Energy storage converter control system optimization method based on hierarchical learning

Abstract

The invention provides an energy storage converter control system optimization method based on hierarchical learning, and the method comprises the steps: firstly, randomly generating a group of particles, and enabling each particle to represent a group of possible controller parameters; thirdly, the fitness of each particle is evaluated according to performance indexes of the system, and the particles are divided into a plurality of layers according to fitness values; different learning strategies are applied to the particles of different levels, the particles of the low level pay more attention to exploration of a new parameter space, and the particles of the high level pay more attention to development in a known excellent parameter area. Two particles are randomly selected from a higher level as samples, and learning of other particles is guided, so that the positions of the particles are updated. The process is repeated until the number of iterations is satisfied or a satisfactory solution is found. According to the method, the LPSO algorithm is adopted, the global search capability and the local development capability of the algorithm are remarkably improved through a hierarchical learning strategy, and a solution with higher quality can be found in a complex optimization problem.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent control optimization of energy storage converters, and in particular relates to an energy storage converter control system optimization method based on hierarchical learning. Background Art

[0002] In today's rapidly developing field of power electronics, energy storage converters (power converters) serve as a crucial bridge between energy sources and loads, and their performance directly impacts the stability and efficiency of the entire power system. The core of an energy storage converter system lies in its controller design, and optimizing controller parameters is crucial to system performance. With the continuous development of new energy technologies, such as the growing demand for grid-connected renewable energy sources like solar and wind power, the control strategies and parameter optimization issues of energy storage converters have become increasingly complex and pressing.

[0003] Traditional controller parameter optimization methods, including classic PID control, fuzzy control, and genetic algorithms, often have limitations. For example, PID control parameter adjustment relies on the engineer's experience and is difficult to adapt to complex and changing system dynamics. While fuzzy control has a certain degree of adaptability, it lacks a systematic optimization method for parameter adjustment.

[0004] As a method that can break away from reliance on engineering experience and optimize parameter adjustment, evolutionary algorithms provide a solution for optimizing the control system of energy storage converter systems. However, traditional evolutionary algorithms also have some drawbacks:

[0005] (1) Traditional evolutionary algorithms may require a large number of individuals and multiple generations of iterations to ensure the quality and diversity of solutions, which may result in high computational costs.

[0006] (2) Traditional evolutionary algorithms require users to manually set and adjust multiple parameters, such as crossover rate, mutation rate, and population size, which may be a challenge for users.

[0007] (3) When traditional evolutionary algorithms deal with complex problems with many local optimal solutions, they may converge to the local optimal solution too early. Summary of the Invention

[0008] Purpose of the Invention: This invention addresses the shortcomings of existing technologies by providing a hierarchical learning-based optimization method for energy storage converter control systems. This method utilizes a hierarchical learning-based particle swarm optimization (LPSO) algorithm to optimize the controller parameters of the energy storage converter. Compared to traditional particle swarm optimization (PSO), LPSO achieves more efficient search and optimization by dividing particles into different hierarchies and applying differentiated learning strategies at each level.

[0009] The method comprises the following steps:

[0010] Step 1: Collect historical power grid data and preprocess it;

[0011] Step 2, by using the fitness function to calculate the fitness value and evaluate the fitness value of each particle to measure the performance of each particle;

[0012] Step 3: Particles are divided into different levels according to their fitness. Particles are divided into low-level particles and high-level particles according to their fitness, forming a hierarchical learning structure, in which low-level particles focus on exploring new parameter spaces, and high-level particles focus on developing within known excellent parameter areas;

[0013] Step 4: Through a hierarchical learning strategy, low-level particles enhance their exploration capabilities by learning from excellent particles at higher levels; high-level particles enhance their development capabilities by learning from excellent particles at lower levels.

[0014] Step 5: Randomly select two particles from the high level as samples to update the individual optimal pbest and neighborhood optimal nbest respectively to guide the learning of low-level particles;

[0015] Step 6: Set up the particle swarm optimization algorithm based on hierarchical learning ( Layered Particle Swarm Optimization ) LPSO parameters;

[0016] Step 7, adjusting the inertia weight of the particle swarm optimization algorithm based on hierarchical learning LPSO through self-adaptation;

[0017] Step 8: Update the particle position and adjust the controller parameters according to the system feedback;

[0018] Step 9: Repeat evaluation, stratification, learning, and updating until the optimal solution is found.

[0019] Step 10: Apply the parameters corresponding to the particle with the best fitness to the energy storage converter system.

[0020] Step 1 includes: the power grid historical data includes the voltage V(t) at time t, the current I(t) at time t, and the power factor cos(φ).

[0021] Step 2 includes: calculating the fitness F using the following formula:

[0022] F=ω1×f1+ω2×f2+ω3×f3,

[0023] Among them, ω1, ω2, and ω3 are weight coefficients, f1 is the tracking speed index of the output voltage; f2 is the output voltage overshoot; and f3 is the output voltage static error.

[0024] Step 3 includes: in the particle swarm optimization algorithm based on hierarchical learning LPSO, it is assumed that there are NP particles divided into NL levels, each level contains the same number of particles, called the level size LS, and the calculation formula is:

[0025]

[0026] The fitness value of the particle F(X i ) determines the position of the particle in the hierarchy, the i-th particle X i The mathematical expression assigned to the lth layer is:

[0027]

[0028] Among them, rank(F(X i )) is the i-th particle X i The ranking of the fitness value among all particles, Indicates rounding up.

[0029] In step 4, the LPSO algorithm is layered through a hierarchical learning strategy. The specific steps include:

[0030] Particle hierarchical division: Sort the particles from high to low according to their fitness values, and divide the sorted particles into L1 to L NL , where L1 is the highest level, L NL is the lowest level, each layer contains the same number of particles;

[0031] Hierarchical learning strategy: low-level particles (L3, L4) enhance their exploration capabilities by learning from excellent particles from higher levels (such as L1, L2); high-level particles increase their development capabilities by learning from excellent particles from lower levels;

[0032] For example, if the total number of particles is 100, divided into 4 layers, each layer has 25 particles. L1 and L2 are higher layers (higher fitness values), and L3 and L4 are lower layers (lower fitness values).

[0033] Through this hierarchical strategy, the particle swarm optimization algorithm can balance local search and global exploration capabilities and improve the overall optimization performance.

[0034] Step 5 includes: and Randomly select particles and As example particles, they are used to update the individual optimal pbest and the neighborhood optimal nbest to guide the learning of low-level particles;

[0035] The higher-level determination is based on the fitness ranking of particles. It represents a level randomly selected from the rl1-th higher level in the particle swarm optimization algorithm LPSO based on hierarchical learning. It contains more than two particles, and the position and velocity of each particle are represented by and respectively; where i represents the number of the particle in the population (the i-th particle), and j represents the dimension of the problem space (the j-th dimension). represents the current position of the i-th particle in the j-th dimensional solution space in represents the moving velocity of the i-th particle in the j-th dimensional space in It represents a level randomly selected from the rl2-th higher level in the particle swarm optimization algorithm LPSO based on hierarchical learning. It contains more than two particles, and the position and velocity of each particle are represented by and respectively; represents the current position of particle i in the j-th dimensional solution space in represents the moving velocity of the i-th particle in the j-th dimensional space in

[0036] represents the k1-th particle randomly selected from ; represents the k2-th particle randomly selected from ; where rl1 < rl2 < i, and k1 and k2 are randomly selected particle indices; the better exemplary particles are used to guide the particles to approach the excellent solution region and enhance the development ability; the relatively worse exemplary particles are used to increase the randomness and diversity of the particles and improve the exploration ability.

[0037] In step 6, the parameters include the number of particles NP, the number of levels NL, the level size LS, and the control parameter φ. Generally, NP = 20, NL = 4, LS = 5, and φ = 0.4 are set.

[0038] In step 7, the inertia weight of the particle swarm optimization algorithm LPSO based on hierarchical learning is adaptively adjusted by the following formula:

[0039]

[0040] where, represents the inertia weight when the particle iterates d times, ω min and ωmax respectively represent the pre - given minimum inertia coefficient and maximum inertia coefficient, generally taking 0.4 and 0.9; represents the fitness of the particle at the d - th iteration, is the average fitness of all particles at the d - th iteration, represents the maximum fitness of all particles at the d - th iteration.

[0041] Step 8 includes: updating the particle velocity and position using the following formula:

[0042]

[0043] where, r1 and r2 are random numbers between 0 and 1, is the particle the updated velocity after d - th iteration of the particle, is the particle X i,j the updated position after d - th iteration; represents the position vector of the k1 - th particle randomly selected from the rl1 - th level after d - th iteration, where rl1 is the index of a randomly selected higher level (rl1 < i, that is, this level is better than the i - th level where the current particle is located), and k1 is the index of a particle randomly selected from this level rl1; represents the position vector of the k2 - th particle randomly selected from the rl2 - th level after d - th iteration, where rl2 is the index of a randomly selected higher level (rl2 < i and rl1 ≠ rl2), and k2 is the index of a particle randomly selected from this level rl2, where rl1 < rl2 < i, and k1 and k2 are randomly selected particle indices, and are two exemplars for the current example learning, coming from two different higher levels respectively; represents the position vector of the k1 - th particle randomly selected from the rl1 - th level, represents the position vector of the k2 - th particle randomly selected from the rl2 - th level, the fitness value of is greater than or equal to the fitness value of is used to enhance the exploitation ability of the particle, is used to enhance the exploration ability of the particle;

[0044] φ is the control parameter of hierarchical learning, between 0 and 1, which affects the second exemplar

[0045] Step 9 includes: During the iteration process of the LPSO algorithm, the following steps are repeated until the termination condition is met:

[0046] Evaluate particle fitness: For each particle in the current population, the fitness function is used to calculate the fitness value. The fitness function comprehensively considers multiple performance indicators, such as the output voltage tracking speed indicator, overshoot, and static error, and obtains the fitness value of each particle by weighted summation through weight coefficients.

[0047] Particle stratification: Particles are assigned to different levels based on their fitness values; particles with higher fitness values ​​are assigned to higher levels (such as L1), and particles with lower fitness values ​​are assigned to lower levels (such as LNL). The purpose of stratification is to apply differentiated learning strategies at different levels to improve the algorithm's search efficiency and solution quality;

[0048] Example selection and learning: Two particles are randomly selected from higher levels (L1 and L2) as example particles, which are used to update the individual optimal solution pbest and the neighborhood optimal solution nbest, respectively. The better example particles are used to guide particles toward excellent solution regions, enhancing their exploration capabilities. The relatively poorer example particles are used to increase the randomness and diversity of particles, improving their exploration capabilities. In this way, the algorithm is able to strike a balance between global and local search.

[0049] Updating particle velocity and position: Based on the example particles and control parameters, each particle's velocity and position are updated. The velocity update formula takes into account the particle's current velocity, the velocity increment learned from the best example particles, and the velocity increment learned from the worst example particles. The position update formula adjusts the particle's position based on the updated velocity. By continuously updating the particle's velocity and position, the algorithm gradually approaches the optimal solution in the search space.

[0050] Check the termination condition: After each iteration, check whether the termination condition has been met. The termination condition can be reaching a preset maximum number of iterations. If the termination condition is met, the iteration process ends; otherwise, the next iteration continues.

[0051] Step 10 includes:

[0052] Determine the optimal solution: During the iteration process of the LPSO algorithm, record the particle with the best fitness value and the corresponding parameters; the best fitness value indicates that the particle has the best performance after comprehensively considering multiple performance indicators.

[0053] Parameter application: The parameters corresponding to the particles with the best fitness are extracted and used as the optimized parameters of the energy storage converter controller. These parameters are applied to the actual energy storage converter system to improve system performance, stability, response speed, and overall efficiency.

[0054] The present invention has the following beneficial effects: the LPSO algorithm significantly improves the global search capability and local development capability of the algorithm through a hierarchical learning strategy, and can find higher quality solutions in complex optimization problems.

[0055] The dynamic sample selection mechanism of the LPSO algorithm increases the randomness and diversity of the algorithm, helps the algorithm escape from local optimality, and improves the global search capability.

[0056] The adaptive inertia weight strategy of the LPSO algorithm dynamically adjusts the inertia weight according to the performance of the particles, enabling the algorithm to better adapt to different search stages.

[0057] The LPSO algorithm demonstrates higher convergence speed and accuracy in the optimization of energy storage converter control systems. It can be adjusted according to the characteristics of specific problems, improving the versatility and adaptability of the algorithm and significantly improving the stability and response speed of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 This is an optimization flow chart of the energy storage converter control system in the present invention.

[0059] Figure 2 It is a control system optimization diagram of the energy storage converter in the present invention.

[0060] Figure 3 This is a comparison chart of output voltage tracking under different algorithms. DETAILED DESCRIPTION

[0061] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, and the above and / or other advantages of the present invention will become more apparent.

[0062] like Figure 1 As shown, this embodiment provides an energy storage converter control system optimization method based on hierarchical learning, comprising the following steps:

[0063] Step 1: Collect historical power grid data, including fitting various combinations of energy storage converter controller parameters kp1, ki1, kp2, ki2, kp3, and ki3, and the corresponding performance indicators time (st), overshoot (ov), and steady-state error (sse). The selection of kp1, ki1, kp2, ki2, kp3, and ki3 is random, but the approximate range of these values ​​is [0.0001, 50].

[0064] Step 2: Preprocess the data. Preprocessing operations include data cleaning, eliminating unreasonable data, reordering the data, and normalizing the data.

[0065] Step 3: Design the fitness function of the LPSO algorithm. The fitness function includes multiple performance indicators. The detailed steps of this example are as follows:

[0066] Step 3.1, according to the experimental requirements, the parameters include fitting various combinations of energy storage converter controller parameters kp1, ki1, kp2, ki2, kp3, ki3, and the corresponding performance indicators time (st), overshoot (ov), and steady-state error (sse).

[0067] The specific adaptation fitness function of step 3.2 is as follows:

[0068]

[0069] Among them, k1 and k2 are the weight coefficients of the performance indicator st, k3 and k4 are the weight coefficients of the performance indicator ov, and k5 and k6 are the weight coefficients of the performance indicator sse, among which k1 = 50, k2 = 0.1, k3 = 50, k4 = 70, k5 = 50, k6 = 550.

[0070] Step 4: Determine the fitness of the particles according to the fitness function, and assign the particles to different levels according to their fitness values. Assume that there are NP particles divided into NL levels, and each level contains the same number of particles, which is called the "level size" (LS). The calculation formula can be defined as:

[0071]

[0072] Among them, NP=20, NL=4, LS=5.

[0073] Step 5: Particles are divided into different levels according to their hierarchical learning optimal strategies. Particles at different levels are treated differently by learning different numbers of particles at higher levels. Specifically, the lower the level a particle belongs to, the more candidate examples ( and ), and vice versa, in this way, particles in lower levels can focus on exploring the search space, while particles in higher levels can focus on exploring the search space.

[0074] Step 6: Select excellent particles from higher levels as samples to guide the lower-level particles. The existing two excellent particles serve as examples for guiding the learning of inferior particles, which is beneficial to enhancing the exploration ability. Different from pbest, nbest, or gbest learning in most PSO algorithm variants, the LPSO algorithm directly uses two high-quality particles randomly selected from two different higher levels to guide particle learning. The two selected exemplary particles are respectively used to replace pbest (personal best) and nbest (neighborhood best) in the particle swarm optimization (PSO) algorithm. Specifically, among the two particles randomly selected from the higher level, the better one is used to update pbest, while the relatively worse one is used to update nbest or gbest;

[0075] Step 7: Optimize the LPSO algorithm using MATLAB code. The particle velocity and position update formulas are as follows:

[0076]

[0077] where r1, r1, and r3 are random numbers between 0 and 1, is the velocity of particle in dimension d, is the position of particle X i,j updated after d iterations, represents the position vector of the k1-th particle randomly selected from the rl1-th level after d iterations, where rl1 is the index of a randomly selected higher level (rl1 < i, that is, this level is better than the i-th level where the current particle is located), and k1 is the index of a particle randomly selected from this level rl1; represents the position vector of the k2-th particle randomly selected from the rl2-th level after d iterations, where rl2 is the index of a randomly selected higher level (rl2 < i and rl1 ≠ rl2), and k2 is the index of a particle randomly selected from this level rl2, where rl1 < rl2 < i, and k1 and k2 are randomly selected particle indices, and are two exemplars for the current example learning, respectively from two different higher levels; represents the position vector of the k1-th particle randomly selected from the rl1-th level, represents the position vector of the k2-th particle randomly selected from the rl2-th level, The fitness value of is greater than or equal to The fitness value of is used to enhance the exploitation ability of particles, is used to enhance the exploration ability of particles;

[0078] φ is the control parameter of hierarchical learning, which is between 0 and 1 and affects the second example

[0079] The formulas for updating particle velocities and positions do not directly apply to updating particles in the first and second levels. To handle this, we employ different additional techniques for the two levels. First, since the particles in the first level are the best particles in the entire population of the current generation, and better solutions are often found near these particles, we simply keep these particles unchanged to preserve the most useful information and protect them from being weakened. Therefore, particles in the first level are carried directly into the next generation. Second, similar adaptations are applied to particles in the second level. Instead of randomly selecting two samples from two randomly selected higher levels, two samples of these particles are randomly selected from the first level.

[0080] Step 8: According to the requirements of step 7, set d = 6, φ = 0.4, and set other parameters: number of particles N = 20, number of iterations G = 1000.

[0081] Step 9: Evaluate the fitness of each particle and update the global optimal position.

[0082] Step 10, termination condition judgment: reaching the set maximum number of evolutionary iterations.

[0083] Step 11: If the convergence conditions of the LPSO algorithm are met, the optimal solution, i.e., the optimal controller, is output. If the convergence conditions of the algorithm are not met, the algorithm returns to step 7, updates the velocity and position of the particle, and continues to iterate.

[0084] Step 12: The LPSO algorithm optimization program is completed and the optimal parameter combination is kp 1best =0.0001, ki 1best =7.7422, kp 2best =29.7730, ki 2best =22.6126, kp 3best =25.0926, ki 3best =50;

[0085] Step 13: Bring the optimal parameter combination obtained in step 12 into the system and compare it with the PSO algorithm, and draw a comparison chart of each performance indicator under different methods, such as Figure 3 shown.

[0086] like Figure 2 As shown, this embodiment provides an optimization diagram of an energy storage converter control system based on hierarchical learning. The core control structure of the system includes:

[0087] Outer loop DC voltage control: By comparing the actual DC voltage U dc and target voltage The target current is generated by the PI regulator It is further used as the input of the inner loop current control.

[0088] Inner loop current control: using target current and (in ) and the actual current i d and i q The error of f is used as input to generate the target voltage through an independent PI controller and

[0089] Coordinate transformation module:

[0090] Park transformation and Clarke transformation: transform the three-phase current i a 、i b 、i c Converted into DC component i d and i q , thus achieving more efficient control.

[0091] Inverse conversion module: convert the target voltage and Converted to three-phase target voltage U a 、U b 、U c .

[0092] SPWM modulation module: Generates three-phase PWM signals through space vector pulse width modulation (SPWM) to drive the switching devices of the energy storage converter.

[0093] Intelligent optimization algorithm module: uses a hierarchical learning-based optimization algorithm (LPSO) to dynamically adjust PI controller parameters to further improve system performance and adapt to complex operating environments.

[0094] like Figure 3 As shown in Figure 1, this embodiment provides a comparison of output voltage tracking using different algorithms. As can be seen from the figure, the LPSO algorithm is able to approach the steady-state output value relatively quickly in the initial stage. This rapid response is due to the LPSO algorithm's hierarchical learning mechanism, which enables more efficient exploration and utilization of the search space, resulting in a relatively optimal solution in a short period of time.

[0095] During the steady-state phase, the LPSO algorithm's output voltage curve is almost identical to the target value (dashed green line). This demonstrates that the LPSO algorithm tracks the target output with greater accuracy and is capable of achieving optimization with minimal error. Although the LPSO algorithm curve exhibits a slight overshoot upon reaching the steady-state value, this overshoot is small and decays rapidly, demonstrating its excellent dynamic response. Compared to the particle swarm optimization (PSO) algorithm, this moderate overshoot allows for rapid target attainment while avoiding excessive fluctuations. The LPSO algorithm, based on hierarchical learning, can dynamically adjust the swarm's learning model during the optimization process, effectively balancing global and local search. This characteristic enables the LPSO algorithm to achieve exceptional performance in complex optimization problems.

[0096] This embodiment also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for optimizing an energy storage converter control system based on hierarchical learning is implemented.

[0097] This embodiment also provides a computer device, including:

[0098] Memory, used to store instructions.

[0099] The processor is configured to execute the instructions so that the computer device performs the operations of the energy storage converter control system optimization method based on hierarchical learning.

[0100] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0101] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0102] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0103] The present invention provides a method for optimizing an energy storage converter control system based on hierarchical learning. While there are numerous methods and approaches for implementing this technical solution, the aforementioned are merely preferred embodiments of the present invention. It should be noted that those skilled in the art may make various improvements and modifications without departing from the principles of the present invention, and such improvements and modifications are also within the scope of protection of the present invention. Any components not specified in this embodiment may be implemented using existing technologies.

Claims

1. A method for optimizing an energy storage converter control system based on hierarchical learning, characterized in that: The following steps are involved: Step 1: Collect historical power grid data and preprocess it; Step 2, by using the fitness function to calculate the fitness value and evaluate the fitness value of each particle to measure the performance of each particle; Step 3: Particles are divided into different levels according to their fitness. Particles are divided into low-level particles and high-level particles according to their fitness, forming a hierarchical learning structure, in which low-level particles focus on exploring new parameter spaces, and high-level particles focus on developing within known excellent parameter areas; Step 4: Through a hierarchical learning strategy, low-level particles enhance their exploration capabilities by learning from excellent particles at higher levels; high-level particles enhance their development capabilities by learning from excellent particles at lower levels. Step 5: Randomly select two particles from the high level as samples, which are used to update the individual optimal pbest and the neighborhood optimal nbest respectively to guide the learning of the low level particles; Step 6, setting the parameters of the particle swarm optimization algorithm LPSO based on hierarchical learning; Step 7, adjusting the inertia weight of the particle swarm optimization algorithm based on hierarchical learning LPSO through self-adaptation; Step 8: Update the particle position and adjust the controller parameters according to the system feedback; Step 9: Repeat evaluation, stratification, learning, and updating until the optimal solution is found. Step 10: Apply the parameters corresponding to the particle with the best fitness to the energy storage converter system.

2. The method according to claim 1, characterized in that Step 1 includes: the power grid historical data includes the voltage V(t) at time t, the current I(t) at time t, and the power factor cos(φ).

3. The method according to claim 2, characterized in that Step 2 includes: calculating the fitness F using the following formula: F=ω1×f1+ω2×f2+ω3×f3, Among them, ω1, ω2, and ω3 are weight coefficients, f1 is the tracking speed index of the output voltage; f2 is the output voltage overshoot; and f3 is the output voltage static error.

4. The method according to claim 3, characterized in that Step 3 includes: in the particle swarm optimization algorithm based on hierarchical learning LPSO, it is assumed that there are NP particles divided into NL levels, each level contains the same number of particles, called the level size LS, and the calculation formula is: The fitness value of the particle F(X i ) determines the position of the particle in the hierarchy, the i-th particle X i The mathematical expression assigned to the lth layer is: Among them, rank(F(X i )) is the i-th particle X i The ranking of the fitness value among all particles, Indicates rounding up.

5. The method according to claim 4, characterized in that In step 4, the LPSO algorithm is layered through a hierarchical learning strategy. The specific steps include: Particle hierarchical division: Sort the particles from high to low according to their fitness values, and divide the sorted particles into L1 to L NL , where L1 is the highest level, L NL is the lowest level, each layer contains the same number of particles; Hierarchical learning strategy: low-level particles enhance their exploration capabilities by learning from excellent particles at higher levels; high-level particles enhance their development capabilities by learning from excellent particles at lower levels.

6. The method according to claim 5, characterized in that Step 5 includes: and Randomly select particles and As example particles, they are used to update the individual optimal pbest and the neighborhood optimal nbest to guide the learning of lower-level particles; The higher level decision is based on the fitness ranking of particles. represents a level randomly selected from the rl1th higher level in the particle swarm optimization algorithm based on hierarchical learning LPSO, There are more than two particles, and the position and velocity of each particle are expressed as and Represents, where i represents the number of particles in the group, j represents the dimension of the problem space, express The current position of the i-th particle in the j-th dimension of the solution space, express The moving speed of the i-th particle in the j-th dimension space; represents a level randomly selected from the rl2th higher level in the Particle Swarm Optimization algorithm based on hierarchical learning LPSO, There are more than two particles, and the position and velocity of each particle are expressed as and express, express The current position of particle i in the j-th dimension of the solution space, express The moving speed of the i-th particle in the j-th dimension space; denotes the k1-th particle randomly selected from ; denotes the k2-th particle randomly selected from ; where rl1 < rl2 < i, and k1 and k2 are the indices of randomly selected particles; the selected from the higher-level is a relatively better example particle, which is used to guide the particles to approach the excellent solution region; the selected from the at a lower level is a relatively worse example particle, which is used to increase the randomness and diversity of the particles.

7. The method according to claim 6, characterized in that In step 6, the parameters include the number of particles NP, the number of layers NL, the layer size LS and the control parameter φ.

8. The method according to claim 7, characterized in that In step 7, the inertia weight of the particle swarm optimization algorithm based on hierarchical learning LPSO is adaptively adjusted by the following formula: in, Represents particles Inertia weight at iteration d, ω min and ω max Respectively represent the predetermined minimum and maximum inertia coefficients; represents the fitness of the particle at the dth iteration, is the average fitness of all particles at the dth iteration, It represents the maximum fitness of all particles at the dth iteration.

9. The method according to claim 8, characterized in that Step 8 includes: updating the particle velocity and position using the following formula: where r1 and r2 are random numbers between 0 and 1, is the velocity updated after d iterations of the particle i,j is the position of particle X i,j updated after d iterations; represents the position vector of the k1-th particle randomly selected from the rl1-th level after d iterations, where rl1 is the index of a randomly selected higher level, rl1 < i, and k1 is the index of a particle randomly selected from that level rl1; represents the position vector of the k2-th particle randomly selected from the rl2-th level after d iterations, where rl2 is the index of a randomly selected higher level, rl2 < i, and rl1 ≠ rl2, and k2 is the index of a particle randomly selected from that level rl2, where rl1 < rl2 < i, and k1 and k2 are randomly selected particle indices, and are two exemplars for the current example learning, coming from two different higher levels respectively; represents the position vector of the k1-th particle randomly selected from the rl1-th level, represents the position vector of the k2-th particle randomly selected from the rl2-th level, the fitness value of is greater than or equal to the fitness value of is used to enhance the development ability of the particle, is used to enhance the exploration ability of the particle; φ is the control parameter for hierarchical learning.

10. The method according to claim 9, characterized in that Step 9 includes: During the iteration process of the LPSO algorithm, the following steps are repeated until the termination condition is met: Evaluate particle fitness: For each particle in the current population, use the fitness function to calculate the fitness value; Particle stratification: particles are assigned to different levels according to their fitness values; particles with higher fitness values ​​are assigned to higher levels, and particles with lower fitness values ​​are assigned to lower levels; Example selection and learning: Two particles are randomly selected from a higher level as example particles, which are used to update the individual optimal pbest and the neighborhood optimal nbest respectively; the better example particles are used to guide particles to approach the excellent solution area and enhance the development ability; the relatively poor example particles are used to increase the randomness and diversity of particles and improve the exploration ability; Update particle speed and position: Update the speed and position of each particle based on the sample particle and control parameters; Check the termination condition: After each iteration, check whether the termination condition is met. If the termination condition is met, the iteration process ends; otherwise, continue to the next round of iteration; Step 10 includes: Determine the optimal solution: During the iteration process of the LPSO algorithm, record the particles with the best fitness value and the corresponding parameters; Parameter application: The parameters corresponding to the particles with the best fitness are extracted and used as the optimized energy storage converter controller parameters.

Images