Multi-target power loss optimization control method for modular multilevel converter

Abstract

The invention discloses a modular multi-level converter multi-target power loss optimization control method, and relates to the technical field of multi-level power electronic converters, and the method comprises the steps: carrying out the construction of a multi-target mantis search algorithm model through hunting strategy simulation, Pareto control and crowding distance, and building a power loss optimization model of an MMC; a multi-target mantis search algorithm MOMSA is adopted to optimize the optimal values of the injected secondary circulation and the third harmonic voltage, and the optimal injection amplitude, the optimal phase angle and the optimal power loss of the MMC under the given condition are given; injecting the optimal secondary circulating current and the third harmonic voltage into the MMC system to carry out multi-target power loss optimization control; compared with other loss optimization methods, additional hardware burden does not need to be added, and meanwhile, compared with an existing method of only considering total power consumption optimization of the sub-modules or single-target loss optimization of the power device with the highest loss in the sub-modules, collaborative optimization of multi-target power loss optimization is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of multilevel power electronic converters, and particularly relates to a multi-objective power loss optimization control method, device and medium for a modular multilevel converter. Background Art

[0002] Modular multilevel converters (MMCs) have been widely used in power grids due to their advantages such as high efficiency and easy expansion; the total power loss of sub-modules and the semiconductor device with the highest loss in the sub-module have an important impact on the operating cost and service life of the MMC; the loss optimization method based on circulating current control has become an important power loss optimization technology because it has the advantages of not requiring additional hardware costs and not deteriorating the output current harmonic performance.

[0003] The MMC consists of multiple sub-modules (SMs), and each sub-module contains multiple power devices, which will result in considerable power loss; the cost generated by these power losses is an important part of the operating expenses, and a small reduction in power loss will also lead to a significant decrease in the annual operating cost of the MMC; therefore, reducing power loss is crucial for improving the efficiency of the MMC, expanding the operating capacity of the MMC, and enhancing the economic operation of the MMC; in addition to efficiency, the operating reliability of the MMC is also a key indicator because the maximum loss of the power devices in the SM affects its reliability; therefore, reducing the total power loss of the SM and the highest loss of the semiconductor devices in the SM is crucial for increasing the service life of the MMC system and prolonging the service life of the MMC.

[0004] In existing research, the loss optimization research on the MMC mainly includes two aspects: the optimization of the total loss of the SM and the optimization of the power device loss in the SM. Among them, the optimization of the total loss will reduce the total loss in the SM, and the optimization of the power device loss in the SM will reduce the highest loss in the sub-module; the current circulating current control methods mainly involve sampling and coordinate transformation. By sampling the output currents of each parallel inverter, calculating the d-axis and q-axis components of the circulating current, and adding them as disturbances to the voltage-current loop control of the droop control, a PWM drive signal is generated to drive the on-off of the inverter bridge arm switch tubes to control the inverter output; however, such circulating current control methods approximate the power loss model and transform the loss optimization problem into a current optimization problem. Since it involves the problem of voltage change, and the power loss is the product of current and voltage, there is a certain error when obtaining the optimal solution of the power loss optimization, and it is difficult to simultaneously minimize the total loss and the loss of the highest loss device in the sub-module in the MMC system, so it is difficult to realize the collaborative optimization of the total loss of the MMC system and the device with the highest loss in the sub-module. Summary of the Invention

[0005] The embodiment of the present invention provides a multi-objective power loss optimization control method for a modular multilevel converter, which can solve the problem in the prior art that it is difficult to achieve the collaborative optimization of the total loss of the MMC system and the power devices with the highest sub-module loss in the current stage.

[0006] The embodiment of the present invention provides a multi-objective power loss optimization control method for a modular multilevel converter, including the following steps: Utilize the structural information of the modular multilevel converter to construct a power loss optimization model of the modular multilevel converter; Regard different second-order circulating currents and third-order harmonics as different injection schemes of the power loss optimization model, input each injection scheme into the power loss optimization model, and obtain the total power loss of the modular multilevel converter and the loss of the highest-loss device under different injection schemes; Search for the injection scheme that minimizes the total power loss of the modular multilevel converter and the loss of the highest-loss device. During the search, regard different injection schemes as different solutions of the multi-objective mantis search algorithm MOMSA, use the elite non-dominated sorting NDS to perform non-dominated sorting on different solutions, regard the solutions with the loss of devices in the modular multilevel converter and multiple sub-modules less than the preset loss threshold as non-dominated solutions, and calculate the crowding distance of each non-dominated solution using the crowding distance CD. The crowding distance of each non-dominated solution represents the sum of the differences between the current solution and all adjacent solutions, and regard the solution with the largest crowding distance as the searched injection scheme; Control the modular multilevel converter using the searched injection scheme.

[0007] Preferably, the construction of the power loss optimization model includes: According to the number N of sub-modules on the arm of the modular multilevel converter, the arm inductance L, and the structure composed of two power switches, two diodes and a DC capacitor included in the sub-module, construct a power loss optimization model of the three-phase modular multilevel converter MMC system; the power loss optimization model is expressed as: ; Where: Represents the loss of the power switch T a in the sub-module, Represents the loss of the power switch T b in the sub-module, Represents the loss of the diode D a in the sub-module, Represents the loss of the diode D b in the sub-module; Is the number of sub-modules SM in each arm of the modular multilevel converter MMC; P con_Ta 、 Pon_Ta and P off_Ta are the conduction loss, turn-on loss, and turn-off loss of the upper power device of the SM, respectively; P con_Tb 、 P on_Tb and P off_Tb are the conduction loss, turn-on loss, and turn-off loss of the lower power device of the SM, respectively; P con_Da and P rec_Da are the conduction loss and reverse recovery loss of the upper diode of the SM, respectively; P con_Db and P rec_Db are the conduction loss and reverse recovery loss of the lower diode of the SM, respectively.

[0008] Preferably, the method for finding the injection scheme that minimizes the total power loss of the modular multilevel converter and the loss of the device with the highest loss includes: According to the number N of sub-modules and the arm inductance L in the three-phase modular multilevel converter MMC system, and two power switches T a and T b in the sub-module, two diodes D a and D b and a DC capacitor C, generate a set of initial solutions, and each initial solution represents an injection scheme for the secondary circulating current and the third harmonic voltage; For each initial solution, inject it into the power loss optimization model of the three-phase modular multilevel converter MMC system to obtain the power loss of the three-phase modular multilevel converter MMC system under this injection scheme; according to the power loss, define a fitness function value for each solution, and the fitness function is the reciprocal of the total power loss, expressed as Fitness= where P total is the total power loss of the MMC system; Use the elitist non-dominated sorting NDS to perform non-dominated sorting on all initial solutions according to the fitness value to obtain non-dominated solutions, which represent solutions that are not worse than other solutions on multiple objectives; and obtain the crowding distance of the solutions in each non-dominated layer; Select a part of the non-dominated solutions as the parent generation, and according to the position update formula of the multi-objective mantis search algorithm MOMSA, combine the global and local information to update the positions of the parent generation solutions to generate new candidate solutions; the injection scheme of the new candidate solutions is injected into the power loss optimization model again to calculate the new fitness value; Merge the parental solution and the newly generated candidate solution, and conduct fitness evaluation, non-dominated sorting, and crowding distance calculation again; according to the result of non-dominated sorting, select a new population, retain the high-quality non-dominated solutions, and replace the inferior solutions; stop the fitness evaluation and update until the maximum number of iterations or the fitness change is not obvious. Output the set of non-dominated solutions obtained through multiple iterations of optimization. These solutions represent the optimal solution set under the given multi-objectives, and obtain the optimal secondary circulation and third harmonic voltage from the optimal solution set.

[0009] Preferably, the position update of the multi-objective mantis search algorithm MOMSA is based on the objectives and the positions of other individuals, and combines global and local information; The position update formula is expressed as: X (t+1) =X （t） +α·Ftrack+β·Fattack; Where: Ftrack represents the direction of an individual tracking the objective; Fattack represents random perturbation or mutation; α, β represent control parameters; X （t） represents the current position at time t; The global information comes from the optimal solution of the entire population, representing the solution that performs best on all objectives in the current iteration; at the same time, this optimal solution represents the best search direction of the population in the global scope, which can guide other individuals to move towards the global optimal region for global search; The local information comes from the individual itself and the information within its neighborhood, which can help the individual conduct detailed search in the local area, adjust its own search direction and step size to find a better solution within the local scope.

[0010] Preferably, the control modular multilevel converter includes: Obtain the optimal injection amplitude and optimal phase angle value according to the optimal values of the secondary circulation and the third harmonic voltage; according to the optimal injection amplitude and optimal phase angle value, obtain the secondary circulation component in the three-phase modular multilevel converter MMC system through algebraic calculation and a band-pass filter; among them, the algebraic calculation is to calculate the corresponding circulation amplitude and phase angle according to the mathematical relationship between the injected voltage and the circulation; the band-pass filter is used to extract the circulation component of a specific frequency from the calculated circulation signal; Obtain the dq components of the secondary circulation from the secondary circulation component according to the method of coordinate transformation; the coordinate transformation is to convert the three-phase circulation signal to the synchronous rotating dq coordinate system; in the dq coordinate system, the d-axis component corresponds to the excitation component, and the q-axis component corresponds to the torque component, and the circulation is controlled by controlling the dq components respectively; The dq components of the secondary circulation are made to follow their reference values through closed-loop vector control; the closed-loop vector control compares the deviation between the actual dq components of the secondary circulation and the reference values, and uses a proportional-integral-derivative (PID) controller to adjust the control signals of the MMC system, so that the actual dq components of the secondary circulation can track the reference values to achieve the injection of the secondary circulation. Based on the injection of the secondary circulation, the optimal third-harmonic voltage reference signal is added to the arm reference signal to achieve the collaborative optimization of multi-objective power losses of the MMC system to control the modular multilevel converter.

[0011] Preferably, the collaborative optimization control of the multi-objective power losses of the MMC system includes: The injection signal of the secondary circulation can be expressed as: i 2nd ( t ) = I 2nd *cos( ωt + ϕ 2nd ); Where: I 2nd is the amplitude of the secondary circulation, ϕ 2nd is the phase angle of the secondary circulation, ω is the angular frequency; The injection signal of the third-harmonic voltage can be expressed as: v 3rd ( t ) = V 3rd *cos(3 ωt + ϕ 3rd ); Where: V 3rd is the amplitude of the third-harmonic voltage, ϕ 3rd is the phase angle of the third-harmonic voltage; The transfer function of the band-pass filter can be expressed as: H ( s ) = ; Where: is the center frequency, Q is the quality factor; The transformation formula from three-phase to two-phase is expressed as: ; Where: θ is the synchronous rotation angle,ia , ib , ic are three-phase circulating current signals, id and iq are the circulating current components in the dq coordinate system; When in closed-loop vector, generate the reference values id ,ref and iq ,ref of the secondary circulating current dq components. The reference values are generated based on the optimization objective and are expressed as: i d,ref = I d,opt × cos( ϕ d,opt ) ; i q,ref = I q,opt × cos( ϕ q,opt ) ; Where: I d,opt and I q,opt are the amplitudes of the optimized dq components, ϕ d,opt and ϕ q,opt are the phase angles of the optimized dq components; The output of the PID controller is expressed as: u ( t ) = K p × e ( t ) + K i × + K d × ; Where: e ( t ) is the error signal, K p , K i , K d are the parameters of the PID controller; When adding the optimal third-harmonic voltage reference signal to the arm reference signal, generate the reference signal v 3rd,ref of the third-harmonic voltage, which is expressed as: v 3rd,ref =V 3rd,opt *cos(3 ωt + ϕ 3 rd,opt )); Wherein: V 3rd,opt is the amplitude of the optimized third - harmonic voltage, ϕ 3 rd,opt is the phase angle of the optimized third - harmonic voltage; The arm reference signal is expressed as: v ref ( t ) = v arm,ref ( t ) + v 3rd,ref ( t ); Wherein: v arm,ref ( t ) is the original reference signal of the arm; When operating under given conditions, the MMC system adjusts the injected secondary circulating current and third - harmonic voltage in real - time according to the current operating state and optimization parameters, so that the MMC system always operates in the state of optimal power loss, realizing the cooperative optimization control of multi - objective power loss of the MMC system.

[0012] An embodiment of the present invention also provides an electronic device, including a memory and a processor; The memory is used to store a computer program; The processor, when executing the computer program stored in the memory, realizes the steps of a method for optimizing multi - objective power loss control of a modular multilevel converter as described above.

[0013] An embodiment of the present invention also provides a computer - readable storage medium for storing a computer program, and when the computer program is executed by a processor, the steps of a method for optimizing multi - objective power loss control of a modular multilevel converter as described above are realized.

[0014] An embodiment of the present invention provides a method for optimizing multi - objective power loss control of a modular multilevel converter. Compared with the prior art, its beneficial effects are as follows: The present invention uses different secondary circulations and third harmonics as different injection schemes for the power loss optimization model. Each injection scheme is input into the power loss optimization model to obtain the total power loss of the modular multilevel converter and the loss of the device with the highest loss under different injection schemes. The different injection schemes are used as different initial solutions of the multi-objective mantis search algorithm MOMSA. Each initial solution is non-dominated sorted to identify solutions that are not worse than other solutions in multiple objectives. At the same time, the crowding distance of the solutions in each non-dominated layer is calculated to evaluate the distribution diversity of the solutions, and finally the high-quality solutions and low-quality solutions are distinguished. After multiple iterative optimizations, a non-dominated solution set is obtained. This set contains the best solution set under the given multi-objectives, from which the optimal secondary circulation and third harmonic voltage can be obtained. This process combines the objective mantis search algorithm MOMSA, elitist non-dominated sorting NDS, and crowding distance CD, and iteratively optimizes to screen out the optimal injection scheme. This optimal injection scheme represents the one that can minimize both the total loss and the loss of the device with the highest loss within the sub-module in the MMC system, thus realizing the collaborative optimization of the total loss of the MMC system and the device with the highest loss in the sub-module.

[0015] Moreover, the present invention does not require adding a hardware burden. Compared with the existing methods that only consider the total power consumption optimization of the sub-module or the single loss optimization of the power device with the highest loss within the sub-module, it realizes the collaborative optimization of multi-objective losses, that is, it realizes the collaborative optimization of the modular multilevel converter and its sub-module devices, reduces the operating cost of the MMC system, and extends the service life of the power device. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a schematic diagram of the overall flow of a multi-objective power loss optimization control method for a modular multilevel converter provided by an embodiment of the present invention; Figure 2 It is a schematic diagram of the topology structure of a three-phase modular multilevel converter of a multi-objective power loss optimization control method for a modular multilevel converter provided by an embodiment of the present invention; Figure 3 It is a schematic diagram of the topology structure of a sub-module of a multi-objective power loss optimization control method for a modular multilevel converter provided by an embodiment of the present invention; Figure 4 It is a schematic diagram of the MMC output waveform after injecting secondary circulation and third harmonic voltage of a multi-objective power loss optimization control method for a modular multilevel converter provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will describe the specific embodiments of the present invention in detail with reference to the accompanying drawings. Many specific details are set forth in the following description to facilitate a thorough understanding of the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0018] See Figure 1 , an embodiment of the present invention provides a multi-objective power loss optimization control method for a modular multilevel converter, including the following steps: constructing a multi-objective mantis search algorithm model using hunting strategy simulation, Pareto dominance, and crowding distance; establishing a power loss optimization model for the MMC; using the multi-objective mantis search algorithm (MOMSA) to optimize the optimal values of the injected secondary circulating current and the third harmonic voltage, and giving the optimal values of the optimal injection amplitude and phase angle and the optimal power loss of the MMC under given conditions; injecting the optimal secondary circulating current and the third harmonic voltage given by the multi-objective mantis search algorithm (MOMSA) into the MMC system for multi-objective power loss optimization control. Compared with other loss optimization methods, the present invention does not require an additional hardware burden. At the same time, compared with the existing methods that only consider the total power consumption optimization of the sub-modules or the single-objective loss optimization of the power device with the highest loss in the sub-module, the multi-objective power loss optimization method based on the multi-objective mantis search algorithm proposed by the present invention can achieve the collaborative optimization of multi-objective power loss optimization.

[0019] Specifically, it includes the following steps: Step S1: According to the system information of the measured modular multilevel converter, including the number N of sub-modules on the bridge arm of the modular multilevel converter, the arm inductance L, and the structural information of the sub-modules; use the system information to construct a power loss model of the three-phase modular multilevel converter.

[0020] Step S2: Construct a multi-objective mantis search algorithm model for optimizing the power loss of the MMC using hunting strategy simulation, Pareto dominance, and crowding distance.

[0021] The mathematical model of the multi-objective mantis search algorithm (Multi-objective Mantis Search Algorithm, MOMSA) is mainly based on the hunting behavior and sexual cannibalism behavior of mantises, and combines the elite non-dominated sorting (NDS) and crowding distance (CD) mechanisms to handle multi-objective optimization problems.

[0022] 1. Elite non-dominated sorting (NDS).

[0023] Elitist non - dominated sorting is used to identify and rank non - dominated solutions, that is, those solutions that are not worse than other solutions in multiple objectives; the specific steps are as follows: Initialization: Generate the initial population P.

[0024] Non - dominated sorting: Divide the population P into multiple non - dominated levels F1, F2, F3... F k , where F1 is the first - level non - dominated solution, F2 is the second - level non - dominated solution, and so on.

[0025] Selection: Starting from F1, select the solutions in the non - dominated levels in turn until the population size N is reached.

[0026] 2. Crowding distance (CD).

[0027] The crowding distance mechanism is used to maintain the diversity of solutions and avoid the over - concentration of solutions in the objective space; the specific steps are as follows: Calculate the crowding distance: For each non - dominated level F i , calculate the crowding distance of each solution in the objective space; the calculation formula for the crowding distance is: CD(i)= .

[0028] Where: M is the number of objective functions, f m (i) is the value of the i - th solution on the m - th objective, f mmax and f mmin are the maximum and minimum values of the m - th objective respectively.

[0029] Selection: In each non - dominated level, select solutions according to the crowding distance, and give priority to solutions with larger crowding distances.

[0030] 3. Convergence mechanism of the Mantis Search Algorithm (MSA).

[0031] MOMSA adopts the same convergence mechanism as MSA, including three stages: searching for prey (exploration), attacking prey (exploitation), and sexual cannibalism; the specific formulas are as follows: Searching for prey (exploration): X i (t + 1)=X i (t)+ α·rand·(X best (t)-X i (t)).

[0032] Where: X i (t) is the position of the i - th solution at the t - th iteration, X best (t) is the position of the current optimal solution, α is the step - size coefficient, and rand is a random number.

[0033] Attacking prey (exploitation): X i (t + 1)=X i (t)+β·rand·(X best (t)-X i (t))。

[0034] Where: β is the step coefficient, usually less than α.

[0035] Cannibalism of the same kind: X i (t + 1)=X i (t)+r·rand·(X mate (t)-X i (t))。

[0036] Where: X mate (t) is the position of the paired solution, and r is the step coefficient.

[0037] Step S3: Use the multi-objective mantis search algorithm model to optimize the optimal values of the secondary circulation and the third harmonic voltage injected into the modular multilevel converter (MMC), calculate the optimal injection amplitude and the optimal phase angle value, and finally calculate the optimal power loss of the MMC under given conditions. Specifically, it includes:

[0038] 1. Initialization: According to the parameters of the MMC system, such as the number of sub-modules, arm inductance, etc., generate a set of initial solutions, and each solution represents a possible injection scheme for the secondary circulation and the third harmonic voltage, including the injection amplitude and the phase angle.

[0039] 2. Fitness evaluation: For each initial solution, inject it into the power loss model of the MMC and calculate the power loss of the MMC under this injection scheme. The power loss model is as described in the text and includes the losses of each power device and diode in the sub-module. According to the calculation result of the power loss, define the fitness function value for each solution. The fitness function can be the reciprocal of the power loss or other functions negatively correlated with the loss, so that the optimization objective is transformed into maximizing the fitness function value.

[0040] The fitness function can be the reciprocal of the total power loss of the MMC, that is, Fitness = , where P total is the total power loss of the MMC. Such a definition means that the higher the fitness value, the better the corresponding solution, that is, the lower the total power loss.

[0041] 3. Non-dominated sorting and crowding distance calculation: Sort all the initial solutions according to the fitness value by non-dominated sorting, and identify the non-dominated solutions, that is, the solutions that are not worse than other solutions in multiple objectives (such as reducing the total loss and reducing the loss of the highest loss device). Then calculate the crowding distance of the solutions in each non-dominated layer to evaluate the distribution diversity of the solutions.

[0042] 4. Selection and Update: Select a part of the solutions from the non-dominated solutions as the parents, and ensure the diversity of the population according to methods such as crowding distance. Then, according to the position update formula of the mantis search algorithm, combining global and local information, update the positions of the parent solutions to generate new candidate solutions. The injection scheme of the new candidate solutions is injected into the MMC power loss model again to calculate the new fitness values.

[0043] 5. Iterative Optimization: Combine the parent solutions and the newly generated candidate solutions, and conduct fitness evaluation, non-dominated sorting, and crowding distance calculation again. According to the result of non-dominated sorting, select a new population, retain the high-quality non-dominated solutions, and replace the inferior solutions. Check whether the termination conditions are met, such as reaching the maximum number of iterations, the fitness change is not obvious, etc. If the termination conditions are not met, continue with fitness evaluation and update until the termination conditions are met.

[0044] 6. Output the Optimal Solution: Finally, output the set of non-dominated solutions obtained through multiple iterations of optimization. These solutions represent the best solution set under the given multi-objectives. From them, the optimal injection amplitudes and phase angles of the secondary circulation and the third harmonic voltage can be obtained, as well as the corresponding optimal power loss of the MMC.

[0045] Step S4: Inject the calculated optimal injection amplitude and optimal phase angle values into the MMC system to achieve multi-objective power loss optimization. Specifically, it includes:

[0046] 1. First, according to the optimal injection amplitude and phase angle given by the multi-objective mantis search algorithm, obtain the secondary circulation component in the MMC through algebraic calculation and a band-pass filter. Here, the algebraic calculation is mainly based on the mathematical relationship between the injected voltage and the circulation, and calculate the corresponding circulation amplitude and phase angle. The band-pass filter is used to extract the circulation component of a specific frequency (such as the second harmonic) from the calculated circulation signal to ensure the accuracy of the injected circulation signal.

[0047] 2. Then, obtain the dq components of the secondary circulation through coordinate transformation. Coordinate transformation is to convert the three-phase circulation signal to the synchronous rotating dq coordinate system for vector control. In the dq coordinate system, the d-axis component usually corresponds to the excitation component, and the q-axis component corresponds to the torque component. By controlling the dq components respectively, precise control of the circulation can be achieved.

[0048] 3. Next, the dq components of the secondary circulation are made to follow their reference values through closed-loop vector control. Closed-loop vector control is an accurate control method. By comparing the deviation between the actual dq components of the secondary circulation and the reference values, and using adjustment means such as a proportional-integral-derivative (PID) controller, the control signals of the MMC system are adjusted so that the actual dq components of the secondary circulation can quickly and accurately track the reference values, thereby achieving the precise injection of the secondary circulation.

[0049] 4. Finally, the optimal third-harmonic voltage reference signal is added to the arm reference signal. This step is to further optimize the power loss of the MMC. On the basis of injecting the secondary circulation, by adding the third-harmonic voltage, the output waveform of the MMC can be improved, the switching frequency of the switching devices can be reduced, and thus the switching loss can be reduced, realizing the collaborative optimization of multiple-objective power loss of the MMC.

[0050] 5. When operating under given conditions, the system will adjust the injected secondary circulation and third-harmonic voltage in real time according to the current operating state and optimization parameters to adapt to the changes in the system operating conditions, ensuring that the MMC system always operates in the state of optimal power loss. This real-time adjustment method can give full play to the advantages of the optimization results of the multi-objective mantis search algorithm and achieve long-term stable power loss optimization control.

[0051] As Figure 4 shown, it is the output waveform diagram of the MMC system; from Figure 4 the intersection of the traditional method and the proposed method on the abscissa in Figure 4 , the optimal values of the calculated secondary circulation and third-harmonic voltage are injected, and the waveform changes, and the experimental results are recorded. The experimental results are as

[0052] shown. a Among them, the sub-module is a half-bridge structure, which consists of two diodes D b , D a , two power switches T b , T and a DC capacitor C, and the power loss model of the three-phase modular multilevel converter is expressed as:

[0053] Where: represents the loss of the power switch T a in the sub-module, represents the loss of the power switch T b in the sub-module, represents the loss of the diode D a in the sub-module, represents the loss of the diode D b in the sub-module; is the number of sub-modules SM in each arm of the modular multilevel converter MMC; in the formula P con_Ta , P on_Ta and P off_Ta are the conduction loss, turn-on loss and turn-off loss of the upper power device of the SM respectively; P con_Tb , P on_Tb and P off_Tb are the conduction loss, turn-on loss and turn-off loss of the lower power device of the SM respectively; P con_Da and P rec_Da are the conduction loss and reverse recovery loss of the upper diode of the SM respectively; P con_Db and P rec_Db are the conduction loss and reverse recovery loss of the lower diode of the SM respectively.

[0054] Mantis Search Algorithm (MSA): 1. Position update in the Mantis Search Algorithm.

[0055] The position update of the mantis individual is based on the target and the positions of other individuals, combining global and local information; its update formula is expressed as: X （t+1） =X （t） +α·Ftrack+β·Fattack.

[0056] Where: Ftrack represents the direction of the individual tracking the target; Fattack represents random perturbation or mutation to enhance the diversity of the population; α, β are control parameters used to adjust the balance between exploration and exploitation; X （t） represents the current position at time t.

[0057] Global information: In the Multi-Objective Mantis Search Algorithm (MOMSA), the global information mainly comes from the optimal solution of the entire population, that is, the solution that performs best on all objectives in the current iteration; this optimal solution represents the best search direction of the population in the global scope and can guide other individuals to move towards the global optimal region, thus achieving global search; specifically including: (1) Location of the optimal solution: Xbest(t) represents the location of the solution that performs best among the entire population in the multi-objective optimization problem in the current iteration; this location is the optimal location obtained by comprehensively considering all objective function values, and it contains the injection scheme (amplitude and phase angle of the secondary circulation and third-harmonic voltage) that can achieve a better balance for all optimization objectives (such as the total power loss of the MMC and the loss of the highest-loss device in the sub-module) in the current iteration.

[0058] (2) Objective function values: The respective objective function values corresponding to the location of the optimal solution, such as the total power loss P total and the loss of the highest-loss device P max ; these objective function values reflect the specific performance of the optimal solution in each optimization objective, provide performance indicators of the optimal solution within the global scope for the algorithm, and help the algorithm evaluate the pros and cons of the current search direction.

[0059] Local information: Local information mainly comes from the individual itself and the information within its neighborhood. It can help the individual conduct a detailed search within the local area, adjust its search direction and step size, and thus find a better solution within the local scope; specifically including: (1) Current location of the individual: Xi ( t ) represents the location of the i th individual in the t nd iteration, that is, the current injection scheme (amplitude and phase angle of the secondary circulation and third-harmonic voltage) of the individual; this location is the starting point of the individual in the local search process. Based on this location, the individual will combine global information and other local information to adjust its own location.

[0060] (2) Relative position of the individual to the optimal solution: X best( t ) − Xi ( t ) represents the difference between the current location of the individual and the location of the global optimal solution, that is, the distance between the individual and the optimal solution in each dimension (amplitude and phase angle); this relative position information reflects the pros and cons of the individual relative to the global optimal solution within the local area, and provides the local search direction for the individual; if the individual differs greatly from the optimal solution in a certain dimension, it means there is a large room for improvement in that dimension, and the individual will adjust the value of that dimension in the direction of the optimal solution; conversely, if the difference is small, it means the individual is already close to the optimal solution in that dimension, and a more detailed local search may be required in that dimension.

[0061] MOMSA realizes the efficient search of the multi-objective solution space by simulating the hunting behavior of mantises. Its core lies in combining natural inspiration and mathematical modeling, and maintaining the diversity and balance of solutions through Pareto optimization.

[0062] Specifically, the calculation process of the multi-objective mantis search algorithm includes: Randomly generate a set of initial solutions (mantises). Each solution is usually defined by multiple power loss objective functions and constraint conditions to ensure that the initial solutions cover the entire search space. Define the parameters of the algorithm, such as the maximum number of iterations, fitness function, etc. For each solution in the population, calculate its fitness value on each objective. Fitness evaluation is the key to multi-objective optimization, and usually, a fitness function needs to be designed according to the specific problem. Perform non-dominated sorting on the solutions in the population to identify non-dominated solutions. These solutions constitute the Pareto front, representing the best balance between different objectives. Select a part of the non-dominated solutions as parents for the next generation generation operation. Methods such as crowding distance can be used to ensure the diversity of the population. According to the selected parent solutions, use movement strategies (such as exploration and exploitation) to generate new candidate solutions. Usually, the hunting behavior of mantises is simulated, and by introducing update mechanisms such as attraction, repulsion, or quantum rotation gates, the new solutions are made closer to high-quality solutions. Combine the parent solutions and the newly generated solutions to form a new solution set. Calculate the fitness values of the merged new solution set again and perform non-dominated sorting. According to the results of non-dominated sorting, select a new population to ensure that the size of the population remains unchanged. Usually, high-quality non-dominated solutions are retained and inferior solutions are replaced. Check whether the termination conditions are met, such as reaching the maximum number of iterations, the fitness change is not obvious, etc. If the conditions are met, stop the algorithm. If the termination conditions are not reached, return to step 2 and continue with fitness evaluation and update.

[0063] Finally, output the set of non-dominated solutions obtained through multiple iterations of optimization, representing the best solution set under the given multi-objectives. The specific representation is as follows:

[0064] 1. Initialize the population: Generate the initial solution set and calculate the fitness value of each solution.

[0065] 2. Non-dominated sorting: Perform non-dominated sorting on the population to identify non-dominated solutions.

[0066] 3. Crowding distance calculation: Calculate the crowding distance of solutions in each non-dominated layer to maintain the diversity of solutions.

[0067] 4. Position update: According to the position update formula of the mantis search algorithm, combine global and local information to update the position of the solution.

[0068] 5. Iterative optimization: Through multiple iterations, continuously optimize the solution set, and finally output the set of non-dominated solutions.

[0069] 6. Clean the population: Clean the population to ensure that the population size remains unchanged, and select high-quality solutions according to the ranking and crowding distance.

[0070] The parameter calculations and optimization processes involved in the code are closely related to the multi-objective mantis search algorithm (MOMSA) model construction and the MMC power loss optimization model in the document. The specific parameter calculations are as follows:

[0071] ① Initialize the parameters.

[0072] name: The name of the multi-objective optimization problem. M: The number of objective functions. fobj: The calculation function of the objective function. dim: The dimension of the solution. lb: The lower bound of the solution. ub: The upper bound of the solution. Max_iter: The maximum number of iterations. SearchAgents_no: The population size. ishow: The display interval.

[0073] ② Control parameter op: The probability of random perturbation.

[0074] A: The archive size. a: The step size coefficient. P: The number of non-dominated levels. alp: The adjustment parameter of the step size coefficient. Pc: The crossover probability.

[0075] ③ Initialize the population.

[0076] Generate the initial solution set Positions, and each solution contains the injection amplitude and phase angle of the secondary circulation and the third harmonic voltage.

[0077] Calculate the fitness value f of each solution, and the fitness function is the reciprocal of the total power loss of the MMC.

[0078] ④ Non-dominated sorting and crowding distance calculation.

[0079] Perform non-dominated sorting to divide the population into multiple non-dominated levels.

[0080] Calculate the crowding distance of the solutions in each non-dominated level, and select the solutions with a larger crowding distance.

[0081] ⑤ Position update.

[0082] Use the position update formula, combine global and local information, and update the position of the solution.

[0083] The position update formula includes: Explore: X i (t + 1) = X i (t) + α·rand·(X best (t) - X i (t)).

[0084] Exploit: X i (t + 1) = X i(t) + β·rand·(X best (t) - X i (t))。

[0085] Cannibalism of the same kind: X i (t + 1) = X i (t) + r·rand·(X mate (t) - X i (t))。

[0086] ⑥ Iterative optimization.

[0087] Combine the parent solutions and the newly generated candidate solutions to form a new solution set.

[0088] Recalculate the fitness values and perform non - dominated sorting.

[0089] Select a new population to ensure that the population size remains unchanged.

[0090] Check whether the maximum number of iterations is reached. If so, stop the algorithm; otherwise, continue the iteration.

[0091] ⑦ Clean up the population.

[0092] Clean up the population to ensure that the population size remains unchanged, and select high - quality solutions according to the ranking and crowding distance.

[0093] Specific parameter calculation examples are as follows: Assume that the MMC system parameters are as follows: The number of sub - modules N = 100; the arm inductor L = 10 μH; the maximum number of iterations T = 100; the population size N = 50; the step - size coefficients α = 0.5, β = 0.2, r = 0.1.

[0094] ① Initialization.

[0095] Generate 50 initial solutions, and each solution contains the injection amplitudes and phase angles of the second - order circulating current and the third - order harmonic voltage.

[0096] Define the algorithm parameters: Max_iter = 100, SearchAgents_no = 50, alpha = 0.5, beta = 0.2, r = 0.1.

[0097] ② Fitness evaluation.

[0098] For each initial solution, calculate the total power loss of the MMC P total.

[0099] Calculate the fitness function value: f(i, 1:M) = 1 / P_total.

[0100] ③ Non - dominated sorting and crowding distance calculation.

[0101] Perform non - dominated sorting to divide the population into multiple non - dominated levels.

[0102] Calculate the crowding distance of the solutions in each non - dominated level and select the solutions with larger crowding distances.

[0103] ④Position update.

[0104] Use the position update formula, combining global and local information, to update the positions of the solutions.

[0105] The position update formula includes: Explore: X i (t + 1)=X i (t)+ 0.5·rand·(X best (t)-X i (t)).

[0106] Exploit: X i (t + 1)=X i (t)+ 0.2·rand·(X best (t)-X i (t)).

[0107] Sexual cannibalism: X i (t + 1)=X i (t)+0.1·rand·(X mate (t)-X i (t)).

[0108] ⑤Iterative optimization.

[0109] Merge the parent solutions and the newly generated candidate solutions to form a new solution set.

[0110] Recalculate the fitness values and perform non - dominated sorting.

[0111] Select a new population to ensure that the population size remains unchanged.

[0112] Check whether the maximum number of iterations 100 is reached. If so, stop the algorithm; otherwise, continue the iteration.

[0113] ⑥Output the optimal solution.

[0114] Output the non - dominated solution set and select the optimal injection amplitudes and phase angles of the secondary circulation and the third - harmonic voltage from it, as well as the corresponding optimal power loss of the MMC.

[0115] Through the above code and parameter calculations, the multi - objective power loss optimization of the MMC system can be achieved, ensuring that the optimal injection scheme is found under given conditions, reducing the operating cost of the system, and extending the service life of power devices.

[0116] In step S4, when injecting the optimal injection amplitude and phase angle given by the multi-objective mantis search algorithm for the circulating current, the secondary circulating current in the MMC can be obtained through algebraic calculation and a band-pass filter; the dq components of the secondary circulating current are obtained through coordinate transformation, and the dq components of the secondary circulating current are made to follow their reference values through closed-loop vector control. Then, the third harmonic voltage signal is added to the arm reference signal to minimize the total loss and the maximum loss of the sub-modules, and finally, the multi-objective power loss optimization control of the MMC is realized. The specific calculation process is as follows:

[0117] 1. Injection of secondary circulating current and third harmonic voltage. It includes:

[0118] ① Algebraic calculation: According to the optimal injection amplitude and phase angle given by the MOMSA algorithm, calculate the injection signals of the secondary circulating current and the third harmonic voltage.

[0119] The injection signal of the secondary circulating current can be expressed as: i 2nd ( t ) = I 2nd * cos( ωt + ϕ 2nd ).

[0120] Where: I 2nd is the amplitude of the secondary circulating current, ϕ 2nd is the phase angle of the secondary circulating current, ω is the angular frequency.

[0121] The injection signal of the third harmonic voltage can be expressed as: v 3rd ( t ) = V 3rd * cos(3 ωt + ϕ 3rd ).

[0122] Where: V 3rd is the amplitude of the third harmonic voltage, ϕ 3rd is the phase angle of the third harmonic voltage.

[0123] ② Band-pass filter: Use a band-pass filter to extract the circulating current component of a specific frequency (such as the secondary) from the calculated circulating current signal to ensure the accuracy of the injected circulating current signal.

[0124] The transfer function of the band-pass filter can be expressed as: H ( s ) = 。

[0125] Where: is the center frequency, Q is the quality factor.

[0126] 2. Coordinate transformation. It includes:

[0127] ① Three-phase to two-phase transformation: Convert the three-phase circulating current signal to the synchronous rotating dq coordinate system for vector control.

[0128] The transformation formula from three-phase to two-phase is: 。

[0129] Where: θ is the synchronous rotation angle, ia 、 ib 、 ic are the three-phase circulating current signals, id and iq are the circulating current components in the dq coordinate system.

[0130] 3. Closed-loop vector control. It includes:

[0131] ① Reference value generation: Generate the reference values id ,ref and iq ,ref of the secondary circulating current dq components.

[0132] The reference values can be generated based on the optimization objective, for example: i d,ref = I d,opt × cos( ϕ d,opt )。

[0133] i q,ref = I q,opt × cos( ϕ q,opt )。

[0134] Where: I d,opt and I q,opt are the amplitudes of the optimized dq components, ϕ d,opt and ϕq,opt is the optimized dq - component phase angle.

[0135] ② PID controller: Use a Proportional - Integral - Derivative (PID) controller to adjust the control signal of the MMC system, so that the actual secondary - circulating dq - components can quickly and accurately track the reference value.

[0136] The output of the PID controller can be expressed as: u ( t ) = K p × e ( t ) + K i × + K d × .

[0137] Where: e ( t ) is the error signal, K p 、 K i 、 K d are the parameters of the PID controller.

[0138] 4. Third - harmonic voltage injection. It includes:

[0139] ① Reference - signal generation: Generate the reference signal of the third - harmonic voltage v 3rd,ref .

[0140] The reference signal can be generated based on the optimization objective, for example: v 3rd,ref = V 3rd,opt *cos(3 ωt + ϕ 3 rd,opt ).

[0141] Where: V 3rd,opt is the optimized third - harmonic voltage amplitude, ϕ 3 rd,opt is the optimized third - harmonic voltage phase angle.

[0142] ② Arm - reference - signal adjustment: Add the optimal third - harmonic voltage reference signal to the arm - reference signal to further optimize the power loss of the MMC.

[0143] The arm reference signal can be expressed as: v ref ( t ) = v arm,ref ( t ) + v 3rd,ref ( t ).

[0144] Where: v arm,ref ( t ) is the original reference signal of the arm.

[0145] 5. Real-time adjustment. It includes:

[0146] ① System operation status monitoring: When operating under given conditions, the system will adjust the injected secondary circulating current and third harmonic voltage in real time according to the current operating status and optimization parameters to adapt to the changes in system operating conditions.

[0147] The operating status of the system is monitored in real time through sensors and measurement devices, such as current, voltage, temperature, etc.

[0148] ② Control signal adjustment: According to the monitored operating status, the parameters of the PID controller are adjusted to ensure the stable operation of the system.

[0149] Through closed-loop control, the actual dq components of the secondary circulating current can track the reference value quickly and accurately, so as to achieve the precise injection of the secondary circulating current.

[0150] ③ Optimization control: By adjusting the injected secondary circulating current and third harmonic voltage in real time, it is ensured that the MMC system always operates in the state of optimal power loss. The present invention proposes a multi-objective loss optimization method for sub-modules of a modular multilevel converter aiming at the loss optimization problem of the modular multilevel converter. The topologies of the three-phase MMC and its sub-modules are as Figure 2 and Figure 3 shown. The MMC is composed of six arms, and each arm contains n (n = 1, 2,..., N) sub-modules (Submodule, SM) with the same topological structure and an arm inductor L; the sub-module is a half-bridge structure, which consists of two diodes Da, Db, two power switches Ta, Tb and a DC capacitor C.

[0151] The multi-objective power loss optimization control method for modular multilevel converters based on the multi-objective mantis search algorithm proposed in the present invention constructs a multi-objective mantis search algorithm model by simulating hunting strategies and using Pareto domination and crowding distance; establishes a power loss optimization model for the MMC; uses the multi-objective mantis search algorithm (MOMSA) to optimize the optimal values of the injected secondary circulating current and the third harmonic voltage, obtains the optimal values of the optimal injection amplitude and phase angle, and the optimal power loss of the MMC under given conditions; injects the given optimal injection amplitude and phase angle into the MMC system for multi-objective power loss optimization control. This realizes the collaborative optimization of the multi-objective power loss of the MMC, reduces the operating cost of the MMC system, and extends the service life of power devices. At the same time, the mantis search algorithm has good global search ability. When solving complex nonlinear problems, it can avoid falling into local optimal solutions and thus find better solutions. Moreover, in the face of system parameter uncertainties, the algorithm can effectively find robust solutions to ensure stable performance under different operating conditions.

[0152] The power loss optimization control method for modular multilevel converters based on the multi-objective mantis search algorithm proposed in the present invention is an improvement on the basis of traditional algorithms. When applied to the MMC, it has the advantages of fast convergence speed, high accuracy, and good robustness. At the same time, compared with traditional multi-objective optimization algorithms such as NSGA-II (Non-dominated Sorting Genetic Algorithm II) or MOEA / D (Multi-Objective Evolutionary Algorithm based on Decomposition), MOMSA has stronger exploration ability and dynamic adaptability, and has broad application prospects in solving complex multi-objective optimization problems.

[0153] The multi-objective power loss optimization control method of modular multilevel converter based on multi-objective mantis search algorithm constructs a multi-objective mantis search algorithm model by simulating hunting strategies, Pareto domination, and crowding distance; establishes an optimization model for the power loss of MMC; uses the multi-objective mantis search algorithm (MOMSA) to optimize the optimal values of the injected secondary circulating current and the third harmonic voltage, obtains the optimal values of the optimal injection amplitude and phase angle, and the optimal power loss of MMC under given conditions; uses the given optimal injection amplitude and phase angle, and then injects them into the MMC system for power loss optimization control. This realizes the collaborative optimization of the total loss of MMC and the power device with the highest sub-module loss, reduces the operating cost of the MMC system, increases the service life of the power device, and enhances the overall reliability of the MMC system. The relatively fast convergence speed of this algorithm can find a solution close to the optimal in a relatively short time, and is suitable for power electronic systems that require real-time control. At the same time, the MOMSO algorithm can adapt to different system configurations and operating conditions, adjust the search strategy through an adaptive mechanism, and optimize the performance. This algorithm can also handle multiple objectives simultaneously, such as reducing the loss of the power device with the highest loss in the sub-module and the total loss of the sub-module power devices at the same time. If only considering the loss optimization of a single objective, it will be difficult to achieve the collaborative optimization of high efficiency and high reliability of MMC. These technologies can help reduce the total power loss of the MMC system and the highest loss in the sub-module, thereby improving the overall operating efficiency and service life of the system.

[0154] The above embodiments only represent several implementation manners of the present invention. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention patent shall be subject to the appended claims.

Claims

1. A multi-objective power loss optimization control method for a modular multi-level converter, characterized in that: The following steps are involved: Using the structural information of modular multilevel converter, a power loss optimization model of modular multilevel converter is constructed; Different secondary circulating currents and third harmonics are used as different injection schemes of the power loss optimization model, and each injection scheme is input into the power loss optimization model to obtain the total power loss of the modular multilevel converter and the loss of the highest loss device under different injection schemes; The injection scheme that minimizes the total power loss of the modular multilevel converter and the loss of the highest loss device is sought. In the search, different injection schemes are used as different solutions of the multi-objective mantis search algorithm MOMSA. The elite non-dominated sorting NDS is used to perform non-dominated sorting on different solutions. The solutions whose losses of the devices in the modular multilevel converter and multiple sub-modules are less than the preset loss threshold are taken as non-dominated solutions. The crowding distance CD is used to calculate the crowding distance of each non-dominated solution. The crowding distance of each non-dominated solution represents the sum of the differences between the current solution and all adjacent solutions. The solution with the largest crowding distance is taken as the injection scheme to be sought. The modular multilevel converter is controlled using the sought injection scheme.

2. A modular multi-level converter multi-objective power loss optimization control method according to claim 1, characterized in that: The construction of the power loss optimization model includes: According to the number N of submodules on the bridge arm of the modular multilevel converter and the bridge arm inductance L, as well as the structure consisting of two power switches, two diodes and a DC capacitor included in the submodule, a power loss optimization model of the three-phase modular multilevel converter MMC system is constructed; the power loss optimization model is expressed as: ； in: Represents the power switch T in the submodule a The loss, Represents the power switch T in the submodule b The loss, Indicates the diode D in the submodule a The loss, Indicates the diode D in the submodule b loss; is the number of submodules SM in each bridge arm of the modular multilevel converter MMC; P con_Ta , P on_Ta and P off_Ta They are the conduction loss, turn-on loss and turn-off loss of the power devices on the SM. P con_Tb , P on_Tb and P off_Tb They are the conduction loss, turn-on loss and turn-off loss of the power devices at the bottom of the SM respectively; P con_Da and P rec_Da are the conduction loss and reverse recovery loss of the upper diode of SM respectively; P con_Db and P rec_Db They are the conduction loss and reverse recovery loss of the diode at the bottom of the SM, respectively.

3. A modular multi-level converter multi-objective power loss optimization control method according to claim 2, characterized in that: The method of finding an injection scheme that minimizes the total power loss of the modular multi-level converter and the loss of the highest loss device comprises: According to the number of submodules N and the bridge arm inductance L in the three-phase modular multilevel converter MMC system, and the two power switch tubes T in the submodule a and T b , two diodes D a and D b , a DC capacitor C, generates a set of initial solutions, each of which represents an injection scheme of secondary circulating current and third harmonic voltage; For each initial solution, it is injected into the power loss optimization model of the three-phase modular multilevel converter MMC system to obtain the power loss of the three-phase modular multilevel converter MMC system under the injection scheme; according to the power loss, a fitness function value is defined for each solution. The fitness function is the inverse of the total power loss, expressed as Fitness= ,in P total is the total power loss of the MMC system; Use elite non-dominated sorting (NDS) to sort all initial solutions according to their fitness values, obtain non-dominated solutions, which are solutions that are no worse than other solutions in multiple objectives, and obtain the crowding distance of the solutions in each non-dominated layer. A part of the solutions are selected from the non-dominated solutions as the parent generation, and the position of the parent generation solution is updated according to the position update formula of the multi-objective mantis search algorithm MOMSA, combining global and local information to generate new candidate solutions; the injection scheme of the new candidate solution is injected into the power loss optimization model again to calculate the new fitness value; The parent solution and the newly generated candidate solution are merged, and fitness evaluation, non-dominated sorting and crowding distance calculation are performed again; according to the result of non-dominated sorting, a new population is selected, and high-quality non-dominated solutions are retained and low-quality solutions are replaced; fitness evaluation and updating are stopped until the maximum number of iterations is reached or the fitness change is not obvious; The non-dominated solution set obtained after multiple iterative optimizations is output. These solutions represent the optimal solution set under given multiple objectives, and the optimal secondary circulating current and third harmonic voltage are obtained from the optimal solution set.

4. A modular multi-level converter multi-objective power loss optimization control method according to claim 3, characterized in that: The position update of the multi-target mantis search algorithm MOMSA is based on the position of the target and other individuals, and combines global and local information; The position update formula is expressed as: X (t+1) =X （t） +α·Ftrack+β·Fattack; Among them: Ftrack represents the direction of the individual tracking target; Fattack represents random disturbance or mutation; α and β represent control parameters; X （t） Indicates the current position at time t; Global information comes from the optimal solution of the entire population, which indicates the solution with the best performance on all targets in the current iteration. At the same time, this optimal solution indicates the best search direction of the population in the global scope, which can guide other individuals to move to the global optimal area for global search. Local information comes from the information of the individual itself and its neighborhood, which can help the individual conduct detailed searches in the local area and adjust its search direction and step size to find a better solution in the local range.

5. The multi-objective power loss optimization control method for a modular multi-level converter according to claim 1, characterized in that: The control modular multi-level converter comprises: According to the optimal values ​​of the secondary circulating current and the third harmonic voltage, the optimal injection amplitude and the optimal phase angle are obtained; according to the optimal injection amplitude and the optimal phase angle, the secondary circulating current component in the three-phase modular multilevel converter MMC system is obtained through algebraic calculation and bandpass filter; wherein, the algebraic calculation is to calculate the corresponding circulating current amplitude and phase angle according to the mathematical relationship between the injection voltage and the circulating current; the bandpass filter is used to extract the circulating current component of a specific frequency from the calculated circulating current signal; The dq components of the secondary circulating current are obtained from the secondary circulating current components by means of coordinate transformation; the coordinate transformation is to transform the three-phase circulating current signal into a synchronously rotating dq coordinate system; in the dq coordinate system, the d-axis component corresponds to the excitation component, and the q-axis component corresponds to the torque component, and the circulating current is controlled by controlling the dq components respectively; The closed-loop vector control allows the secondary circulating current dq component to follow its reference value; the closed-loop vector control compares the deviation between the actual secondary circulating current dq component and the reference value, and uses the proportional-integral-differential PID controller to adjust the control signal of the MMC system so that the actual secondary circulating current dq component can track the reference value to achieve the injection of the secondary circulating current; On the basis of injecting secondary circulating current, the optimal third harmonic voltage reference signal is added to the bridge arm reference signal to realize the collaborative optimization of multi-objective power loss of MMC system to control modular multilevel converter.

6. A modular multi-level converter multi-objective power loss optimization control method according to claim 5, characterized in that: The collaborative optimization control of multi-objective power loss of the MMC system includes: The injection signal of the secondary circulation can be expressed as: i 2nd ( t )= I 2nd *cos( ωt + ϕ 2nd ); in: I 2nd is the amplitude of the secondary circulation, ϕ 2nd is the phase angle of the secondary circulation, ω is the angular frequency; The injection signal of the third harmonic voltage can be expressed as: v 3rd ( t )= V 3rd *cos(3 ωt + ϕ 3rd ); in: V 3rd is the amplitude of the third harmonic voltage, ϕ 3rd is the phase angle of the third harmonic voltage; The transfer function of the bandpass filter can be expressed as: H ( s )= ； in: is the center frequency, Q is the quality factor; The three-phase to two-phase transformation formula is expressed as: ； in: θ is the synchronous rotation angle, ia , ib , ic is the three-phase circulating current signal, id and iq is the circulation component in the dq coordinate system; In closed loop vector, the reference value of the secondary circulating current dq component is generated id ,ref and iq ,ref,The reference value is generated based on the optimization goal and is expressed as: i d,ref = I d,opt ×cos( ϕ d,opt ); i q,ref = I q,opt ×cos( ϕ q,opt ); in: I d,opt and I q,opt is the optimized dq component amplitude, ϕ d,opt and ϕ q,opt is the optimized dq component phase angle; The output of the PID controller is expressed as: u ( t )= K p × e ( t )+ K i × + K d × ； in: e ( t ) is the error signal, K p , K i , K d are the parameters of the PID controller; When the optimal third harmonic voltage reference signal is added to the bridge arm reference signal, a third harmonic voltage reference signal is generated based on the optimization target. v 3rd,ref , expressed as: v 3rd,ref = V 3rd,opt *cos(3 ωt + ϕ 3 rd,opt ); in: V 3rd,opt is the optimized third harmonic voltage amplitude, ϕ 3 rd,opt is the optimized third harmonic voltage phase angle; The bridge arm reference signal is expressed as: v ref ( t )= v arm,ref ( t )+ v 3rd,ref ( t )； in: v arm,ref ( t ) is the original reference signal of the bridge arm; When operating under given conditions, the MMC system adjusts the injected secondary circulating current and third harmonic voltage in real time according to the current operating status and optimization parameters, so that the MMC system always operates in the state of optimal power loss, realizing the coordinated optimization control of multi-objective power loss of the MMC system.

7. An electronic device, characterized in that: include: Memory and processor; The memory is used to store computer programs; The processor is used to implement the steps of a modular multi-level converter multi-objective power loss optimization control method as described in any one of claims 1 to 6 when executing the computer program stored in the memory.

8. A computer-readable storage medium, characterized in that: Used to store a computer program, which, when executed by a processor, implements the steps of a modular multi-level converter multi-objective power loss optimization control method as described in any one of claims 1 to 6.

Images