the number of sub-modules to be tried in each phase bridge arm; and select the first sub-module number combination that minimizes the first cost function.

[0054] In this embodiment, the output current tracking deviation is the output current of the next cycle Reference value of grid-connected current The absolute value of the difference, that is:

[0055] (6)

[0056] Among them, the grid-connected current reference value The calculation process is as follows:

[0057] Get the DC voltage measurement value and convert the DC voltage measurement value V dc Compare with the reference value to get the active power reference , the grid-connected current is calculated through active power reference, reactive power reference, and grid voltage Reference; the specific calculation formula is as follows:

[0058] (7)

[0059] (8)

[0060] in, is the three-phase grid voltage value in the kth cycle, , is the grid voltage value in the αβ coordinate system within the k+1 period, , are respectively the active power reference value and the reactive power reference value within the k cycle; , They are the grid-connected current reference values ​​in the αβ coordinate system within the k cycle respectively; finally, the three-phase grid-connected current reference values ​​are obtained by using the Clarke inverse transformation .

[0061] Tracking deviation according to output current , determine the number of submodule combinations to try, specifically:

[0062] when , ;

[0063] when , ;

[0064] when , .

[0065] Among them, the interval ( , ) is divided into segment, along with Increase the number of options to try; specifically:

[0066] when , the selection is empty;

[0067] when When the options are [ ];

[0068] when When the options are [ ]、[ ]、[ ];

[0069] And so on;

[0070] when When the options are [ ]、[ ]、[ ], ..., [ ]、[ ].

[0071] As the deviation increases, the number of options increases; we traverse the number of submodule combinations we try and select the one that makes the first cost function The smallest combination of the first submodule number .

[0072] in, and The sum is N, so Can be directly based on to be sure.

[0073] (4) Select three candidate combinations and substitute them into the circulation prediction equation to obtain the circulation prediction value of the system in the next cycle.

[0074] In this embodiment, the current SM input number is combined , and Substituting the three candidate combinations into formula (4), we can obtain the circulation of the system in the next cycle: (also after delay compensation).

[0075] From formula (4), we can see that the parameters required for circulation model prediction are: , and ,in, represents the circulation of phase x in the kth period, It is half of the sum of the upper and lower bridge arm currents, that is, through the input , you can get . , are the average capacitor voltages of x-phase SMs, namely Figure 3 In ;therefore, Figure 3 The input parameters in are and , we can obtain .

[0076] (5) Calculating the deviation of circulating current tracking based on the predicted circulating current value of the next cycle, determining the number of sub-modules to be tried in each phase bridge arm according to the deviation of circulating current tracking; and selecting the second sub-module number combination that minimizes the second cost function.

[0077] The deviation of circulation tracking is calculated based on the circulation prediction value of the next cycle, specifically:

[0078] (9)

[0079] in, Indicates the reference value of the circulating current. The circulating current contains DC and AC components, of which the DC component must exist, while the AC component needs to be suppressed, so the circulating current reference can be set to the DC component P* / 3Vdc; Vdc is the DC side voltage.

[0080] The number of submodules to be tried in each phase bridge arm is determined according to the deviation of circulating current tracking, specifically:

[0081] when hour, ;

[0082] when hour, ;

[0083] when hour, .

[0084] According to the determined combination of submodule numbers to be attempted, select the second cost function The smallest combination of the second submodule number .

[0085] (6) Based on the obtained first submodule number combination and second submodule number combination, a submodule capacitor voltage sorting algorithm is used to realize capacitor voltage balancing, and the switch state of each submodule is output.

[0086] The capacitor voltage sorting algorithm refers to the method of sorting the capacitor voltage after knowing the input number of each bridge arm submodule. Under the premise of Sort in ascending or descending order. Then sort by current The submodules with the front or back direction selection order are put into operation first, so that the submodules with high capacitor voltage are discharged first, and the submodules with low voltage are charged first, so as to achieve voltage balance; this method is a prior art and will not be described in detail.

[0087] This embodiment achieves flexible variability of candidate submodule combinations and flexible switching of attempted submodule combinations with control target deviation by associating candidate submodule combinations with the total number of submodules and associating attempted submodule combinations with deviation in each control cycle, thereby balancing the relationship between computational burden and dynamic performance. When the total number of submodules is small, the number of candidate submodule combinations is reduced, which can reduce the computational burden; when the total number of submodules is large, the number of candidate submodule combinations is increased, making the selection more flexible; when the control target deviation is large, the number of attempted submodule combinations in each control cycle is large, quickly approaching the optimum, and improving dynamic performance; when the control target deviation is small, the number of attempted submodule combinations in each control cycle is small, which ensures performance while reducing computational burden.

[0088] Embodiment 2

[0089] In one or more embodiments, a grid-connected MMC optimization model predictive control system for offshore wind power is disclosed, including:

[0090] A candidate submodule determination module is used to establish a relationship between the total number N of half-bridge submodules in each phase bridge arm and the number of candidate submodule combinations, and determine the number of candidate submodule combinations in each phase bridge arm according to the total number of half-bridge submodules in each phase bridge arm;

[0091] The prediction current control module is used to select three candidate submodule number combinations in the current cycle and bring them into the current prediction equation to obtain the output current of the system in the next cycle; calculate the output current tracking deviation based on the output current of the next cycle, and determine the submodule number combination to be tried in each phase bridge arm according to the output current tracking deviation; select the first submodule number combination that minimizes the first cost function;

[0092] The predicted circulation control module is used to select three candidate sub-module combinations in the current cycle and substitute them into the circulation prediction equation to obtain the circulation prediction value of the system in the next cycle; calculate the deviation of the circulation tracking based on the circulation prediction value of the next cycle, and determine the sub-module combination to be tried in each phase bridge arm according to the deviation of the circulation tracking; select the second sub-module combination that minimizes the second cost function.

[0093] The specific implementation of the above modules is the same as that in Example 1 and will not be described in detail.

[0094] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. A grid-connected MMC optimization model predictive control method for offshore wind power, characterized in that: include: Establishing a relationship between the total number N of half-bridge submodules in each phase bridge arm and the number of candidate submodule combinations, and determining the number of candidate submodule combinations in each phase bridge arm according to the total number of half-bridge submodules in each phase bridge arm; The relationship between the total number N of half-bridge submodules per phase bridge arm and the number of candidate submodule combinations is established, specifically: ； ； in, ... All are fitting coefficients; represents the set of selectable submodules in the K+1th cycle; Indicates the number of optimal submodules selected in the Kth cycle; Select three candidate submodule number combinations of the current cycle and bring them into the current prediction equation to obtain the output current of the system in the next cycle; calculate the output current tracking deviation based on the output current of the next cycle, and determine the submodule number combination to be tried in each phase bridge arm according to the output current tracking deviation; select the first submodule number combination that minimizes the first cost function; The current prediction equation is: ； Where k represents the kth sampling period, , , and is the average capacitor voltage of x-phase SMs, SM represents submodule, SMs represents the complex form of submodule; represents the AC grid current output by phase x in the k+1th cycle; and is the number of lower bridge arms and upper bridge arms of the submodule SM; and are the grid current and voltage in the Kth cycle respectively; The first cost function is: ； in, is the output current of the next cycle, is the grid-connected current reference value; Select three candidate submodule number combinations of the current cycle and substitute them into the circulation prediction equation to obtain the circulation prediction value of the system in the next cycle; calculate the deviation of the circulation tracking based on the circulation prediction value of the next cycle, and determine the submodule number combination to be tried for each phase bridge arm according to the deviation of the circulation tracking; select the second submodule number combination that minimizes the second cost function; The circulation prediction equation is: ； Where k represents the kth sampling period, , and is the average capacitor voltage of x-phase SMs; represents the circulation of phase x in the k+1th period; represents the circulation of phase x in the kth period, , are the average capacitor voltages of x-phase SMs, respectively; and is the number of lower bridge arms and upper bridge arms of the submodule SM; Indicates the DC side voltage; The second cost function is specifically: ； in, The circulation of the next cycle, Indicates the reference value of the circulation; The method further includes: based on the obtained first submodule number combination and second submodule number combination, utilizing a submodule capacitor voltage sorting algorithm to realize capacitor voltage balancing, and outputting a switch state of each submodule.

2. A grid-connected MMC optimization model predictive control method for offshore wind power according to claim 1, characterized in that: The output current tracking deviation is calculated based on the output current of the next cycle, specifically: The absolute value of the difference between the output current of the next cycle and the grid-connected current reference value is the output current tracking deviation; wherein the grid-connected current reference value is calculated based on the active power reference, the reactive power reference and the grid voltage.

3. The MMC optimization model prediction control method for offshore wind power grid connection according to claim 1, characterized in that: Tracking deviation according to output current , determine the number of submodules to be tried in each phase bridge arm, specifically: when , ; when , ; when , ; Among them, the interval ( , ) is divided into segment, along with Increase the number of options to try.

4. The MMC optimization model prediction control method for offshore wind power grid connection according to claim 1, characterized in that: The deviation of circulation tracking is calculated based on the circulation prediction value of the next cycle, specifically: in, represents the reference value of the circulation, represents the circulation prediction value of the next cycle, Indicates the deviation of circulation tracking.

5. The method for predictive control of an offshore wind power grid-connected MMC optimization model according to claim 1, characterized in that: Deviation according to circulation tracking Determine the number of submodule combinations to try, specifically: when , ; when , ; when , ; Indicates the number of optimal sub-modules of the half-bridge arm selected in the Kth cycle.

6. A grid-connected MMC optimization model predictive control system for offshore wind power, characterized in that: include: A candidate submodule determination module is used to establish a relationship between the total number N of half-bridge submodules in each phase bridge arm and the number of candidate submodule combinations, and determine the number of candidate submodule combinations in each phase bridge arm according to the total number of half-bridge submodules in each phase bridge arm; The relationship between the total number N of half-bridge submodules per phase bridge arm and the number of candidate submodule combinations is established, specifically: ； ； in, ... All are fitting coefficients; represents the set of selectable submodules in the K+1th cycle; Indicates the number of optimal submodules selected in the Kth cycle; The prediction current control module is used to select three candidate submodule number combinations in the current cycle and bring them into the current prediction equation to obtain the output current of the system in the next cycle; calculate the output current tracking deviation based on the output current of the next cycle, and determine the submodule number combination to be tried in each phase bridge arm according to the output current tracking deviation; select the first submodule number combination that minimizes the first cost function; The current prediction equation is: ； Where k represents the kth sampling period, , , and is the average capacitor voltage of x-phase SMs, SM represents submodule, SMs represents the complex form of submodule; represents the AC grid current output by phase x in the k+1th cycle; and is the number of lower bridge arms and upper bridge arms of the submodule SM; and are the grid current and voltage in the Kth cycle respectively; The first cost function is: ； in, is the output current of the next cycle, is the grid-connected current reference value; The predicted circulation control module is used to select three candidate submodule number combinations in the current cycle and substitute them into the circulation prediction equation to obtain the circulation prediction value of the system in the next cycle; calculate the deviation of the circulation tracking based on the circulation prediction value of the next cycle, and determine the submodule number combination to be tried in each phase bridge arm according to the deviation of the circulation tracking; select the second submodule number combination that minimizes the second cost function; The circulation prediction equation is: ； Where k represents the kth sampling period, , and is the average capacitor voltage of x-phase SMs; represents the circulation of phase x in the k+1th period; represents the circulation of phase x in the kth period, , are the average capacitor voltages of x-phase SMs, respectively; and is the number of lower bridge arms and upper bridge arms of the submodule SM; Indicates the DC side voltage; The second cost function is specifically: ； in, The circulation of the next cycle, Indicates the reference value of the circulation; The method further includes: based on the obtained first submodule number combination and second submodule number combination, utilizing a submodule capacitor voltage sorting algorithm to realize capacitor voltage balancing, and outputting a switch state of each submodule.

Images