A method and system for optimizing microgrid capacity configuration

Abstract

The present invention relates to the technical field of microgrid capacity configuration, and in particular to a method and system for optimizing the capacity configuration of a microgrid. The present invention constructs a total cost objective function with the goal of minimizing the total cost of all entities in a wind, solar and storage microgrid, constructs a wind and solar abandonment objective function with the goal of minimizing the penalty cost for abandoning wind and solar power, and constructs an ACG wind and solar frequency regulation objective function with the goal of maximizing the ACG wind and solar frequency regulation revenue. Power balance constraints and wind and solar abandonment constraints are set as constraints to establish a microgrid capacity non-cooperative game model; based on the microgrid capacity non-cooperative game model and its constraints, an improved particle swarm algorithm is used to solve the microgrid capacity non-cooperative game model to obtain the capacity target configuration result. The present invention optimizes the power output of each entity by considering the wind and solar abandonment factors, achieves the Nash equilibrium of the system, maximizes the system revenue, and effectively improves the overall performance and economy of the microgrid system.

Description

Technical Field

[0001] The present invention relates to the technical field of microgrid capacity configuration, and in particular to a microgrid capacity configuration optimization method and system. Background Art

[0002] Microgrids, as small energy systems that can integrate multiple distributed energy sources and flexibly realize island and grid-connected operation, play a vital role in the transformation of energy structure. With the popularization of renewable energy (such as wind power and photovoltaics), how to efficiently connect such energy to microgrids and maintain stable operation is the current research focus. However, due to the randomness and intermittency of wind and photovoltaic power generation, the operation of microgrids is very challenging. Currently, capacity configuration is mainly achieved through two types of optimization algorithms: traditional optimization methods and intelligent optimization algorithms.

[0003] Traditional optimization algorithms such as linear programming and dynamic programming are the earliest widely used methods. These algorithms perform well when dealing with small-scale microgrid systems and can provide relatively accurate solutions. However, when facing large-scale and complex microgrids, the computational efficiency and solution accuracy of traditional optimization algorithms often cannot meet the requirements. In the technical solution of microgrid capacity configuration, the economy, reliability and renewable energy utilization rate of the microgrid system are usually converted into a single objective function through weighting. When solving a single objective function, the intelligent optimization algorithm can use the global search mechanism and quickly locate the optimal solution area through information sharing between individuals. Compared with other intelligent optimization algorithms, the particle swarm algorithm can effectively explore the solution space, avoid local optimality, and improve the global search ability through the cooperation and competition of particles in the group. The parallel computing capability of the particle swarm algorithm makes it superior in dealing with large-scale problems such as distribution network capacity configuration optimization, and can significantly shorten the solution time. Therefore, the application of the particle swarm algorithm in microgrid capacity configuration can provide an efficient and reliable solution, providing strong support for optimizing the operation and management of microgrids.

[0004] However, the distribution network capacity configuration optimization model integrating a single objective function often ignores the characteristics and requirements of different modules, resulting in a lack of flexibility in the overall optimization of the system. When the requirements of a certain module change, adjusting the overall objective function may cause the performance of other modules to deteriorate, making it difficult to adapt to the rapidly changing environment. However, when converted to multiple objective functions, the particle swarm algorithm needs to effectively balance the relationship between different objectives. The introduction of multiple objective functions significantly increases the complexity of the optimization problem, making it more difficult to achieve a balance between exploration (discovering new solutions) and utilization (optimizing existing solutions). When the particle swarm algorithm is dealing with multi-objective optimization, information sharing between particles may lead to insufficient search in the target space, making it easier to fall into a local optimal solution, making the microgrid capacity configuration strategy less than ideal. This is because when looking for the best balance between different objectives, particles may over-rely on existing better solutions, thereby ignoring potential better solutions. In addition, the trade-off between multiple objectives requires more computing resources and time, which also increases the complexity of the algorithm. In addition, the existing microgrid capacity configuration optimization model ignores the frequency regulation benefits of wind and solar power, but tends to pursue a stable output of power generation. The inability to flexibly adjust the frequency regulation capabilities of wind and solar power generation to cope with frequency fluctuations in the power grid will lead to insufficient stability and flexibility of the system, and will cause inefficient use of renewable energy, thereby increasing wind and solar power abandonment and causing energy waste, which will in turn lead to the inability of capacity configuration plans to maximize benefits, thereby reducing the overall economy of the system. Summary of the invention

[0005] To this end, the technical problem to be solved by the present invention is to overcome the defect that the existing intelligent optimization algorithm is prone to fall into the local optimal solution when facing multiple objective functions, resulting in insufficient accuracy of the optimal solution for capacity configuration, and the existing capacity configuration optimization method ignores the frequency regulation benefits of wind and solar power, which results in the capacity configuration scheme being unable to achieve maximum benefit.

[0006] In order to solve the above technical problems, the present invention provides a microgrid capacity configuration optimization method, comprising the following steps:

[0007] Obtain the power model of each subject in the wind-solar-storage microgrid in the area to be optimized, and the AGC frequency regulation benefit model of the wind power generation unit and the photovoltaic power generation unit. Construct a total cost objective function with the goal of minimizing the total cost of all subjects in the wind-solar-storage microgrid, construct a wind and solar abandonment objective function with the goal of minimizing the penalty cost of abandoning wind and solar power, and construct an ACG wind and solar frequency regulation objective function with the goal of maximizing the ACG wind and solar frequency regulation benefit. Set power balance constraints and wind and solar abandonment constraints as constraints, and establish a microgrid capacity non-cooperative game model; among which, the subjects of the wind-solar-storage microgrid include wind power generation units, photovoltaic power generation units and energy storage systems;

[0008] The parameter data of the time period to be optimized are obtained and input into the non-cooperative game model of microgrid capacity. The improved particle swarm algorithm is used to solve the non-cooperative game model of microgrid capacity to obtain the capacity target configuration result.

[0009] Preferably, the method of obtaining the power model of each subject in the wind-solar-storage microgrid in the area to be optimized, and the AGC frequency regulation benefit model of the wind power generation unit and the photovoltaic power generation unit includes:

[0010] The wind power generation model of the wind power generation unit is:

[0011] ,

[0012] in, for The wind power generation capacity in the period, is the rated power of the wind power generation unit, is the cut-in wind speed of the wind power generation unit, is the rated wind speed of the wind power generation unit, is the cut-out wind speed of the wind power generation unit, For wind power generation units Wind speed during the time period;

[0013] The photovoltaic power generation model of the photovoltaic power generation unit is:

[0014] ,

[0015] in, for Photovoltaic power generation during the period, is the rated power of the PV module, is the temperature coefficient, for The temperature of the time period, is the rated temperature of the PV module;

[0016] The energy storage system power model of the energy storage system is:

[0017] ,

[0018] ,

[0019] in, for The output power of the energy storage system during the period, is the charging power of the energy storage system, represents the discharge power of the energy storage system, for The net power of the microgrid during the period, for The wind power generation capacity in the period, for The photovoltaic power generation during the period, for The load of the microgrid during the period;

[0020] The AGC frequency regulation revenue model of wind power generation units is:

[0021] ,

[0022] in, for Wind power frequency regulation benefits during the period, is the wind power frequency regulation benefit per unit time, is the regulation coefficient of wind power frequency regulation, for The total number of wind power frequency modulations during the period, for During the period The depth of regulation of secondary wind power frequency regulation;

[0023] The AGC frequency regulation benefit model of the photovoltaic power generation unit is:

[0024] ,

[0025] in, for The photovoltaic frequency modulation income of the time period, is the photoelectric frequency modulation gain per unit time, is the adjustment coefficient of photoelectric frequency modulation, for The total number of photoelectric frequency modulations in the time period, for During the period The adjustment depth of sub-photoelectric frequency modulation.

[0026] Preferably, the non-cooperative game model of microgrid capacity is:

[0027] ,

[0028] ,

[0029] ,

[0030] Power balance constraints:

[0031] ,

[0032] Wind and solar curtailment constraints:

[0033] ,

[0034] in, is the total cost objective function, is the wind and solar curtailment objective function, is the ACG wind and solar frequency modulation objective function, for Time period The electricity purchase cost of each entity, for Time period The operation and maintenance costs of each entity, for Time period The energy storage system cost of each entity, for The penalty fees for abandoning wind and solar power during the period, for The penalty fees incurred by all entities for abandoning wind and solar power during the period, Indicates the time interval, is the total duration, As a penalty factor for abandoning wind and solar power, for The photovoltaic frequency modulation income of the time period, for Wind power frequency regulation benefits during the period, for The power purchased by the power grid during the period, for The wind power generation capacity in the period, for Photovoltaic power generation during the period, for The output power of the energy storage system during the period, for The load of the microgrid during the period, for The power abandoned by wind energy during the period, for The power abandoned by photovoltaic power during the period, is the maximum power of wind power generation, is the maximum power of photovoltaic power generation.

[0035] Preferably, the improved particle swarm algorithm is used to solve the microgrid capacity non-cooperative game model to obtain the capacity target configuration result, including:

[0036] Step S21: Set the particle swarm size, particle dimension, maximum number of iterations, initial weight, end weight, individual learning factor, group learning factor, and external archive size; Initialize , the Sobol sequence is used to generate the initial position of each particle, and the initial position of each particle is updated by the mean shift clustering algorithm to obtain the updated initial position of each particle; the initial velocity of each particle is randomly generated, where, Indicates the current iteration number. The updated initial position of each particle represents a capacity target configuration result.

[0037] Step S22: the positions of each particle after each iteration are externally archived. If the particle positions stored in the external archive exceed the external archive size, the particle positions in the external archive are deleted based on the objective function values ​​corresponding to the positions of each particle.

[0038] Step S23: using a fast stratification strategy to select a particle position in the external archive as the global optimal position of the current iteration;

[0039] Step S24: based on the objective function value corresponding to each historical iteration position of each particle, obtain the Pareto optimal solution in each historical iteration position of each particle as the individual learning position of each particle in the current iteration;

[0040] Step S25: Calculate the speed inertia weight of the current iteration based on the initial weight, the end weight, the current iteration number and the maximum iteration number; Update the speed of each particle based on the individual learning factor, the group learning factor, the current iteration number, the speed inertia weight of the current iteration, the global best position of the current iteration and the individual learning position of each particle in the current iteration;

[0041] Step S26: Based on the objective function value corresponding to the current iteration position of each particle and the global optimal position of the current iteration, the position inertia weight of each particle in the current iteration is calculated; based on the individual learning factor, the group learning factor, the position inertia weight of each particle in the current iteration, the current iteration number, the global optimal position of the current iteration and the individual learning position of each particle in the current iteration, the position of each particle is updated. ;

[0042] Step S27: Determine the current number of iterations Whether the maximum number of iterations is reached, if so, the global optimal position of the current iteration is output as the capacity target configuration result, otherwise, steps S22-S27 are repeated.

[0043] Preferably, the method of using the Sobol sequence to generate the initial position of each particle, and updating the initial position of each particle by a mean shift clustering algorithm to obtain the updated initial position of each particle comprises:

[0044] The Sobol sequence is used to generate Q Sobol sequence sample points, each of which represents the initial position of a particle, where Q is the size of the particle group;

[0045] The mean shift value of each Sobol sequence sample point is calculated by the mean shift clustering algorithm;

[0046] Based on the mean offset value of each Sobol sequence sample point, each Sobol sequence sample point is updated to obtain the updated initial position of each particle.

[0047] Preferably, the speed inertia weight of the current iteration is calculated based on the initial weight, the end weight, the current iteration number and the maximum iteration number, and the calculation formula is:

[0048] ,

[0049] in, For the The velocity inertia weight of the iteration, For the ending weight, is the initial weight, is the maximum number of iterations.

[0050] Preferably, the speed of each particle is updated based on the individual learning factor, the group learning factor, the current iteration number, the speed inertia weight of the current iteration, the global optimal position of the current iteration and the individual learning position of each particle in the current iteration, and the formula is:

[0051] ,

[0052] in, For the The iteration The speed of a particle, For the The iteration The speed of a particle, For the The iteration The position of a particle, For the The velocity inertia weight of the iteration, is the individual learning factor, is the group learning factor, is the current iteration number, is the maximum number of iterations, For the The iteration The individual learning positions of particles, For the The global best position of the iteration, is a random number in the interval [0, 1].

[0053] Preferably, the position inertia weight of each particle in the current iteration is calculated based on the objective function value corresponding to the current iteration position of each particle and the global optimal position of the current iteration, and the calculation formula is:

[0054] ,

[0055] in, For the The iteration The position inertia weight of each particle, is a weight distribution factor in the range [0, 1], For the In the iteration The target fitness of each particle, , For the Particle No. The total cost objective function value corresponding to the iteration position, For the Particle No. The wind and solar curtailment objective function value corresponding to the iteration position is: For the Particle No. The ACG wind and solar frequency modulation objective function value corresponding to the iteration position, To set a positive number, For all particles in The highest target fitness in iterations, For the The target fitness corresponding to the global best position of the iteration.

[0056] Preferably, the position of each particle is updated based on the individual learning factor, the group learning factor, the position inertia weight of each particle in the current iteration, the current iteration number, the global optimal position of the current iteration and the individual learning position of each particle in the current iteration, and the formula is:

[0057] ，

[0058] in, For the The iteration The position of a particle, For the The iteration The position of a particle, For the The iteration The speed of a particle, For the The iteration The position inertia weight of each particle.

[0059] The present invention also provides a microgrid capacity configuration optimization system, comprising:

[0060] The model building module is used to obtain the power model of each subject in the wind-solar-storage microgrid in the area to be optimized, the AGC frequency regulation benefit model of the wind power generation unit and the photovoltaic power generation unit, and to build a total cost objective function with the goal of minimizing the total cost of all subjects in the wind-solar-storage microgrid, to build a wind and photovoltaic abandonment objective function with the goal of minimizing the penalty cost of abandoning wind and photovoltaic, and to build an ACG wind and photovoltaic frequency regulation objective function with the goal of maximizing the ACG wind and photovoltaic frequency regulation benefit. The power balance constraint and wind and photovoltaic abandonment constraint are set as constraint conditions to establish a non-cooperative game model of microgrid capacity; among which, the subjects of the wind-solar-storage microgrid include wind power generation units, photovoltaic power generation units and energy storage systems;

[0061] The solving module is used to obtain the parameter data of the time period to be optimized, and input it into the microgrid capacity non-cooperative game model. The improved particle swarm algorithm is used to solve the microgrid capacity non-cooperative game model to obtain the capacity target configuration result.

[0062] The present invention discloses a method for optimizing the capacity configuration of a microgrid, which obtains the power model of each subject in a wind-solar-storage microgrid, the AGC frequency regulation benefit model of the wind power generation unit and the photovoltaic power generation unit, constructs a total cost objective function with the goal of minimizing the total cost of all subjects in the wind-solar-storage microgrid, constructs a wind and photovoltaic abandonment objective function with the goal of minimizing the penalty cost for abandoning wind and photovoltaic, constructs an ACG wind and photovoltaic frequency regulation objective function with the goal of maximizing the ACG wind and photovoltaic frequency regulation benefit, and establishes a non-cooperative game model for the capacity of the microgrid; it enables multiple independent subjects in the microgrid (such as wind power, photovoltaic and energy storage units) to maximize their own interests, minimize the penalty cost for abandoning wind and photovoltaic, and maximize the ACG wind and photovoltaic frequency regulation benefit. Maximizing the frequency regulation benefits constitutes a non-cooperative game environment. In the process of the system reaching Nash equilibrium, different power generation units can work in coordination to achieve the global optimal power output and capacity configuration, thereby improving the economic benefits and operational stability of the entire system. Power balance constraints and wind and solar power abandonment constraints are set as constraints for the microgrid capacity non-cooperative game model. The power balance constraint can ensure the supply and demand balance of the microgrid, avoid excess or shortage of electricity, and ensure the safe and stable operation of the system. The wind and solar power abandonment constraints can promote the effective use of renewable energy, reduce economic losses caused by waste of resources, improve the economy and environmental benefits of the overall system, and ensure the operating efficiency of the distribution network.

[0063] In addition, based on the non-cooperative game model of microgrid capacity and its constraints, the improved particle swarm algorithm is used to solve the non-cooperative game model of microgrid capacity, and the capacity target configuration result is obtained; the existing particle swarm algorithm position initialization method often uses random initialization or Sobol sequence to generate the initial particle position. Although the Sobol sequence has better uniform distribution than the random method, the points generated by the Sobol sequence evenly cover the space, but do not consider the high-density area of ​​the data space, resulting in the inability to effectively concentrate the initial position of the particle near the optimal area of ​​the problem solution space. Therefore, only using the Sobol sequence for initialization may cause the algorithm to converge slowly in complex and multi-peak optimization problems, and may fall into the local optimum. The final microgrid capacity configuration scheme is difficult to achieve the best in terms of economy and stability. Therefore, the present invention proposes an improved particle swarm algorithm that uses the Sobol sequence to generate the initial particle position to ensure uniform distribution in high-dimensional space and avoid the initial points being concentrated in a limited area; these points are further processed by mean shift clustering to concentrate the particles in the high-density area and improve the effectiveness of the initial position. In this way, mean shift clustering can guide the particles to the vicinity of a better solution space, thereby accelerating the convergence speed and improving the overall optimization effect of the configuration scheme. In addition, the weight selection method in the existing particle swarm algorithm usually adopts a fixed or simple linear decreasing method. This method uses a larger weight in the initial stage to enhance the global search capability, and gradually reduces the weight in the later iteration to increase the precision of the local search. However, in complex multi-objective optimization problems, it is difficult to accurately control the balance between global and local searches by only using this method, which can easily cause the algorithm to fall into the local optimum or the convergence speed is poor, so that the final microgrid capacity configuration scheme cannot reach the global optimum. Therefore, the present invention dynamically adjusts the position inertia weight and the speed inertia weight by the target fitness corresponding to the global optimal position and the highest target fitness of all particles at each iteration. The position and speed of each particle in the current iteration are updated by the position inertia weight and speed inertia weight corresponding to each iteration. This adaptive adjustment method can flexibly allocate global and local search capabilities at different iteration stages, avoid falling into the local optimum too early, and improve the convergence speed, thereby improving the accuracy of the solution of the microgrid capacity configuration scheme. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In order to make the content of the present invention more clearly understood, the present invention is further described in detail below according to specific embodiments of the present invention in conjunction with the accompanying drawings, wherein:

[0065] Figure 1 It is a flow chart of the steps of a microgrid capacity configuration optimization method of the present invention.

[0066] Figure 2 It is a system structure diagram of a microgrid capacity configuration optimization system of the present invention. DETAILED DESCRIPTION

[0067] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it, but the embodiments are not intended to limit the present invention.

[0068] Reference Figure 1 As shown, the first embodiment of the present invention provides a microgrid capacity configuration optimization method, comprising the following steps:

[0069] Step S1: Obtain the power model of each subject in the wind-solar-storage microgrid in the area to be optimized, and the AGC frequency regulation benefit model of the wind power generation unit and the photovoltaic power generation unit, construct a total cost objective function with the goal of minimizing the total cost of all subjects in the wind-solar-storage microgrid, construct a wind abandonment objective function with the goal of minimizing the penalty cost of abandoning wind and solar power, and construct an ACG wind-solar-storage frequency regulation objective function with the goal of maximizing the ACG wind-solar-storage frequency regulation benefit, set power balance constraints and wind abandonment constraints as constraints, and establish a microgrid capacity non-cooperative game model; wherein, the subjects of the wind-solar-storage microgrid include wind power generation units, photovoltaic power generation units and energy storage systems;

[0070] In this embodiment, preferably, the wind power generation unit is any one of a wind turbine generator set, a micro wind turbine generator, and an offshore wind turbine generator;

[0071] The photovoltaic power generation unit is any one of a rooftop photovoltaic system, a ground photovoltaic power station, and a building-integrated photovoltaic system;

[0072] The energy storage system is any one of a battery energy storage system, a supercapacitor energy storage system, and a flywheel energy storage system.

[0073] In this embodiment, preferably, the power model of each subject in the wind-solar-storage microgrid in the area to be optimized, and the AGC frequency regulation benefit model of the wind power generation unit and the photovoltaic power generation unit are obtained, including:

[0074] The wind power generation model of the wind power generation unit is:

[0075] ,

[0076] in, for The wind power generation capacity in the period, is the rated power of the wind power generation unit, is the cut-in wind speed of the wind power generation unit, is the rated wind speed of the wind power generation unit, is the cut-out wind speed of the wind power generation unit, For wind power generation units Wind speed during the time period;

[0077] The photovoltaic power generation model of the photovoltaic power generation unit is:

[0078] ,

[0079] in, for The photovoltaic power generation during the period, is the rated power of the PV module, is the temperature coefficient, for The temperature of the time period, is the rated temperature of the PV module;

[0080] The energy storage system power model of the energy storage system is:

[0081] ,

[0082] ,

[0083] in, for The output power of the energy storage system during the period, is the charging power of the energy storage system, represents the discharge power of the energy storage system, for The net power of the microgrid during the period, for The wind power generation capacity in the period, for The photovoltaic power generation during the period, for The load of the microgrid during the period;

[0084] The AGC frequency regulation revenue model of wind power generation units is:

[0085] ,

[0086] in, for Wind power frequency regulation benefits during the period, is the wind power frequency regulation benefit per unit time, is the regulation coefficient of wind power frequency regulation, for The total number of wind power frequency modulations during the period, for During the period The depth of regulation of secondary wind power frequency regulation;

[0087] The AGC frequency regulation benefit model of the photovoltaic power generation unit is:

[0088] ,

[0089] in, for The photovoltaic frequency modulation income of the time period, is the photoelectric frequency modulation gain per unit time, is the adjustment coefficient of photoelectric frequency modulation, for The total number of photoelectric frequency modulations in the time period, for During the period The adjustment depth of sub-photoelectric frequency modulation.

[0090] In this embodiment, preferably, the non-cooperative game model of microgrid capacity is:

[0091] ,

[0092] ,

[0093] ,

[0094] Power balance constraints:

[0095] ,

[0096] Wind and solar curtailment constraints:

[0097] ,

[0098] in, is the total cost objective function, is the wind and solar curtailment objective function, is the ACG wind and solar frequency modulation objective function, for Time period The electricity purchase cost of each entity, for Time period The operation and maintenance costs of each entity, for Time period The energy storage system cost of each entity, for The penalty fees for abandoning wind and solar power during the period, for The penalty fees incurred by all entities for abandoning wind and solar power during the period, Indicates the time interval, is the total duration, As a penalty factor for abandoning wind and solar power, for The photovoltaic frequency modulation income of the time period, for Wind power frequency regulation benefits during the period, for The power purchased by the power grid during the period, for The wind power generation capacity in the period, for Photovoltaic power generation during the period, for The output power of the energy storage system during the period, for The load of the microgrid during the period, for The power abandoned by wind energy during the period, for The power abandoned by photovoltaic power during the period, is the maximum power of wind power generation, is the maximum power of photovoltaic power generation.

[0099] The payoff function of each subject is:

[0100]

[0101] in, for Time period The payoff function of an agent is for Time period The electricity sales revenue of each entity, for Wind power frequency regulation benefits during the period, for Photovoltaic frequency modulation benefits during the period.

[0102] The present invention adopts a non-cooperative game model to simulate the multi-agent collaborative game process of wind power, photovoltaic and energy storage systems. In this model, each subject, as an independent game participant, optimizes its strategy by maximizing its own benefits. In the process of the system reaching Nash equilibrium, different power generation units can work in coordination to achieve the global optimal power output and capacity configuration, improve the economic benefits and operational stability of the entire system, and take into account the ACG wind and solar frequency regulation benefits and the penalty fees for wind and solar abandonment, encourage more efficient use of renewable energy, and avoid energy waste caused by excessive pursuit of benefits.

[0103] By rationally optimizing the configuration of wind power, photovoltaic and energy storage systems, the present invention reduces the volatility of the power generation system and effectively smoothes the power fluctuations through the dynamic scheduling of energy storage units. In particular, when the wind and solar power generation power is higher than the load demand, the energy storage system can be charged in time to avoid the waste of excess power, thereby effectively reducing the phenomenon of wind and solar power abandonment and improving the utilization rate of renewable energy.

[0104] Step S2: Obtain the parameter data of the time period to be optimized, and input them into the microgrid capacity non-cooperative game model, and use the improved particle swarm algorithm to solve the microgrid capacity non-cooperative game model to obtain the capacity target configuration result.

[0105] In this embodiment, the parameter data of the time period to be optimized include: the rated power of the wind power generation unit, the cut-in wind speed of the wind power generation unit, the rated wind speed of the wind power generation unit, the cut-out wind speed of the wind power generation unit, the wind speed of the wind power generation unit in the time period to be optimized, the rated power of the photovoltaic module, the temperature coefficient, the temperature in the time period to be optimized, the rated temperature of the photovoltaic module, the charging power of the energy storage system, the discharging power of the energy storage system, the net power of the microgrid in the time period to be optimized, the wind power generation power in the time period to be optimized, the photovoltaic power generation power in the time period to be optimized, and the load of the microgrid in the time period to be optimized.

[0106] In this embodiment, preferably, the improved particle swarm algorithm is used to solve the microgrid capacity non-cooperative game model to obtain the capacity target configuration result, including:

[0107] Step S21: Set the particle swarm size, particle dimension, maximum number of iterations, initial weight, end weight, individual learning factor, group learning factor, and external archive size; Initialize , the Sobol sequence is used to generate the initial position of each particle, and the initial position of each particle is updated by the mean shift clustering algorithm to obtain the updated initial position of each particle; the initial velocity of each particle is randomly generated, where, Indicates the current iteration number. The updated initial position of each particle represents a capacity target configuration result.

[0108] Step S22: the positions of each particle after each iteration are externally archived. If the particle positions stored in the external archive exceed the external archive size, the particle positions in the external archive are deleted based on the objective function values ​​corresponding to the positions of each particle.

[0109] Step S23: using a fast stratification strategy to select a particle position in the external archive as the global optimal position of the current iteration;

[0110] Step S24: based on the objective function value corresponding to each historical iteration position of each particle, obtain the Pareto optimal solution in each historical iteration position of each particle as the individual learning position of each particle in the current iteration;

[0111] Step S25: Calculate the speed inertia weight of the current iteration based on the initial weight, the end weight, the current iteration number and the maximum iteration number; Update the speed of each particle based on the individual learning factor, the group learning factor, the current iteration number, the speed inertia weight of the current iteration, the global best position of the current iteration and the individual learning position of each particle in the current iteration;

[0112] Step S26: Based on the objective function value corresponding to the current iteration position of each particle and the global optimal position of the current iteration, the position inertia weight of each particle in the current iteration is calculated; based on the individual learning factor, the group learning factor, the position inertia weight of each particle in the current iteration, the current iteration number, the global optimal position of the current iteration and the individual learning position of each particle in the current iteration, the position of each particle is updated. ;

[0113] Step S27: Determine the current number of iterations Whether the maximum number of iterations is reached, if so, the global optimal position of the current iteration is output as the capacity target configuration result, otherwise, steps S22-S27 are repeated.

[0114] In this embodiment, preferably, the Sobol sequence is used to generate the initial position of each particle, and the initial position of each particle is updated by a mean shift clustering algorithm to obtain the updated initial position of each particle, including:

[0115] The Sobol sequence is used to generate Q Sobol sequence sample points, each of which represents the initial position of a particle, where Q is the size of the particle group;

[0116] The mean shift value of each Sobol sequence sample point is calculated by the mean shift clustering algorithm;

[0117] Based on the mean offset value of each Sobol sequence sample point, each Sobol sequence sample point is updated to obtain the updated initial position of each particle.

[0118] The Sobol sequence is a type of low-discrepancy sequence. Its excellent uniformity makes it of great significance to generate samples in high-dimensional space. The distribution of the Sobol sequence is not only uniform, but also does not require the number of samples to be determined in advance or to be stored, and can generate an infinite number of samples as needed. Therefore, the Sobol sequence is widely used in the problem of high-dimensional space sampling.

[0119] The generation formula of Sobol sequence is as follows:

[0120]

[0121] in, Represents the index of the Sobol sequence sample point, is the index of the dimension, Indicates The Sobol sequence sample point is The value of the dimension, is the bit index in binary representation, For the The Sobol sequence sample point is Binary representation of bits.

[0122] Mean shift clustering is a density-based nonparametric clustering method. Its core idea is to determine the cluster center by finding high-density areas in the data space. For any Sobol sequence sample point, its mean shift can be updated by the following formula:

[0123]

[0124] in, is the mean shift value of the current Sobol sequence sample point, For the Sobol sequence sample points, is the current Sobol sequence sample point, is the kernel function, usually the Gaussian kernel function is selected. Indicates The center is The Sobol sequence sample point set, is the current Sobol sequence sample point and the The Euclidean distance between the sample points of a Sobol sequence.

[0125] The iterative process of mean shift continuously updates the position of the Sobol sequence sample points until it converges to the density maximum area, that is, the cluster center.

[0126] Existing particle swarm position initialization methods often use random initialization or Sobol sequence to generate initial particle positions in high-dimensional space. Although the Sobol sequence has better uniform distribution than the random method, there are still some shortcomings when using the Sobol sequence alone. First, although the points generated by the Sobol sequence evenly cover the space, they do not take into account the high-density areas of the data space, resulting in the inability to effectively concentrate the initial positions of the particles near the optimal area of ​​the problem solution space, increasing the number of optimization iterations and reducing the convergence speed and accuracy of the algorithm. Therefore, using only the Sobol sequence for initialization may cause the algorithm to converge slowly in complex, multi-peak optimization problems and may fall into local optimality. The resulting microgrid capacity configuration scheme is difficult to achieve optimality in terms of economy and stability.

[0127] The present invention overcomes these defects of the existing particle swarm position initialization method by combining the Sobol sequence with the mean shift clustering algorithm. Specifically, the Sobol sequence is first used to generate the initial position of each particle to ensure uniform distribution in the high-dimensional space and avoid the initial points being concentrated in a limited area. Then, these points are further processed by mean shift clustering to concentrate the particles in high-density areas and improve the effectiveness of the initial position. In this way, mean shift clustering can guide particles to the vicinity of a better solution space, thereby accelerating the convergence speed, reducing the number of iterations, and increasing the probability of finding the global optimal solution. In this way, the microgrid capacity configuration scheme finally solved can better balance the economy and stability of the system, and improve the overall optimization effect of the configuration scheme.

[0128] In this embodiment, preferably, the speed inertia weight of the current iteration is calculated based on the initial weight, the end weight, the current iteration number and the maximum iteration number, and the calculation formula is:

[0129] ,

[0130] in, For the The velocity inertia weight of the iteration, For the ending weight, is the initial weight, is the maximum number of iterations.

[0131] In this embodiment, preferably, the speed of each particle is updated based on the individual learning factor, the group learning factor, the current iteration number, the speed inertia weight of the current iteration, the global optimal position of the current iteration and the individual learning position of each particle in the current iteration, and the formula is:

[0132] ,

[0133] in, For the The iteration The speed of a particle, For the The iteration The speed of a particle, For the The iteration The position of a particle, For the The velocity inertia weight of the iteration, is the individual learning factor, is the group learning factor, is the current iteration number, is the maximum number of iterations, For the The iteration The individual learning positions of particles, For the The global best position of the iteration, is a random number in the interval [0, 1].

[0134] In this embodiment, preferably, the position inertia weight of each particle in the current iteration is calculated based on the objective function value corresponding to the current iteration position of each particle and the global optimal position of the current iteration, and the calculation formula is:

[0135] ,

[0136] in, For the The iteration The position inertia weight of each particle, is a weight distribution factor in the range [0, 1], For the In the iteration The target fitness of each particle, , For the Particle No. The total cost objective function value corresponding to the iteration position, For the Particle No. The wind and solar curtailment objective function value corresponding to the iteration position is: For the Particle No. The ACG wind and solar frequency modulation objective function value corresponding to the iteration position, To set a positive number, For all particles in The highest target fitness in iterations, For the The target fitness corresponding to the global best position of the iteration.

[0137] In this embodiment, preferably, the position of each particle is updated based on the individual learning factor, the group learning factor, the position inertia weight of each particle in the current iteration, the current number of iterations, the global optimal position of the current iteration and the individual learning position of each particle in the current iteration, and the formula is:

[0138] ，

[0139] in, For the The iteration The position of a particle, For the The iteration The position of a particle, For the The iteration The speed of a particle, For the The iteration The position inertia weight of each particle.

[0140] The weight selection method in the existing particle swarm algorithm usually adopts a fixed or simple linear decreasing method. This method uses a larger weight in the initial stage to enhance the global search capability, and gradually reduces the weight in the later iteration to increase the precision of the local search. However, in complex multi-objective optimization problems, it is difficult to accurately control the balance between global and local searches using only this method, which can easily cause the algorithm to fall into the local optimum or converge poorly, making the final microgrid capacity configuration plan unable to reach the global optimum.

[0141] Therefore, the present invention dynamically adjusts the position inertia weight and the speed inertia weight through the target fitness corresponding to the global optimal position and the highest target fitness of all particles at each iteration, and updates the position and speed of each particle in the current iteration through the position inertia weight and speed inertia weight corresponding to each iteration. This adaptive adjustment method can flexibly allocate global and local search capabilities at different iteration stages, avoid falling into local optimality too early, and improve the convergence speed, thereby improving the accuracy of the solution of the microgrid capacity configuration solution.

[0142] Based on the sum of the distances between each particle position and each objective function value of its adjacent particle position in the external archive and the absolute value of the difference between each particle position and each objective function value of its adjacent particle position, the particle positions in the external archive are deleted, including:

[0143] Calculate the sum of the distances between each particle position in the external archive and the objective function values ​​of its adjacent particle positions respectively, and sort all particle positions in the external archive in descending order according to the sum of the distances between each particle position in the external archive and the objective function values ​​of its adjacent particle positions, and delete the last particle positions after descending order in sequence until the external archive size is less than or equal to the preset external archive size;

[0144] Among them, if the sum of the distances between the multiple particle positions in the external archive and the objective function values ​​of their adjacent particle positions is equal, the absolute values ​​of the differences between the distances between the multiple particle positions and the objective function values ​​of their adjacent particle positions are calculated respectively, and according to the absolute values ​​of the differences between the distances between the multiple particle positions and the objective function values ​​of their adjacent particle positions, the multiple particle positions in the external archive are sorted in ascending order, and the last particle positions after the ascending sorting are deleted in sequence.

[0145] In this embodiment, preferably, the calculation formula for the sum of the distances between each particle position and each objective function value of its adjacent particle position in the external archive is:

[0146] ,

[0147] The absolute value of the difference between the distances of each particle position in the external archive and the objective function values ​​of its neighboring particle positions is calculated as:

[0148] ,

[0149] in, Indicates the number of The sum of the distances between the position of a particle and the objective function values ​​of its adjacent particle positions, Indicates the number of The absolute value of the difference between the distances of each particle position and the objective function values ​​of its adjacent particle positions, represents the total number of objective functions of the non-cooperative game model for microgrid capacity, Indicates the number of The particle position corresponds to the The objective function value, Indicates the number of The particle position corresponds to the The objective function value, Indicates the number of The particle position corresponds to the The objective function value, Indicates the number of The particle position, Indicates the number of The particle position, Indicates the number of The particle position, represents the total number of particle positions in the external archive scale, Indicates the particle position.

[0150] In this embodiment, the use of a fast stratification strategy to select a particle position in an external archive as the global optimal position of the current iteration includes:

[0151] Calculate the fast stratification value of each particle position in the external archive respectively, and sort all particle positions in the external archive in ascending order according to the size of the fast stratification value of each particle position in the external archive, and take the first particle position after ascending order as the global optimal position of the current iteration;

[0152] If there are multiple particle positions with the same minimum fast stratification value after ascending sorting, one of the particle positions is randomly selected as the global optimal position of the current iteration;

[0153] The quick layering value for each particle position in the external archive is calculated as:

[0154] ,

[0155] in, Indicates the number of Fast layered values ​​of particle positions, represents the total number of objective functions of the non-cooperative game model for microgrid capacity, The first column representing the positions of all particles in the external archive The minimum value among the objective function values, The first column representing the positions of all particles in the external archive The maximum value among the objective function values.

[0156] In this embodiment, , which are the total cost objective function, the wind and solar power abandonment objective function, and the ACG wind and solar power frequency regulation objective function.

[0157] The present invention solves many problems in the optimization of microgrid capacity configuration, including the problem of economic efficiency, renewable energy utilization rate and system stability. The present invention forms a non-cooperative game environment by allowing multiple independent entities in the microgrid (such as wind power, photovoltaics and energy storage units) to maximize their own interests and minimize the penalty costs for abandoning wind and solar power. In this multi-agent, complex constraint and dynamic environment, it is difficult for traditional intelligent optimization methods to effectively balance the interests of each entity, the penalty costs for abandoning wind and solar power and the interests of the total cost of all entities, resulting in unsatisfactory solution accuracy and optimization effect.

[0158] Therefore, the improved particle swarm algorithm is used to solve the non-cooperative game model of microgrid capacity and obtain the capacity target configuration result; when the external archive size exceeds the preset maximum size, the improved particle swarm algorithm will effectively delete the particles that exceed the size, which can not only improve the algorithm's search ability, but also maintain the diversity of the particle population, making it easy to retain and develop excellent particles in the subsequent evolution process, effectively improving the diversity of the population, and using a fast hierarchical strategy to select the global optimal position to improve the convergence speed and accuracy of the solution. The microgrid capacity configuration result is optimal, and the economic benefits of the microgrid system are maximized, the operation stability is improved, and the efficient use of renewable energy is achieved.

[0159] By optimizing capacity configuration, the present invention can flexibly respond to the access of large-scale renewable energy (such as wind power and photovoltaics). As the proportion of wind power and photovoltaic power generation in the power grid continues to increase, how to stably and efficiently utilize these renewable energy sources has become a key challenge. The optimization model used in the present invention can effectively solve this problem and ensure that the system can still maintain efficient operation when a high proportion of renewable energy is connected.

[0160] This second embodiment takes the actual data of a microgrid in a certain area as an example to demonstrate how to perform capacity optimization configuration through an improved multi-objective particle swarm optimization algorithm (MOPSO).

[0161] First, based on the historical wind speed data of the area, the wind speed probability density function of the Weibull distribution is fitted, and the wind speed shape parameter , scale parameter The input parameter of photovoltaic power generation is light intensity. The light intensity in the area follows the Beta distribution. The first solar radiation intensity shape parameter , the second solar radiation intensity shape parameter The microgrid load curve is obtained based on historical data. Set the conversion efficiency of the photovoltaic panel , the photovoltaic panel area is 2,000 square meters.

[0162] According to the historical wind speed data of the area, the wind speed probability density function of the Weibull distribution is fitted, and the formula is:

[0163] ,

[0164] in, is the probability density function of wind speed, is the wind speed shape parameter, is the scale parameter, is an exponential function with a natural constant as base, is the wind speed.

[0165] The formula for calculating the output power of photovoltaic power generation is:

[0166] ,

[0167] in, is the output power of photovoltaic power generation, is the conversion efficiency of the photovoltaic panel, is the photovoltaic panel area, is the solar radiation intensity.

[0168] Solar radiation intensity is modeled based on geographic location and climatic conditions and can be represented by a Beta distribution:

[0169] ,

[0170] in, is the solar radiation intensity represented by Beta distribution, and are the first solar radiation intensity shape parameter and the second solar radiation intensity shape parameter, is the Beta function.

[0171] The capacity of the energy storage system is initially set at 500kWh, its charging and discharging efficiency is 90%, the maximum charging power is 100kW, and the maximum discharging power is 100kW. The energy storage system balances the volatility of wind power and photovoltaic power by controlling its charging and discharging state.

[0172] Based on the above data, the output power of wind power generation and photovoltaic power generation per hour is calculated respectively.

[0173] The energy storage system smoothes the fluctuations of wind power and photovoltaic power generation by adjusting the charging and discharging status. The specific scheduling strategy is as follows:

[0174] (1) During the period when the power generation is greater than the load, the energy storage system is charged, and the charging power is as follows:

[0175] ,

[0176] in, For the remaining power, is the maximum charging power, In order to obtain the minimum value, the charging power of the energy storage system during the peak period of photovoltaic power generation during the day is calculated.

[0177] (2) During the period when the power generation is less than the load, the energy storage system discharges, and the discharge power is as follows:

[0178] ,

[0179] in, is the discharge power of the energy storage system, For power shortage, is the maximum discharge power.

[0180] The improved particle swarm algorithm is used to solve the non-cooperative game model of microgrid capacity, and the capacity target configuration results are obtained. The capacity target configuration results are: wind power capacity 800kW, photovoltaic capacity 600kW, and energy storage capacity 550kW. At the same time, the optimization scheme of the present invention reduces the total cost by 15% compared with the traditional scheme, and the overall economy of the system is significantly improved. It shows that through the scheduling of the energy storage system, the system power generation and load are balanced in all time periods, and the phenomenon of wind and light abandonment is significantly reduced.

[0181] Reference Figure 2 As shown, this third embodiment also provides a microgrid capacity configuration optimization system, including:

[0182] The model building module 10 is used to obtain the power model of each subject in the wind-solar-storage microgrid to be optimized, the AGC frequency regulation benefit model of the wind power generation unit and the photovoltaic power generation unit, and to construct a total cost objective function with the goal of minimizing the total cost of all subjects in the wind-solar-storage microgrid, to construct a wind abandonment objective function with the goal of minimizing the penalty cost of abandoning wind and photovoltaic power, and to construct an ACG wind-solar-storage frequency regulation objective function with the goal of maximizing the ACG wind-solar-storage frequency regulation benefit, and to set power balance constraints and wind abandonment constraints as constraint conditions to establish a microgrid capacity non-cooperative game model; wherein, the subjects of the wind-solar-storage microgrid include wind power generation units, photovoltaic power generation units and energy storage systems;

[0183] The solution module 20 is used to obtain the parameter data of the time period to be optimized, and input it into the non-cooperative game model of the microgrid capacity, and use the improved particle swarm algorithm to solve the non-cooperative game model of the microgrid capacity to obtain the capacity target configuration result. It should be understood by those skilled in the art that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0184] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, 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 generate 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 flowchart and / or block diagram. 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.

[0185] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate 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 A function specified in one or more boxes.

[0186] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0187] Obviously, the above embodiments are merely examples for the purpose of clear explanation and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived therefrom are still within the scope of protection of the present invention.

Claims

1. A microgrid capacity configuration optimization method, characterized in that: The following steps are involved: Obtain the power model of each subject in the wind-solar-storage microgrid in the area to be optimized, and the AGC frequency regulation benefit model of the wind power generation unit and the photovoltaic power generation unit. Construct a total cost objective function with the goal of minimizing the total cost of all subjects in the wind-solar-storage microgrid, construct a wind and solar abandonment objective function with the goal of minimizing the penalty cost of abandoning wind and solar power, and construct an ACG wind and solar frequency regulation objective function with the goal of maximizing the ACG wind and solar frequency regulation benefit. Set power balance constraints and wind and solar abandonment constraints as constraints, and establish a microgrid capacity non-cooperative game model; among which, the subjects of the wind-solar-storage microgrid include wind power generation units, photovoltaic power generation units and energy storage systems; Obtain the parameter data of the time period to be optimized and input them into the microgrid capacity non-cooperative game model. Use the improved particle swarm algorithm to solve the microgrid capacity non-cooperative game model to obtain the capacity target configuration results, including: Step S21: Set the particle swarm size, particle dimension, maximum number of iterations, initial weight, end weight, individual learning factor, group learning factor, and external archive size; Initialize , the Sobol sequence is used to generate the initial position of each particle, and the initial position of each particle is updated by the mean shift clustering algorithm to obtain the updated initial position of each particle; the initial velocity of each particle is randomly generated, where, Indicates the current iteration number. The updated initial position of each particle represents a capacity target configuration result. Step S22: the positions of each particle after each iteration are externally archived. If the particle positions stored in the external archive exceed the external archive size, the particle positions in the external archive are deleted based on the objective function values ​​corresponding to the positions of each particle. Step S23: using a fast stratification strategy to select a particle position in the external archive as the global optimal position of the current iteration; Step S24: based on the objective function value corresponding to each historical iteration position of each particle, obtain the Pareto optimal solution in each historical iteration position of each particle as the individual learning position of each particle in the current iteration; Step S25: Calculate the speed inertia weight of the current iteration based on the initial weight, the end weight, the current iteration number and the maximum iteration number; Update the speed of each particle based on the individual learning factor, the group learning factor, the current iteration number, the speed inertia weight of the current iteration, the global best position of the current iteration and the individual learning position of each particle in the current iteration; Step S26: Based on the objective function value corresponding to the current iteration position of each particle and the global optimal position of the current iteration, the position inertia weight of each particle in the current iteration is calculated; based on the individual learning factor, the group learning factor, the position inertia weight of each particle in the current iteration, the current iteration number, the global optimal position of the current iteration and the individual learning position of each particle in the current iteration, the position of each particle is updated. ; Step S27: Determine the current number of iterations Whether the maximum number of iterations has been reached, if so, the global best position of the current iteration is output as the capacity target configuration result, otherwise, steps S22-S27 are repeated; The velocity inertia weight of the current iteration is calculated based on the initial weight, the end weight, the current number of iterations and the maximum number of iterations. The calculation formula is: ， in, For the The velocity inertia weight of the iteration, For the ending weight, is the initial weight, is the maximum number of iterations; The speed of each particle is updated based on the individual learning factor, the group learning factor, the current iteration number, the speed inertia weight of the current iteration, the global optimal position of the current iteration and the individual learning position of each particle in the current iteration. The formula is: ， in, For the The iteration The speed of a particle, For the The iteration The speed of a particle, For the The iteration The position of a particle, For the The velocity inertia weight of the iteration, is the individual learning factor, is the group learning factor, is the current iteration number, is the maximum number of iterations, For the The iteration The individual learning positions of particles, For the The global best position of the iteration, is a random number in the interval [0, 1]; Based on the objective function value corresponding to the current iteration position of each particle and the global optimal position of the current iteration, the position inertia weight of each particle in the current iteration is calculated, and the calculation formula is: ， in, For the The iteration The position inertia weight of each particle, is a weight distribution factor in the range [0, 1], For the In the iteration The target fitness of each particle, , For the Particle No. The total cost objective function value corresponding to the iteration position, For the Particle No. The wind and solar curtailment objective function value corresponding to the iteration position is: For the Particle No. The ACG wind and solar frequency modulation objective function value corresponding to the iteration position, To set a positive number, For all particles in The highest target fitness in iterations, For the The target fitness corresponding to the global best position of the iteration; The position of each particle is updated based on the individual learning factor, the group learning factor, the position inertia weight of each particle in the current iteration, the current number of iterations, the global optimal position of the current iteration and the individual learning position of each particle in the current iteration. The formula is: ， in, For the The iteration The position of a particle, For the The iteration The position of a particle, For the The iteration The speed of a particle, For the The iteration The position inertia weight of each particle.

2. A microgrid capacity configuration optimization method according to claim 1, characterized in that: The method of obtaining the power model of each subject in the wind-solar-storage microgrid in the area to be optimized, and the AGC frequency regulation benefit model of the wind power generation unit and the photovoltaic power generation unit includes: The wind power generation model of the wind power generation unit is: ， in, for The wind power generation capacity in the period, is the rated power of the wind power generation unit, is the cut-in wind speed of the wind power generation unit, is the rated wind speed of the wind power generation unit, is the cut-out wind speed of the wind power generation unit, For wind power generation units Wind speed during the time period; The photovoltaic power generation model of the photovoltaic power generation unit is: ， in, for The photovoltaic power generation during the period, is the rated power of the PV module, is the temperature coefficient, for The temperature of the time period, is the rated temperature of the PV module; The energy storage system power model of the energy storage system is: ， ， in, for The output power of the energy storage system during the period, is the charging power of the energy storage system, represents the discharge power of the energy storage system, for The net power of the microgrid during the period, for The wind power generation capacity in the period, for The photovoltaic power generation during the period, for The load of the microgrid during the period; The AGC frequency regulation revenue model of the wind power generation unit is: ， in, for Wind power frequency regulation benefits during the period, is the wind power frequency regulation benefit per unit time, is the regulation coefficient of wind power frequency regulation, for The total number of wind power frequency modulations during the period, for During the period The depth of regulation of secondary wind power frequency regulation; The AGC frequency regulation benefit model of the photovoltaic power generation unit is: ， in, for The photovoltaic frequency modulation income of the time period, is the photoelectric frequency modulation gain per unit time, is the adjustment coefficient of photoelectric frequency modulation, for The total number of photoelectric frequency modulations in the time period, for During the period The adjustment depth of sub-photoelectric frequency modulation.

3. A microgrid capacity configuration optimization method according to claim 2, characterized in that: The non-cooperative game model of microgrid capacity is: ， ， ， Power balance constraints: ， Wind and solar curtailment constraints: ， in, is the total cost objective function, is the wind and solar curtailment objective function, is the ACG wind and solar frequency modulation objective function, for Time period The electricity purchase cost of each entity, for Time period The operation and maintenance costs of each entity, for Time period The energy storage system cost of each entity, for The penalty fees for abandoning wind and solar power during the period, for The penalty fees incurred by all entities for abandoning wind and solar power during the period, Indicates the time interval, is the total duration, As a penalty factor for abandoning wind and solar power, for The photovoltaic frequency modulation income of the time period, for Wind power frequency regulation benefits during the period, for The power purchased by the power grid during the period, for The wind power generation capacity in the period, for The photovoltaic power generation during the period, for The output power of the energy storage system during the period, for The load of the microgrid during the period, for The power abandoned by wind energy during the period, for The power abandoned by photovoltaic power during the period, is the maximum power of wind power generation, is the maximum power of photovoltaic power generation.

4. A microgrid capacity configuration optimization method according to claim 1, characterized in that: The Sobol sequence is used to generate the initial position of each particle, and the initial position of each particle is updated by the mean shift clustering algorithm to obtain the updated initial position of each particle, including: The Sobol sequence is used to generate Q Sobol sequence sample points, each of which represents the initial position of a particle, where Q is the size of the particle group; The mean shift value of each Sobol sequence sample point is calculated by the mean shift clustering algorithm; Based on the mean offset value of each Sobol sequence sample point, each Sobol sequence sample point is updated to obtain the updated initial position of each particle.

5. A microgrid capacity configuration optimization system, characterized in that: include: The model building module is used to obtain the power model of each subject in the wind-solar-storage microgrid in the area to be optimized, the AGC frequency regulation benefit model of the wind power generation unit and the photovoltaic power generation unit, and to build a total cost objective function with the goal of minimizing the total cost of all subjects in the wind-solar-storage microgrid, to build a wind and photovoltaic abandonment objective function with the goal of minimizing the penalty cost of abandoning wind and photovoltaic, and to build an ACG wind and photovoltaic frequency regulation objective function with the goal of maximizing the ACG wind and photovoltaic frequency regulation benefit. The power balance constraint and wind and photovoltaic abandonment constraint are set as constraint conditions to establish a non-cooperative game model of microgrid capacity; among which, the subjects of the wind-solar-storage microgrid include wind power generation units, photovoltaic power generation units and energy storage systems; The solution module is used to obtain the parameter data of the time period to be optimized and input it into the microgrid capacity non-cooperative game model. The improved particle swarm algorithm is used to solve the microgrid capacity non-cooperative game model to obtain the capacity target configuration results, including: Step S21: Set the particle swarm size, particle dimension, maximum number of iterations, initial weight, end weight, individual learning factor, group learning factor, and external archive size; Initialize , the Sobol sequence is used to generate the initial position of each particle, and the initial position of each particle is updated by the mean shift clustering algorithm to obtain the updated initial position of each particle; the initial velocity of each particle is randomly generated, where, Indicates the current iteration number. The updated initial position of each particle represents a capacity target configuration result. Step S22: the positions of each particle after each iteration are externally archived. If the particle positions stored in the external archive exceed the external archive size, the particle positions in the external archive are deleted based on the objective function values ​​corresponding to the positions of each particle. Step S23: using a fast stratification strategy to select a particle position in the external archive as the global optimal position of the current iteration; Step S24: based on the objective function value corresponding to each historical iteration position of each particle, obtain the Pareto optimal solution in each historical iteration position of each particle as the individual learning position of each particle in the current iteration; Step S25: Calculate the speed inertia weight of the current iteration based on the initial weight, the end weight, the current iteration number and the maximum iteration number; Update the speed of each particle based on the individual learning factor, the group learning factor, the current iteration number, the speed inertia weight of the current iteration, the global best position of the current iteration and the individual learning position of each particle in the current iteration; Step S26: Based on the objective function value corresponding to the current iteration position of each particle and the global optimal position of the current iteration, the position inertia weight of each particle in the current iteration is calculated; based on the individual learning factor, the group learning factor, the position inertia weight of each particle in the current iteration, the current iteration number, the global optimal position of the current iteration and the individual learning position of each particle in the current iteration, the position of each particle is updated. ; Step S27: Determine the current number of iterations Whether the maximum number of iterations has been reached, if so, the global best position of the current iteration is output as the capacity target configuration result, otherwise, steps S22-S27 are repeated; The velocity inertia weight of the current iteration is calculated based on the initial weight, the end weight, the current number of iterations and the maximum number of iterations. The calculation formula is: ， in, For the The velocity inertia weight of the iteration, For the ending weight, is the initial weight, is the maximum number of iterations; The speed of each particle is updated based on the individual learning factor, the group learning factor, the current iteration number, the speed inertia weight of the current iteration, the global optimal position of the current iteration and the individual learning position of each particle in the current iteration. The formula is: ， in, For the The iteration The speed of a particle, For the The iteration The speed of a particle, For the The iteration The position of a particle, For the The velocity inertia weight of the iteration, is the individual learning factor, is the group learning factor, is the current iteration number, is the maximum number of iterations, For the The iteration The individual learning positions of particles, For the The global best position of the iteration, is a random number in the interval [0, 1]; Based on the objective function value corresponding to the current iteration position of each particle and the global optimal position of the current iteration, the position inertia weight of each particle in the current iteration is calculated, and the calculation formula is: ， in, For the The iteration The position inertia weight of each particle, is a weight distribution factor in the range [0, 1], For the In the iteration The target fitness of each particle, , For the Particle No. The total cost objective function value corresponding to the iteration position, For the Particle No. The wind and solar curtailment objective function value corresponding to the iteration position is: For the Particle No. The ACG wind and solar frequency modulation objective function value corresponding to the iteration position, To set a positive number, For all particles in The highest target fitness in iterations, For the The target fitness corresponding to the global best position of the iteration; The position of each particle is updated based on the individual learning factor, the group learning factor, the position inertia weight of each particle in the current iteration, the current number of iterations, the global optimal position of the current iteration and the individual learning position of each particle in the current iteration. The formula is: ， in, For the The iteration The position of a particle, For the The iteration The position of a particle, For the The iteration The speed of a particle, For the The iteration The position inertia weight of each particle.

Images