Wind power stabilizing method based on adaptive particle swarm algorithm

Abstract

The invention provides a wind power stabilizing method based on an adaptive particle swarm algorithm in the technical field of wind power integration. The method comprises the following steps: S1, initializing a particle swarm to obtain an initial population; s2, setting a target function; s3, inputting the initial population and the energy storage parameters into a target function to obtain initial fitness; s4, dividing the initial population into sub-populations, and calculating global optimal particles and local optimal particles based on the initial fitness; s5, setting an inertia weight updating rule and a population updating normal form to update the population, and storing global optimal particles and local optimal particles of each generation; s6, selecting a global optimal individual from each global optimal particle and each local optimal particle to calculate the optimal fitness; and S7, determining the optimal rated capacity and the optimal rated power based on the global optimal individual, and determining the minimum cost based on the optimal fitness so as to adjust the system configuration to stabilize the wind power output. The method has the advantage that the smoothness of wind power output is greatly improved.

Description

Technical Field

[0001] The invention relates to the technical field of wind power grid connection, and in particular to a wind power level suppression method based on an adaptive particle swarm algorithm. Background Art

[0002] As traditional energy sources are depleted and the need to transition to clean energy increases, wind power is gradually gaining attention as a prominent renewable energy option. However, the intermittent and unpredictable nature of wind power generation leads to significant fluctuations related to wind speed, which pose challenges to the stability of grid operation and cause harmonic pollution. In order to ensure the safety and reliability of grid operation, the power output of wind power needs to be leveled to smooth the power output.

[0003] Smoothing power output solutions can be classified according to whether they are equipped with energy storage devices. Solutions without energy storage devices rely on the power equipment's own capabilities to adjust power output and achieve power smoothing, but the smoothing effect is limited; solutions with energy storage devices can significantly improve power smoothing capabilities. Battery energy storage systems (BESS) are commonly used as energy storage systems for smoothing wind power output due to their high efficiency, fast response, low cost and easy installation. When configuring an energy storage system, it is necessary to consider technical and economic indicators in single or multiple application scenarios. In some scenarios, it is also necessary to consider the impact of factors such as prediction errors and scheduling plan changes. At present, the mainstream energy storage system configuration methods are as follows:

[0004] 1. Configuration method based on sequential simulation: With technical goals as the direction and a single objective function as the basis for configuration; first obtain historical data of renewable energy power stations, and predict future electricity according to different actual application scenarios, and formulate corresponding energy storage system charging and discharging operation strategies. After weighing factors such as return on investment and application value, the optimal energy storage rated power and rated capacity are obtained according to the time series power demand.

[0005] 2. Deterministic configuration method: Taking the rated power and rated capacity of the energy storage system as the decision variables of the model, based on the deterministic assumption of the data sample and combined with the historical operation data, a renewable energy storage system model with optimized energy storage configuration is established, and the economic or technical aspects are taken into consideration. Then, a solution algorithm is used to obtain the optimal configuration result of the energy storage system.

[0006] 3. Uncertainty configuration method: Due to the uncertainty of renewable energy output, the corresponding power forecast and grid absorption capacity are also uncertain, which makes the energy storage demand uncertain. In view of this uncertainty, an uncertainty configuration method is proposed to solve this problem; uncertainty configuration methods (uncertainty planning) include random planning, fuzzy planning, robust optimization, etc. The application scenarios mainly include improving the fault ride-through capability, grid connection capability, reliable capacity and active support capability of renewable energy units.

[0007] The PSO (Particle Swarm Optimization) algorithm is an evolutionary algorithm inspired by nature. It conceptualizes each candidate solution to an optimization problem as a particle. The collection of these particles is also called a swarm. It navigates in the D-dimensional search space inherent to the problem. The position of the particle is affected by its own experience and the experience of its neighboring particles. It has a good effect on the optimization model with energy storage system configuration as the decision variable.

[0008] However, the traditional PSO algorithm has fixed parameters and inertia weights that cannot be adjusted as the group is updated, resulting in insufficient utilization of group information, poor optimization effect, and thus affecting the wind level suppression effect. Traditionally, there are also variant algorithms of PSO: 1. The CLPSO (Comprehensive Learning Particle Swarm Optimizer) algorithm adopts an innovative learning strategy to update the speed of a given particle using the historical best information of all other particles; 2. Nasir et al. further extended the CLPSO algorithm by restricting the demonstration particles to a dynamic neighborhood; 3. The Dynamic Neighborhood-based Switching PSO (DNSPSO) algorithm designs a new speed update mechanism to adjust the personal best position and the global best position based on the distance-related dynamic neighborhood to make full use of the evolutionary information in the entire group; 4. A PSO algorithm using an iterative equation containing inertia weights; 5. A PSO algorithm in which the inertia weight of each particle in each dimension is determined by the particle's performance and the distance to the historical best position; 6. A PSO algorithm based on chaos-based nonlinear inertia weights, called MPSO; 7. A single-objective optimization PSO variant called PSO-sono, in which a new adaptive scheme for paradigm ratio and shrinkage coefficient is introduced during the iteration process.

[0009] Due to the shortcomings of traditional PSO and PSO variants, such as fixed search mode, inability to adjust the search mode during the search process, and frequent trapping in local optimality, the optimization effect is poor, which in turn affects the smoothness of wind power output, that is, the wind power suppression effect is poor.

[0010] Therefore, how to provide a wind speed suppression method based on an adaptive particle swarm algorithm to achieve the smoothness of improving wind power output has become a technical problem that needs to be solved urgently. Summary of the invention

[0011] The technical problem to be solved by the present invention is to provide a wind speed suppression method based on an adaptive particle swarm algorithm to improve the smoothness of wind power output.

[0012] The present invention is implemented as follows: a wind level suppression method based on an adaptive particle swarm algorithm comprises the following steps:

[0013] Step S1, setting population parameters of a particle swarm of a wind power energy storage system, and initializing the particle swarm based on the population parameters to obtain an initial population;

[0014] Step S2, obtaining energy storage parameters of the wind power energy storage system and setting the objective function of the wind power energy storage system;

[0015] Step S3, inputting the initial population and the initial wind power in the energy storage parameter into the objective function to obtain the initial fitness;

[0016] Step S4, dividing the initial population into two sub-populations, and calculating the global optimal particle of the initial population and the local optimal particle of the sub-population based on the initial fitness;

[0017] Step S5, setting an inertia weight update rule and a population update paradigm, updating the initial population and the sub-population based on the inertia weight update rule and the population update paradigm, and saving the global optimal particles and the local optimal particles updated in each generation;

[0018] Step S6, selecting a global optimal individual from the global optimal particles and the local optimal particles, and calculating the optimal fitness of the global optimal individual;

[0019] Step S7: determining the optimal rated capacity and the optimal rated power of the wind energy storage system based on the global optimal individual, determining the minimum cost of the wind energy storage system based on the optimal fitness, and adjusting the system configuration of the wind energy storage system based on the optimal rated capacity, the optimal rated power and the minimum cost to stabilize the wind power output.

[0020] Furthermore, in step S1, the population parameters at least include an upper bound lb of the solution space, a lower bound ub of the solution space, a dimension dim of the solution, an initialization population number N, and an iteration number T; the value of the dimension dim is 2, including two dimensions of rated capacity and rated power;

[0021] The formula for the initial population is:

[0022] X = lb + rand * (ub - lb);

[0023] Among them, X represents the initial population; rand represents a random matrix with dimension dim and size N.

[0024] Furthermore, in step S2, the energy storage parameters include at least unit power, capacity cost, charge and discharge efficiency, energy storage status, capital recovery rate, initial wind power, rated capacity and rated power.

[0025] Furthermore, in step S2, the formula of the objective function is:

[0026] minF=C+αE penalty +βE curtail +C cycle ;

[0027] C penalty +C curtail =αE penalty +βE curtail ;

[0028] Among them, minF represents the minimum cost, that is, fitness; C represents the annual equivalent cost; E penalty Indicates the number of years with insufficient wind suppression; E curtail represents the annual amount of wind power abandoned; α represents E penalty The unit penalty cost of E curtail Unit penalty cost; C cycle represents the life cycle cost of the wind power storage system; C penalty Indicates the penalty for insufficient wind level suppression; C curtail It indicates the penalty for wind power abandonment.

[0029] Furthermore, in step S5, the formula of the inertia weight update rule is:

[0030]

[0031] ω2=1-ω1;

[0032] Among them, ω2 represents the inertia weight after update; ω1 represents the inertia weight before update; sin() represents the sine function; gen represents the number of iterations; gen max Indicates the maximum number of iterations.

[0033] Furthermore, in step S5, the population update paradigm is specifically:

[0034] Set an inertia weight threshold A1, an inertia weight threshold A2, and a random value r in the range of [0,1];

[0035] When ω2<A1, the formula for population update is:

[0036]

[0037] When ω1>A2, the formula for population update is:

[0038]

[0039] When r<0.6 and ω1<A2, the formula for population update is:

[0040]

[0041] When r>0.6 and ω1<A2, the formula for population update is;

[0042]

[0043] Among them, V i,G+1 represents the search speed of the i-th particle of population X in the (G+1)th generation; X i,G represents the i-th particle of population X in the Gth generation; X i,G+1 represents the ith particle of population X in the (G+1)th generation; step represents the search step; Xmax represents the upper limit of the search space; Xmin represents the lower limit of the search space; d1 represents the population search speed control parameter in the second search stage, and d1∈{-1,1}; d2 represents the population search speed control parameter in the third search stage; d3 represents the population search speed control parameter in the fourth search stage; c represents the control parameter used to speed up the search in each search stage, and c=2; X gbest,G Represents the global optimal particle of population X in the Gth generation; r is used to cooperate with d1, d2, d3 and c to control the population search speed.

[0044] Furthermore, in step S5, the population update paradigm further includes:

[0045] Use migration operators to facilitate information exchange among subpopulations:

[0046] X mworse,G =X fbest,G ;

[0047] X fworse,G =X mbest,G ;

[0048] Among them, X mworse,G represents the global worst particle of subpopulation m in the Gth generation; X fbest,G represents the global optimal particle of subpopulation f in the Gth generation; X fworse,G represents the global worst particle of subpopulation f in the Gth generation; X mbest,G Represents the global optimal particle of subpopulation m in the Gth generation.

[0049] The advantages of the present invention are:

[0050] By setting the population parameters of the particle swarm of the wind power energy storage system, the particle swarm is initialized based on the population parameters to obtain the initial population; then the energy storage parameters of the wind power energy storage system are obtained, the objective function of the wind power energy storage system is set, the initial population and the initial wind power in the energy storage parameters are input into the objective function to obtain the initial fitness, and then the initial population is divided into two sub-populations, and the global optimal particles of the initial population and the local optimal particles of the sub-population are calculated based on the initial fitness; then the inertia weight update rule and the population update paradigm are set to update the initial population and the sub-population, and the global optimal particles and local optimal particles updated in each generation are saved, and the global optimal individuals are selected from each global optimal particle and local optimal particle, and the optimal fitness of the global optimal individual is calculated. Finally, the optimal rated capacity and the optimal rated power of the wind power energy storage system are determined based on the global optimal individual, and the minimum cost of the electric energy storage system is determined based on the optimal fitness. The inertia weight update rule dynamically adjusts the search step and search speed of the particles during the iterative update process. Compared with the weight update scheme of traditional PSO or PSO variants, it reduces the premature convergence of the algorithm during the search process, increases the probability of the algorithm searching for the global optimal particle, and ensures the balance between the exploration and development stages; the population update paradigm mixes Levy search, quasi-reverse learning strategy and migration operator, and uses different update strategies in different search stages, so that different search stages have different search focuses, avoiding the problem that the algorithm searches in a fixed direction, resulting in a high probability that the search direction deviates from the correct direction, and cannot be adjusted in the middle of the search, thereby improving the optimization performance of the population update, and ultimately greatly improving the system configuration optimization effect of the wind power storage system, thereby greatly improving the smoothness of the wind power output. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The present invention will be further described below in conjunction with embodiments with reference to the accompanying drawings.

[0052] Figure 1 The present invention is a flow chart of a wind level suppression method based on an adaptive particle swarm algorithm. DETAILED DESCRIPTION

[0053] The technical solution in the embodiment of the present application has the following overall idea: the particle swarm is updated through the inertia weight update rule and the population update paradigm. The inertia weight update rule dynamically adjusts the search step size and search speed of the particles during the iterative update process. Compared with the weight update scheme of the traditional PSO or PSO variant, it reduces the premature convergence of the algorithm during the search process, increases the probability of the algorithm searching for the global optimal particle, and ensures the balance between the exploration and development stages; the population update paradigm mixes Levy search, quasi-reverse learning strategy and migration operator, and uses different update strategies in different search stages, so that different search stages have different search focuses, avoiding the problem that the algorithm searches in a fixed direction, resulting in a high probability that the search direction deviates from the correct direction, and cannot be adjusted in the middle of the search, that is, the system configuration optimization effect of the wind power storage system is improved, and then the smoothness of the wind power output is improved.

[0054] Please refer to Figure 1 As shown, a preferred embodiment of a wind level suppression method based on an adaptive particle swarm algorithm of the present invention comprises the following steps:

[0055] Step S1, setting population parameters of a particle swarm of a wind power energy storage system, and initializing the particle swarm based on the population parameters to obtain an initial population;

[0056] Step S2, obtaining energy storage parameters of the wind power energy storage system and setting the objective function of the wind power energy storage system;

[0057] Step S3, inputting the initial population and the initial wind power in the energy storage parameter into the objective function to obtain the initial fitness; the fitness is used to measure the quality of the solution;

[0058] Step S4, dividing the initial population into two sub-populations, and calculating the global optimal particle of the initial population and the local optimal particle of the sub-population based on the initial fitness;

[0059] Step S5, setting an inertia weight update rule and a population update paradigm, updating the initial population and the sub-population based on the inertia weight update rule and the population update paradigm, and saving the global optimal particles and the local optimal particles updated in each generation; the population update paradigm includes different strategies, which can effectively utilize group information, improve the coefficient of the iterative paradigm, and adopt an adaptive scheme to enhance performance, that is, by improving the overall algorithm framework and using different strategies at different stages, the search efficiency is effectively improved;

[0060] That is, the population is updated through the adaptive particle swarm algorithm (APSO), and the dimensional concept of particles (individuals) in the adaptive particle swarm algorithm is combined with the specific wind power energy storage system; the rated power and rated capacity are used as two dimensions of individuals in the adaptive particle swarm algorithm for optimization iteration, and are used to measure the fitness of the individual quality, that is, the quality of the solution, and are calculated using the minimum cost formula in the wind power energy storage system. This can better combine the abstract algorithm with the configuration optimization problem of the actual wind power energy storage system, optimize the configuration of the wind power energy storage system and reduce the cost. The optimized system configuration can give full play to the effect of the wind power energy storage system on smoothing wind power fluctuations, and reduce the impact of wind power fluctuations on the grid connection process and the resulting risks;

[0061] Step S6, selecting a global optimal individual from the global optimal particles and the local optimal particles, and calculating the optimal fitness of the global optimal individual;

[0062] Step S7: Determine the optimal rated capacity and the optimal rated power of the wind energy storage system based on the global optimal individual, determine the minimum cost of the electric energy storage system based on the optimal fitness, and adjust the system configuration of the wind energy storage system based on the optimal rated capacity, the optimal rated power and the minimum cost to stabilize the wind power output. That is, calculate the wind power, fluctuation rate and other indicators after the smoothing based on the optimal configuration, and judge the ability to smooth the wind power fluctuation.

[0063] The present invention optimizes the particle swarm of the wind power storage system through an adaptive particle swarm algorithm (APSO), that is, optimizes the configuration of the wind power storage system, and then adjusts the wind power output based on the optimized wind power storage system to perform wind power suppression.

[0064] In step S1, the population parameters include at least the upper bound lb of the solution space, the lower bound ub of the solution space, the dimension dim of the solution, the number of initialized populations N, and the number of iterations T; the value of the dimension dim is 2, including the two dimensions of rated capacity and rated power;

[0065] The formula for the initial population is:

[0066] X = lb + rand * (ub - lb);

[0067] Among them, X represents the initial population; rand represents a random matrix with dimension dim and size N.

[0068] In step S2, the energy storage parameters at least include unit power, capacity cost, charge and discharge efficiency, energy storage state, capital recovery rate, initial wind power, rated capacity and rated power. The energy storage parameters also include annual equivalent cost, annual number of wind power suppression, annual number of wind power abandonment, unit penalty cost, life cycle cost, penalty for insufficient wind power suppression, penalty for wind power abandonment, discount rate, life cycle of wind power storage system, capital recovery coefficient, investment cost, operation and maintenance cost, unit capacity, rated power cost, unit power cost, unit power processing cost, unit capacity processing cost, capacity of wind power storage system, and number of replacement times of lithium batteries.

[0069] In step S2, the formula of the objective function is:

[0070] minF=C+αE penalty +βE curtail +C cycle ;

[0071] C penalty +C curtail =αE penalty +βE curtail ;

[0072] Among them, minF represents the minimum cost, that is, fitness; C represents the annual equivalent cost; E penalty Indicates the number of years with insufficient wind suppression; E curtail represents the annual amount of wind power abandoned; α represents E penalty The unit penalty cost of E curtail Unit penalty cost; C cycle represents the life cycle cost of the wind power storage system; C penalty Indicates the penalty for insufficient wind level suppression; C curtail It indicates the penalty for wind power abandonment.

[0073]

[0074] C install =C capacity C rated +P power P rated ;

[0075] Among them, d r represents the discount rate; T represents the life cycle of the wind power storage system; represents the capital recovery coefficient; C install represents the initial installation investment cost of the wind power energy storage system; C om represents the operation and maintenance cost of the wind power storage system; C capacity Indicates the rated capacity set for the wind power energy storage system; C ratedIndicates unit capacity; P power represents the rated power cost of the wind power storage system; P rated represents the unit power cost;

[0076] C cycle =(k pr_power P HN +k pr_capacity E HN )(ε+1)(1+r) -T ;

[0077] Among them, k pr_power P represents the unit power processing cost set by the wind power storage system; HN Indicates wind power; k pr_capacity E represents the unit capacity processing cost set by the wind power energy storage system; HN represents the capacity of the wind power storage system; ε represents the number of times the lithium battery of the wind power storage system is replaced; r represents a random value in the range of [0,1].

[0078] In the specific implementation, constraints are also set, that is, to ensure that the wind power fluctuation passing through the wind power storage system meets the grid connection standard and reduce the risk of wind power grid connection, a sliding average filter is used to estimate the expected wind power output of the wind power storage system, denoted as P R (t), expressed as follows:

[0079]

[0080] Where k represents time; R(k) represents the original wind power output; P R (t) represents the expected wind power output at time t; W represents the length of the sliding average window. By selecting an appropriate W, the ability of the wind power storage system to smooth wind power can be effectively enhanced while ensuring compliance with grid standards and avoiding grid connection risks caused by over-smoothing or under-smoothing.

[0081] In order to control the fluctuation of wind power output, it is necessary to constrain the charging and discharging capacity of the wind power storage system. The formula for controlling the charging and discharging power of the wind power storage system at each time point is:

[0082]

[0083] Among them, P c (t) represents the charging capacity of the wind power storage system; P d (t) represents the discharge capacity of the wind power storage system; represents the maximum charging power that the wind energy storage system can accept; B represents the preset maximum capacity of the wind energy storage system; S(t) represents the energy storage state of the wind energy storage system at time t; ηΔt represents the charging efficiency within the time period Δt; R(t) represents the original wind power output at time t; D(t) represents the load power; Represents the constraint condition of the maximum power that the wind power storage system can accept; represents the maximum discharge power; β' represents the discharge efficiency; max{0,R(t)-D(t)} represents the charging demand of the wind energy storage system when the original wind power is in excess; max{0,D(t)-R(t)} represents the discharge demand of the wind energy storage system when the original wind power is low; Used to ensure that the wind power energy storage system does not fall below the minimum capacity during discharge.

[0084] During the stabilization process of the wind power storage system, charging and discharging cannot occur at the same time, so the constraints are set:

[0085] P d (t)·P c (t) = 0;

[0086] The calculation formula of the energy storage system's state of storage (SOC) at each time sampling point is:

[0087]

[0088] When t = 1, S(1) represents the energy storage state of the wind power energy storage system at the initial stage. In addition, SOC is subject to the following constraints:

[0089] SOC max ≥S(t)≥SOC mini ;

[0090] Among them, SOC max Indicates the upper limit of SOC; SOC mini Indicates the SOC lower limit.

[0091] Based on the above description, the objective function of the wind power energy storage system configuration optimization problem can be expressed as follows:

[0092] Minimize:

[0093] min F=C+αE penalty +βE curtail +C cycle ;

[0094] subject to:

[0095]

[0096] P d(t)·P c (t) = 0;

[0097]

[0098] SOC max ≥S(t)≥SOC mini ;

[0099] Where:

[0100]

[0101] C penalty +C curtail =αE penalty +βE curtail ;

[0102] C cycle =(k pr_power P HN +k pr_capacity E HN )(ε+1)(1+r) -T .

[0103] In step S5, the formula of the inertia weight update rule is:

[0104]

[0105] ω2=1-ω1;

[0106] Among them, ω2 represents the inertia weight after update; ω1 represents the inertia weight before update; sin() represents the sine function; gen represents the number of iterations; gen max Indicates the maximum number of iterations.

[0107] Setting the inertia weight is crucial to keeping particles at a certain search speed. However, using a fixed inertia weight may lead to an imbalance between exploration and exploitation, resulting in premature convergence. In order to solve this problem and achieve a balance between the two stages of exploration and exploitation, the present invention introduces a linear inertia weight update rule based on a sine function, so that particles can maintain different search speeds and directions in different search stages; for example, in the initial stage, a larger weight is maintained to promote rapid exploration of the entire search space, and in the later search stage, a smaller inertia weight is used to navigate near promising candidate solutions to improve search accuracy.

[0108] In step S5, the population update paradigm is specifically:

[0109] Set an inertia weight threshold A1, an inertia weight threshold A2, and a random value r in the range of [0,1];

[0110] In the initial stage, when particles need to explore the entire search space, Levy search is used, that is, when ω2<A1, the formula for population update is:

[0111]

[0112] It is necessary to guide the particles to search effectively in the right direction to ensure that the update mode can improve the quality of candidate solutions. That is, when ω1>A2, the formula for population update is:

[0113]

[0114] The particle swarm completes the exploration of the search space and finds promising individuals. The goal is to concentrate the search near these promising individuals to improve the precision and accuracy of the search results. In order to avoid falling into the local optimum, the linear inertia weight is combined with the new guide position based on the principle of quasi-adversarial learning. That is, when r < 0.6 and ω1 < A2, the formula for population update is:

[0115]

[0116] When r>0.6 and ω1<A2, the formula for population update is;

[0117]

[0118] Among them, V i,G+1 represents the search speed of the i-th particle of population X in the (G+1)th generation; X i,G represents the i-th particle of population X in generation G; X i,G+1 represents the i-th particle of population X in the (G+1) generation; step represents the search step length, u and v are both random values ​​between [0, 1]; Xmax represents the upper limit of the search space; Xmin represents the lower limit of the search space; d1 represents the population search speed control parameter in the second search stage, and d1∈{-1,1}; d2 represents the population search speed control parameter in the third search stage; d3 represents the population search speed control parameter in the fourth search stage; c represents the control parameter used to speed up the search in each search stage, and c=2; X gbest,G Represents the global optimal particle of population X in the Gth generation; r is used to cooperate with d1, d2, d3 and c to control the population search speed.

[0119]

[0120] Among them, X mgbest,G represents the best individual in one of the subpopulations; f() represents the fitness of the individual; X f,Grepresents the i-th particle in one of the subpopulations in the G generation; ε represents a positive number that is infinitely close to 0; X mi,G represents the i-th particle in one of the subpopulations in the G-th generation.

[0121] In step S5, the population update paradigm further includes:

[0122] Use migration operators to facilitate information exchange among subpopulations:

[0123] X mworse,G =X fbest,G ;

[0124] X fworse,G =X mbest,G ;

[0125] Among them, X mworse,G represents the global worst particle of subpopulation m in the Gth generation; X fbest,G represents the global optimal particle of subpopulation f in the Gth generation; X fworse,G represents the global worst particle of subpopulation f in the Gth generation; X mbest,G Represents the global optimal particle of subpopulation m in the Gth generation.

[0126] In order to enhance the selection of high-quality individuals and improve the population quality, poor individuals are removed from the two sub-populations. In addition, a migration operator is used to promote information exchange between sub-populations, thereby improving search accuracy and maintaining a balance between exploration and exploitation.

[0127] In summary, the advantages of the present invention are:

[0128] By setting the population parameters of the particle swarm of the wind power energy storage system, the particle swarm is initialized based on the population parameters to obtain the initial population; then the energy storage parameters of the wind power energy storage system are obtained, the objective function of the wind power energy storage system is set, the initial population and the initial wind power in the energy storage parameters are input into the objective function to obtain the initial fitness, and then the initial population is divided into two sub-populations, and the global optimal particles of the initial population and the local optimal particles of the sub-population are calculated based on the initial fitness; then the inertia weight update rule and the population update paradigm are set to update the initial population and the sub-population, and the global optimal particles and local optimal particles updated in each generation are saved, and the global optimal individuals are selected from each global optimal particle and local optimal particle, and the optimal fitness of the global optimal individual is calculated. Finally, the optimal rated capacity and the optimal rated power of the wind power energy storage system are determined based on the global optimal individual, and the minimum cost of the electric energy storage system is determined based on the optimal fitness. The inertia weight update rule dynamically adjusts the search step and search speed of the particles during the iterative update process. Compared with the weight update scheme of traditional PSO or PSO variants, it reduces the premature convergence of the algorithm during the search process, increases the probability of the algorithm searching for the global optimal particle, and ensures the balance between the exploration and development stages; the population update paradigm mixes Levy search, quasi-reverse learning strategy and migration operator, and uses different update strategies in different search stages, so that different search stages have different search focuses, avoiding the problem that the algorithm searches in a fixed direction, resulting in a high probability that the search direction deviates from the correct direction, and cannot be adjusted in the middle of the search, thereby improving the optimization performance of the population update, and ultimately greatly improving the system configuration optimization effect of the wind power storage system, thereby greatly improving the smoothness of the wind power output.

[0129] Although the specific implementation modes of the present invention are described above, those skilled in the art should understand that the specific implementation modes described are only illustrative and are not intended to limit the scope of the present invention. Equivalent modifications and changes made by those skilled in the art in accordance with the spirit of the present invention should be included in the scope of protection of the claims of the present invention.

Claims

1. A wind level suppression method based on an adaptive particle swarm algorithm, characterized in that: The steps include: Step S1, setting population parameters of a particle swarm of a wind power energy storage system, and initializing the particle swarm based on the population parameters to obtain an initial population; Step S2, obtaining energy storage parameters of the wind power energy storage system and setting the objective function of the wind power energy storage system; Step S3, inputting the initial population and the initial wind power in the energy storage parameter into the objective function to obtain the initial fitness; Step S4, dividing the initial population into two sub-populations, and calculating the global optimal particle of the initial population and the local optimal particle of the sub-population based on the initial fitness; Step S5, setting an inertia weight update rule and a population update paradigm, updating the initial population and the sub-population based on the inertia weight update rule and the population update paradigm, and saving the global optimal particles and the local optimal particles updated in each generation; Step S6, selecting a global optimal individual from the global optimal particles and the local optimal particles, and calculating the optimal fitness of the global optimal individual; Step S7: determining the optimal rated capacity and the optimal rated power of the wind energy storage system based on the global optimal individual, determining the minimum cost of the wind energy storage system based on the optimal fitness, and adjusting the system configuration of the wind energy storage system based on the optimal rated capacity, the optimal rated power and the minimum cost to stabilize the wind power output.

2. The wind level suppression method based on adaptive particle swarm algorithm according to claim 1, characterized in that: In step S1, the population parameters include at least the upper bound lb of the solution space, the lower bound ub of the solution space, the dimension dim of the solution, the number of initialized populations N, and the number of iterations T; the value of the dimension dim is 2, including the two dimensions of rated capacity and rated power; The formula for the initial population is: X = lb + rand * (ub - lb); Among them, X represents the initial population; rand represents a random matrix with dimension dim and size N.

3. The wind level suppression method based on adaptive particle swarm algorithm according to claim 1, characterized in that: In step S2, the energy storage parameters include at least unit power, capacity cost, charge and discharge efficiency, energy storage status, capital recovery rate, initial wind power, rated capacity and rated power.

4. The wind level suppression method based on adaptive particle swarm algorithm according to claim 1, characterized in that: In step S2, the formula of the objective function is: minF=C+αE penalty +βE curtail +C cycle ; C penalty +C curtail =αE penalty +βE curtail ; Among them, minF represents the minimum cost, that is, fitness; C represents the annual equivalent cost; E penalty Indicates the number of years with insufficient wind suppression; E curtail represents the annual amount of wind power abandoned; α represents E penalty The unit penalty cost of E curtail Unit penalty cost; C cycle represents the life cycle cost of the wind power storage system; C penalty Indicates the penalty for insufficient wind level suppression; C curtail It indicates the penalty for wind power abandonment.

5. The wind level suppression method based on adaptive particle swarm algorithm according to claim 1, characterized in that: In step S5, the formula of the inertia weight update rule is: ω2=1-ω1; Among them, ω2 represents the inertia weight after update; ω1 represents the inertia weight before update; sin() represents the sine function; gen represents the number of iterations; gen max Indicates the maximum number of iterations.

6. A wind level suppression method based on adaptive particle swarm algorithm as claimed in claim 5, characterized in that: In step S5, the population update paradigm is specifically: Set an inertia weight threshold A1, an inertia weight threshold A2, and a random value r in the range of [0,1]; When ω2<A1, the formula for population update is: When ω1>A2, the formula for population update is: When r<0.6 and ω1<A2, the formula for population update is: When r>0.6 and ω1<A2, the formula for population update is; Among them, V i,G+1 represents the search speed of the i-th particle of population X in the (G+1)th generation; X i,G represents the i-th particle of population X in the Gth generation; X i,G+1 represents the ith particle of population X in the (G+1)th generation; step represents the search step; Xmax represents the upper limit of the search space; Xmin represents the lower limit of the search space; d1 represents the population search speed control parameter in the second search stage, and d1∈{-1,1}; d2 represents the population search speed control parameter in the third search stage; d3 represents the population search speed control parameter in the fourth search stage; c represents the control parameter used to speed up the search in each search stage, and c=2; X gbest,G Represents the global optimal particle of population X in the Gth generation; r is used to cooperate with d1, d2, d3 and c to control the population search speed.

7. The wind level suppression method based on adaptive particle swarm algorithm according to claim 1, characterized in that: In step S5, the population update paradigm further includes: Use migration operators to facilitate information exchange among subpopulations: in, represents the global worst particle of subpopulation m in the Gth generation; represents the global optimal particle of subpopulation f in the Gth generation; represents the global worst particle of subpopulation f in the Gth generation; Represents the global optimal particle of subpopulation m in the Gth generation.

Images