Power distribution network energy storage system site selection and capacity configuration strategy based on multi-factor consideration

Abstract

A power distribution network energy storage system site selection and capacity configuration strategy based on multi-factor consideration comprises the steps that access of photovoltaic power generation and wind power generation and operation limitation of an energy storage system are considered, and a power distribution network system model is established; a node voltage stability index and a line overload rate index are introduced to comprehensively evaluate the stability of the power distribution network, and nodes with poor stability are selected as candidate nodes for installation of the energy storage system; comprehensively considering investment and operation cost, maintenance cost and annual electric energy loss of the power distribution network, and establishing a multi-objective optimization model of the power distribution network; setting node voltage, power flow balance, power output, battery capacity and charge and discharge constraint conditions of an energy storage system; and improving a particle swarm algorithm, and solving the power distribution network multi-objective optimization model to obtain the capacity of the energy storage system in the power distribution network. According to the method, the capacity of the energy storage system can be precisely planned and optimally configured, so that the optimal balance between the reliability and the economical efficiency of the power system is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of energy storage planning in a distribution network, and particularly relates to a siting and capacity configuration strategy for an energy storage system in a distribution network based on multi-factor consideration. Background Art

[0002] In many regions with relatively limited development levels, distribution networks generally face the problem of low voltage, which seriously restricts the stable operation and efficient development of the local power supply system, and further affects the further improvement of the regional economy. Therefore, using an energy storage system to address the low voltage problem has become an inevitable trend for the power grid to conform to the development of the times.

[0003] During the actual operation of a distribution network, due to the influence of natural conditions, the output power of distributed power sources exhibits significant intermittency and volatility characteristics. This not only causes instability in the frequency and voltage of the microgrid, leading to a decline in power quality and affecting the normal operation of various electrical equipment, but also increases the system operation risk and poses a potential threat to the safe and stable operation of the power grid. To address this series of problems, by scientifically and reasonably connecting an energy storage system to the distribution network, it can quickly respond and store and release energy when the output power of the distributed power source fluctuates, thereby effectively suppressing power fluctuations and further enhancing the stability and reliability of the distribution network.

[0004] In the existing research and practice on the planning of energy storage systems in distribution networks, in most cases, only the planning and design of the energy storage system itself are unilaterally focused on, and the overall situation of the new distribution network in terms of grid structure, load characteristics, distributed power source distribution, etc. is not comprehensively considered. This limitation makes it impossible for the energy storage system to fully meet the complex and changing operation requirements of the new distribution network during actual operation, resulting in its working efficiency being difficult to reach the ideal state and unable to fully exert the potential maximum effectiveness of the energy storage system in improving power quality and enhancing system stability in the new distribution network.

[0005] In addition, the traditional particle swarm optimization algorithm, as an algorithm widely used in solving various optimization problems, has achieved certain results in many fields. However, with the continuous increase in the complexity of the actual application scenario, some inherent defects of the algorithm gradually emerge. Among them, the problem of premature convergence is particularly prominent, that is, during the algorithm iteration process, the particle swarm is prone to falling into a local optimal solution and unable to find the global optimal solution. Summary of the Invention

[0006] In view of the limitations of the existing technologies in the planning of distribution network energy storage devices, the present invention provides a siting and capacity configuration strategy for a distribution network energy storage system based on multi-factor consideration. The present invention comprehensively considers the accommodation problem of new energy power generation devices accessing the distribution network, the stability problem of the distribution network, and the optimal economic operation of the distribution network. It can accurately plan and optimize the configuration of the capacity of the energy storage system to achieve the optimal balance between the reliability and economy of the power system.

[0007] The technical solution adopted by the present invention is as follows:

[0008] A siting and capacity configuration strategy for a distribution network energy storage system based on multi-factor consideration, comprising the following steps:

[0009] Step 1: Considering the access of photovoltaic power generation and wind power generation, as well as the operation limitations of the energy storage system, establish a distribution network system model;

[0010] Step 2: Introduce a node voltage stability index and a line overload rate index to comprehensively evaluate the stability of the distribution network, and select the nodes with relatively poor stability as the candidate nodes for installing the energy storage system;

[0011] Step 3: Comprehensively consider the investment and operation costs, maintenance costs, and annual power loss of the distribution network, and establish a multi-objective optimization model for the distribution network;

[0012] Step 4: Set the node voltage, power flow balance, power output, battery capacity, and charge and discharge constraints of the energy storage system;

[0013] Step 5: Improve the particle swarm optimization algorithm and solve the multi-objective optimization model of the distribution network to obtain the capacity of the energy storage system in the distribution network.

[0014] In the said Step 1, the distribution network system model is specifically as follows:

[0015] The present invention will select the IEEE 33-node system for simulation, thereby effectively verifying the effectiveness of the method proposed in the invention. The topology diagram of this system is as Figure 5 shown. The base voltage is 12.66 kV. The total load of the system is 3715 kW + j2300 kvar. And MATLAB is used for simulation, and 300 kW of wind power (WT) is connected to nodes 9 and 20, and 300 kW of photovoltaic power (PV) is connected to nodes 24 and 28 in this simulation. The access of wind power and photovoltaic power is to simulate the wind power generation system and photovoltaic system accessed by the distribution network in the actual line. The present invention will subsequently verify the influence of connecting the energy storage system to the nodes of the distribution network on the actual operation of the distribution network.

[0016] The distribution network system model includes an energy storage system operation strategy: when the difference between the power generation of distributed power sources and the load in the distribution network is greater than zero and the SOC value of the energy storage system is less than the maximum SOC value, the energy storage system starts to charge at this time to prevent the problem of voltage over-limit in the distribution network. When the difference between the power generation of distributed power sources and the load in the distribution network is less than zero and the SOC value of the energy storage system is greater than the minimum SOC value, the energy storage system discharges, thereby preventing the problem of low voltage in the distribution network.

[0017] In step 2, the node voltage stability index α i+1 can effectively reflect the voltage fluctuation of the node under different conditions. For the node voltage stability index, α i+1 the smaller it is, the more unstable the node is, and α i+1 The expression is shown in Equation (1):

[0018]

[0019] In Equation (1), represents the voltage of node i; P i+1 represents the active power of node i + 1; R i represents the equivalent resistance of line i; Q i+1 represents the reactive power of node i + 1; X i represents the equivalent reactance of line i.

[0020] In step 2, a dynamic thermal rating is introduced into the line overload rate. The dynamic thermal rating can monitor the temperature of the line, environmental conditions, and current load condition parameters in real time. To more accurately evaluate the current overload rate, under the dynamic thermal rating, the value of the line current is shown in Equation (2):

[0021]

[0022] In Equation (2), I DTR represents the actual current magnitude under the dynamic thermal rating; q c represents the convective heat dissipation of the wind; q r represents the radiative heat dissipation; q s represents the absorbed heat of the line currently irradiated by the sun; R represents the resistance of the line.

[0023] The current overload rate is an important indicator for measuring whether the current in the distribution network line or electrical equipment exceeds its rated carrying capacity. When the current overload rate in the line is high, it indicates that the stability of the line is poor, which also means that the risk faced by this line is greater than that of other lines. The current overload rate η m is shown in Equation (3):

[0024]

[0025] In Equation (3), In Represents the rated current of the line.

[0026] To avoid the limitations of a single indicator, the weighted average method will be used to combine the above two indicators into a comprehensive indicator;

[0027] First, for α i+1 and η m After normalization, we get α norm,i+1 and η norm,m ; α norm,i+1 Represents the node voltage stability index of node i after normalization; η norm,m Represents the node voltage stability index of branch m after normalization; Since the smaller the value of α norm,i+1 , the more unstable the distribution network is, and the larger the value of η norm,m , the more unstable the distribution network is. Since it needs to be unified in the stability evaluation index. Therefore, η norm,m is equivalent to 1 / η norm,m . The comprehensive stability evaluation index of the distribution network is shown in Equation (4):

[0028]

[0029] In Equation (4), λ i Represents the comprehensive stability evaluation index of the distribution network; ω1 and ω2 represent the weight coefficients.

[0030] In step 2, select the nodes with relatively poor stability as the candidate nodes for installing the energy storage system. The stability results are as Figure 6 shown. According to the definition, when λ i is smaller, the line is more unstable. That is, the stability of line 17 is the worst. Therefore, node 18 is selected as the candidate node for installing the energy storage device.

[0031] In step 3, to determine the capacity of the energy storage system under the optimal economic operation of the distribution network, it is necessary to establish the objective function of the multi-objective optimization model of the distribution network. The objective function takes into account minimizing the cost while ensuring that the power system provides a stable power supply to meet the needs of users. Therefore, the selection of the objective function includes: the investment and operation costs of distributed power sources and energy storage systems, the annual power loss of the distribution network, and the power generation income of distributed power sources, as shown in Equation (5):

[0032] F = min(F1 + F2 + F3)(5);

[0033] In Equation (5), F represents the objective function for optimizing the capacity of the energy storage system; F1 represents the average annual investment cost of the system; F2 represents the maintenance cost of the system within one year; F3 represents the annual power loss of the distribution network;

[0034] The investment cost of the distributed power source and energy storage system is related to its installed capacity. The average annual investment cost F1 of the system is shown in Equation (6):

[0035]

[0036] In Equation (6), P e represents the installed capacity of the energy storage system; C i e represents the installation cost of the energy storage system; n is the service life of the energy storage system (unit: year); r z is the discount rate; C OM (t) is the operating cost in the t-th year; t represents the year.

[0037] Due to the operating costs generated by the charging and discharging of the energy storage system, it is necessary to consider the charging and discharging conditions of the distribution network energy storage system. The maintenance cost F2 of the system in one year is shown in Equation (7):

[0038]

[0039] In Equation (7), k ch represents the maintenance cost coefficient during the charging of the energy storage system; k dis represents the maintenance cost coefficient during the discharging of the energy storage system; η e represents the annual maintenance cost coefficient of the energy storage system according to its capacity; P ch (t) represents the charging power of the energy storage system; P dis (t) represents the charging and discharging power of the energy storage system; T represents the total time in a day.

[0040] The annual power loss F3 of the distribution network is shown in Equation (8):

[0041]

[0042] In Equation (8), π L represents the average electricity price; P peak represents the maximum load value of the distribution network; P valley represents the minimum load value of the distribution network; π p represents the electricity price of the power grid during the peak load; π v represents the electricity price of the power grid during the low load; R i represents the equivalent resistance of node i; P i represents the active power of node i; Q i represents the reactive power of node i; U i represents the voltage of node i. In Step 4,

[0043] 1) Node voltage constraint:

[0044] 0.95U n≤U i ≤1.05U n (9);

[0045] In Equation (9), U n represents the rated value of the node voltage of the distribution network; U i represents the voltage value of node i;

[0046] 2) Power flow equation constraint:

[0047]

[0048] In Equation (10), P G,i represents the active power output at node i; P L,i represents the active power load at node i; Q G,i represents the reactive power output at node i; Q L,i represents the reactive power load at node i; G i , B i respectively represent the conductance and susceptance between nodes i and i + 1; U i represents the voltage of node i; U i+1 represents the voltage of node i + 1; n represents the number of nodes.

[0049] 3) Power output constraint:

[0050] To ensure that the charging and discharging powers of the distributed generation and energy storage systems operate within a specific range during time t. The specific constraint formula for the power output constraint is shown in Equation (11):

[0051]

[0052] In Equation (11), P pv (t) represents the power generation of the photovoltaic system; P wt (t) represents the power generation of the wind power generation; P ch (t) represents the charging power of the energy storage system; P dis (t) represents the discharging power of the energy storage system respectively represent the minimum and maximum values of the power generation of the photovoltaic system; respectively represent the minimum and maximum values of the power generation of the wind power generation; respectively represent the minimum and maximum values of the charging power of the energy storage system; respectively represent the minimum and maximum values of the discharging power of the energy storage system;

[0053] 4) Battery capacity constraint:

[0054] SOC min < SOC t < SOC max(12);

[0055] In formula (12), SOC min represents the minimum value of the state of charge of the battery in the energy storage system; SOC t represents the current state of charge of the battery in the energy storage system; SOC max represents the maximum value of the state of charge of the battery in the energy storage system;

[0056] 5) Charging and discharging constraints of the energy storage system:

[0057] Since the energy storage system cannot perform charging and discharging operations simultaneously, it needs to be constrained. When the energy storage system discharges, at this time P dis (t) = 0; at any time t, the sum of the charging and discharging powers is 0. The specific constraint formula is shown in formula (13):

[0058] P dis (t) × P ch (t) = 0 (13);

[0059] In formula (13), P dis (t) represents the discharging power of the energy storage system; P ch (t) represents the charging power of the energy storage system.

[0060] In step 5, an improvement of cross-genetic is made on the basis of the traditional particle swarm algorithm;

[0061] In the search space of the traditional particle swarm algorithm, each particle is a potential solution, and the particle will update its velocity and position according to its own and historical optimal positions. The specific formula is as follows:

[0062]

[0063]

[0064] In the formula: ω represents the inertia weight in this formula; i represents the particle number; c1, c2 represent the acceleration factors; d represents the dimension number; r1, r2 represent random numbers between (0, 1); t represents the number of iterations; p t id represents the component of the i-th particle in the optimal position in the d-th dimension at time t, x t id represents the component of the population optimal position in the d-th dimension at time t.

[0065] In the traditional particle swarm algorithm, problems existing in the algorithm itself may lead to premature convergence during the iteration process. Therefore, an improvement of the cross-genetic operation of the algorithm is made on this basis:

[0066] By introducing a method for measuring particle similarity and comprehensively considering the differences in the numerical values of multiple objective functions for the $i$-th particle and the $j$-th particle, their similarity $S$ ij is calculated as follows:

[0067]

[0068] where $M$ represents the number of objective functions; $f$ m $(x$ i ) represents the value of the $i$-th particle on the $m$-th objective function; $\beta_1$ and $\beta_2$ represent weight coefficients; $f$ m $(x$ j ) represents the value of the $j$-th particle on the $m$-th objective function; $x$ id represents the $d$-th dimensional component of the $i$-th particle in the solution space; $x$ jd represents the $d$-th dimensional component of the $j$-th particle in the solution space; $D$ represents the total number of dimensions in the solution space; $d$ represents the dimension number in the solution space; $m$ represents the objective function number.

[0069] Among them, the selection of the crossover point is judged according to the crossover degree of the particles. For particles with high similarity, crossover operations are performed on them. Specifically, it includes:

[0070] First, the crossover point is selected. In the improved particle swarm optimization algorithm, for particle pairs with high similarity, assuming the solution space dimension is $D$, calculate the sum of the absolute values of the differences in the parameter values of the two particles in each dimension and find the vicinity of the dimension with the largest difference as the crossover point.

[0071] After determining the crossover point, exchange part of the parameter values at the crossover point for the selected particle pair to generate new particles. Assume that particles $i$ and $j$ are two similar particle pairs, and the crossover point is in dimension $k$. So, crossover occurs at $x$ ik and $x$ jk . Because the dimension of the solution space is $D$, particle $i$ is arranged from $x$ i1 to $x$ id . So, particle $i$ is represented as $[x$ i1 , $x$ i2 ,.... $x$ ik ,..., $x$ iD . Similarly, particle $j$ is represented as

[0072] [$x$ j1 , $x$ j2 ,.... $x$ jk ,..., $x$ jD . After the crossover operation, the new particle $i'$ is $[x$ i1 , $x$ i2 ,.... $x$ jk ,..., $x$ iD and the new particle $j'$ is

[0073] [x j1 ,x j2 ,....x ik ,...,x jD , the x in the two new particles ik and x jk generate an exchange.

[0074] During the operation of the improved particle swarm algorithm, by calculating the fitness of the particles, including:

[0075] Mean μ of fitness: The mean fitness is the average of the fitnesses of all particles in the population, which reflects the overall fitness level of the population. Assume there are N particles in the population, and the fitness of the i-th particle is f i , then the calculation formula for the mean fitness μ is

[0076] Standard deviation σ:

[0077] The standard deviation measures the degree to which the fitness of the particles in the population deviates from the mean, reflecting the diversity of the population. Its calculation formula is When σ is large, it means that the fitness of the particles is highly dispersed, the population has rich diversity, and there are many different potential solutions; conversely, when σ is small, the fitness of the particles is relatively concentrated, the population has low diversity, and it may be close to the convergence state.

[0078] Range R of fitness: The range of fitness is the difference between the maximum fitness and the minimum fitness in the population, that is, R = f max -f min , where: f max is the maximum fitness of the particles in the population, and f min is the minimum fitness. The larger R is, the greater the difference in the fitness of the particles in the population, and there is still great exploration potential in the search space; the smaller R is, the closer the fitness of the particles is, and the algorithm may tend to converge. Thus, it reflects the diversity and convergence degree of the population.

[0079] Subsequently, the crossover probability P c is adjusted accordingly. For particles with fitness close to the mean and high population diversity, a lower crossover probability is assigned. For particles with fitness far from the mean or low population diversity, their crossover probabilities are increased. The calculation formula is as follows:

[0080]

[0081] In the formula: P cmin represents the minimum value of the crossover probability; P cmax represents the maximum value of the crossover probability; f i represents the fitness of the i-th particle; σ0 and R0 represent preset thresholds; P c(i) represents the dynamic adaptive crossover probability of the i-th particle.

[0082] Solve the multi-objective optimization model of the distribution network using the improved particle swarm optimization algorithm:

[0083] S1: Initialize the particle swarm: Randomly generate a certain number of particles in the feasible solution space. Each particle represents a potential solution to the multi-objective optimization problem of the distribution network. Each particle has two attributes, position and velocity. The position represents the parameter values of the current solution, and the velocity determines the moving direction and step size of the particle in the solution space.

[0084] S2: Calculate the objective function value of each particle: According to the specific objective function of the multi-objective optimization model of the distribution network, use formula (5) to calculate the objective function value corresponding to each particle.

[0085] S3: Evaluate the particle fitness: According to the objective function value F calculated by formula (5), in the present invention, the goal is to minimize the cost. Therefore, the smaller the objective function value F of the particle, the higher the fitness of the particle, which also means that the energy storage system configuration plan represented by the particle is better in the multi-objective optimization problem. Fitness reflects the quality of the particle in the multi-objective optimization problem.

[0086] S4: Compare the current fitness of each particle with its historical best fitness. If the current is better, update its historical best solution; at the same time, find the global best solution among the historical best solutions of all particles; specifically as follows:

[0087] In the improved particle swarm optimization algorithm, the fitness of the particle is measured by the objective function value. Assume that the fitness of the i-th particle at the current iteration t is f i (t), and its historical best fitness is f i,best . If f i (t) < f i,best , then update the historical best solution. That is, record the position and objective function value of the current particle as the historical best. Then find the global best solution. After all particles update their historical best solutions, comprehensively consider the historical best solutions of all particles and find the solution with the smallest objective function value as the global best solution.

[0088] S5: Calculate the particle similarity: Use formula (16) to calculate the similarity between particles to measure the similarity degree of particles in the solution space.

[0089] S6: Select the crossover particle pairs and crossover points according to the similarity:

[0090] First is the selection of the crossover point. In the improved particle swarm optimization algorithm, for particle pairs with high similarity, assume the solution space dimension is D, calculate the sum of the absolute values of the parameter differences of the two particles in each dimension Find the vicinity of the dimension with the largest difference as the crossover point.

[0091] S7: Calculate the dynamic adaptive crossover probability according to formula (17): Dynamically adjust the crossover probability based on information such as the diversity of the population and the fitness of the particles.

[0092] S8: Determine whether to perform crossover:

[0093] According to the calculated crossover probability P c (i), generate a random number r between (0, 1), r ∈ (0, 1). If r < P c (i), then perform the crossover operation on the selected pair of particles; if r ≥ P c (i), then do not perform the crossover operation.

[0094] S9: Perform the crossover operation:

[0095] If it is decided to perform crossover, then exchange part of the parameter values at the crossover point for the selected pair of particles to generate new particles. After determining the crossover point, exchange part of the parameter values at the crossover point for the selected pair of particles to generate new particles. Suppose particle i and j are two similar pairs of particles, and the crossover point is at dimension k. Particle i is represented as [x i1 , x i2 ,....x ik ,..., x iD , particle j is represented as [x j1 , x j2 ,....x jk ,..., x jD , and the new particle i' generated after the crossover operation is [x i1 , x i2 ,....x jk ,..., x iD , and the new particle j' is [x j1 , x j2 ,....x ik ,..., x jD .

[0096] S10: Local search optimization after crossover: Perform local search on the particles after crossover to further optimize their positions; first determine the neighborhood of the current position of the particle. Let the particle position vector X = [x1, x2, …, x n , x1, x2, …, x n be n particles in the vector X. Define its neighborhood as a series of positions after a small perturbation in each dimension. For example, in the i-th dimension, the new position x i ′ can be expressed as x i ′ = x i + δ, where δ is a small perturbation value and n represents the number of dimensions of the solution space.

[0097] Compare the objective function value obtained from formula (5) at the current particle position with the objective function values of the neighborhood positions. If there exists a neighborhood position with a better objective function value, update the particle position to that neighborhood position. Assume the objective function value corresponding to the current particle position X is F(X), and the objective function value corresponding to a certain neighborhood position X′ is F(X′). If F(X′) < F(X), then update the particle position to X′.

[0098] Continuously repeat this process until no better position can be found within the current neighborhood. At this time, it is considered that a local search optimization has been completed, and the particle position has been further optimized.

[0099] S11: Update the particle velocity and position:

[0100] For each particle, first calculate its new velocity in each dimension according to the velocity update formula (14) This new velocity comprehensively considers the current velocity of the particle, the difference between the particle's own historical best position and the current position, and the difference between the global best position of the population and the current position. Then, according to the position update formula (15), add the current position of the particle and the new velocity to obtain the new position of the particle in each dimension Thus, the position update of the particle in the solution space is realized, guiding the particle to search in a better solution space region.

[0101] S12: Recalculate the particle objective function value and fitness:

[0102] Based on the updated particle position, recalculate the objective function value F and the fitness evaluation according to formula (5). The smaller the objective function value F of the particle, the higher the fitness of the particle.

[0103] S13: Update the particle historical best solution and non-dominated solution set:

[0104] Compare the current fitness of the particle with the historical best fitness and update the historical best solution. In each iteration, the current fitness of the particle needs to be compared with the historical best fitness. Assume the fitness of the i-th particle at the current iteration t is f i (t), and its historical best fitness is f i,best . If f i (t) < f i,best, the historical optimal solution is updated. At this time, the position and objective function value of the current particle are recorded as the historical optimum. Meanwhile, the non-dominated solutions are added to the non-dominated solution set. In a multi-objective optimization problem, for two solutions A and B, if solution A is not worse than solution B in all objective functions and is better than solution B in at least one objective function, then solution A is said to dominate solution B. If a certain solution is not dominated by any other solution, then this solution is a non-dominated solution. In each iteration, for the solution corresponding to each particle, it is necessary to compare with the solutions of other particles to determine whether it is a non-dominated solution. The judgment formula can be abstractly expressed as: for solution X i (corresponding to the i-th particle) and X j (corresponding to the j-th particle), if for all objective functions m = 1, 2,..., M, where M represents the number of objective functions. There is f m (X i ) < f m (X j ), f m (X i ) represents the value of the solution X i of the i-th particle on the m-th objective function. f m (X j ) represents the value of the solution X j of the j-th particle on the m-th objective function.

[0105] And there is at least one objective function m', such that f m′ (X i ) < f m′ (X j ), f m′ (X i ) represents the value of the solution X i of the i-th particle on the m'-th objective function. f m′ (X j ) represents the value of the solution X j of the j-th particle on the m'-th objective function. Then X i dominates X j . If there is no such X i that dominates X j solution, then X i is a non-dominated solution. When it is determined that the solution corresponding to a certain particle is a non-dominated solution, it is added to the non-dominated solution set. The non-dominated solution set is the set of all non-dominated solutions. As the iteration progresses, the non-dominated solution set is continuously updated, and the final optimal Pareto solution set is selected from the non-dominated solution set.

[0106] S14: Select the global optimal solution of the population based on the density distance:

[0107] Calculate the crowding distance of particles in the non-dominated solution set, and select the particles with a larger crowding distance as the global optimal solution. The density calculation formula is as follows:

[0108]

[0109] In the formula: I i represents the crowding distance of the i-th particle in the non-dominated solution set. f m (i + 1) represents the value of the (i + 1)-th particle on the objective function m after sorting the non-dominated solution set in ascending order according to the value of a certain objective function m. f m (i - 1) represents the value of the (i - 1)-th particle on the objective function m after sorting the non-dominated solution set in ascending order according to the value of a certain objective function m. f m max and f m min are the maximum and minimum values of the m-th objective function on the non-dominated solution set.

[0110] S15: Determine whether the termination condition is satisfied: Check whether the preset termination conditions are reached, such as the maximum number of iterations, convergence of the objective function, etc. If satisfied, output the optimal Pareto solution set; otherwise, return to step 5 to continue the iteration.

[0111] Obtain the capacity of the energy storage system in the distribution network, specifically including:

[0112] The optimal cost is analyzed using the traditional particle swarm optimization algorithm and the improved particle swarm optimization algorithm respectively as shown in Table 1. Through data comparison, the optimal cost obtained by the improved particle swarm optimization algorithm is reduced by 47.37% compared with the traditional algorithm. This result shows the superiority of the improved multi-objective particle swarm optimization algorithm.

[0113] Table 1 Comparison of the optimal costs of the traditional particle swarm optimization algorithm and the improved particle swarm optimization algorithm

[0114]

[0115] Through the improved particle swarm optimization algorithm, the optimal capacity of the energy storage system is 23,229 yuan. According to the objective function F1, the capacity of the energy storage system is P e is 386.2 kWh.

[0116]

[0117] In the above formula: F1 represents the average annual investment cost. P e represents the installed capacity of the energy storage system; C i e represents the installation cost of the energy storage system; n is the service life of the energy storage system (unit: year); r z is the discount rate; COM (t) is the operating cost in the t-th year.

[0118] A siting and capacity configuration strategy for a distribution network energy storage system based on multi-factor consideration of the present invention has the following technical effects:

[0119] 1) When exploring the stability of the distribution network, the line overload rate is taken as the core factor. The traditional evaluation methods for line conditions often have certain limitations and are difficult to accurately calculate the real-time dynamic characteristics of the distribution network. In this context, it is particularly necessary and significant to introduce the dynamic thermal rating technology in the strategy of the present invention. This technology can closely combine the actual situations of various aspects such as real-time environmental parameters, load characteristics, and the physical properties of the line itself during the operation of the distribution network, and make a more accurate judgment on the actual carrying capacity and operating state of the line. Therefore, by constructing a comprehensive evaluation index system based on the dynamic thermal rating technology, the lines in the unstable operating state can be accurately screened out in the distribution network.

[0120] 2) In the process of optimizing the capacity of the energy storage system, the present invention not only ensures the fundamental principle of stable power supply of the power system to meet the continuously growing power demand of users. And considering the economic factors in the construction and operation of the power system, the present invention controls the cost to the minimum on the premise of ensuring stable power supply. This includes the analysis of the life cycle cost of the energy storage system, including equipment purchase cost, installation and commissioning cost, operation and maintenance cost, and decommissioning and disposal cost, etc. By comprehensively applying optimization algorithms and models and fully considering the uncertain factors in the operation process of the power system, such as the intermittency of renewable energy power generation and the dynamic changes of load demand, etc., the capacity of the energy storage system is accurately planned and optimally configured to achieve the optimal balance between the reliability and economy of the power system.

[0121] 3) Based on the in-depth study of the traditional particle swarm algorithm, the present invention proposes an improved strategy to optimize the crossover and genetic operations of the algorithm. By introducing an adaptive crossover probability and genetic operator, the algorithm can dynamically adjust the intensity of crossover and genetics according to the evolutionary state of the population during the iteration process, enhancing the global search ability of the algorithm and the ability to jump out of the local optimal solution. At the same time, combined with specific practical problems, a detailed theoretical analysis and simulation experiment verification are carried out on the improved algorithm. The experimental results show that the improved particle swarm algorithm shows significant advantages in terms of convergence speed, optimization accuracy, and stability.

[0122] 4) First, through the analysis of the stability of the distribution network, the present invention selects two dimensions to deeply evaluate the stability of the distribution network system. Through the systematic analysis of the distribution network stability, the obtained results will be established as the key index basis for the access of distributed power sources and energy storage systems to the distribution network, ensuring that the distributed power sources and energy storage systems can be reasonably, efficiently, safely and reliably accessed in the distribution network environment. By combining the two indicators, the weak nodes of the distribution network can be more accurately judged, so as to realize the most efficient access of the energy storage system to the distribution network.

[0123] 5) The objective function for selecting the capacity of the energy storage system proposed by the strategy of the present invention combines factors such as the investment costs of the distributed power source and the energy storage system, the operation and maintenance costs for one year of the system, the power loss of the distribution network, and the loss cost caused by the load valley difference. Considering various factors, the comprehensive cost after the energy storage system is accessed to the distribution network can effectively judge how to achieve the stable and efficient operation of the distribution network after accessing the energy storage system while saving costs.

[0124] 6) When considering the stability of the distribution network, the present invention strategy introduces the dynamic thermal rating technology to analyze the line overload rate. This technology can more accurately reflect the actual current-carrying capacity of the line. The traditional method uses a fixed thermal rating, which is determined based on the worst environmental conditions and does not consider the actual environmental changes. The dynamic thermal rating technology uses real-time meteorological data, such as temperature, wind speed, sunlight, etc., to more accurately reflect the actual current-carrying capacity of the line. And it takes into account the changes in line characteristics. For example, during the long-term operation of the line, its physical characteristics such as resistance and heat transfer coefficient will change, affecting the current-carrying capacity. The dynamic thermal rating technology monitors the line parameters in real time, can timely reflect these changes, and accurately evaluate the overload rate. Description of the Drawings

[0125] Figure 1 It is a schematic diagram of the operation strategy of the energy storage system.

[0126] Figure 2 It is a flow chart of the improved particle swarm optimization algorithm for solving.

[0127] Figure 3 It is a comparison chart of the optimal cost iteration curves of the traditional particle swarm optimization algorithm and the improved particle swarm optimization algorithm.

[0128] Figure 4(a) is the system node voltage curve under Scenario 1;

[0129] Figure 4(b) is the system node voltage curve under Scenario 2;

[0130] Figure 4(c) is the system node voltage curve under Scenario 3.

[0131] Figure 5 It is the topology diagram of the IEEE33-node system.

[0132] Figure 6 Schematic diagram of line stability indicators. DETAILED DESCRIPTION

[0133] The present invention proposes a distribution network energy storage system site selection and capacity configuration strategy based on multiple factors, which comprehensively considers the distribution network's own conditions, the voltage crossing caused by the access of distributed power sources, and the economic efficiency of distribution network operation. The following is a methodological description of this strategy.

[0134] The distribution network energy storage system site selection and capacity configuration strategy based on multiple factors includes the following steps:

[0135] Step (1): Considering the access of photovoltaic power generation and wind power generation, as well as the operation limitations of the energy storage system, a distribution network system model is established;

[0136] Step (2): Considering the differences in the stability of different nodes in the distribution network, it is necessary to analyze the stability of the distribution network. The node voltage stability index and the line overload rate index are introduced to comprehensively evaluate the stability of the distribution network. After evaluating the stability of different nodes in the distribution network, the nodes with poor stability are selected as candidate nodes for the installation of the energy storage system.

[0137] Step (3): Taking into account the investment and operating costs, maintenance costs, and annual power loss of the distribution network, a multi-objective optimization model for the distribution network is established to optimize the capacity of the energy storage system; at the same time, it is considered to minimize the cost while ensuring that the power system provides a stable power supply to meet user needs.

[0138] Step (4): In order to ensure the normal operation of the distribution network system, set the node voltage, power flow balance, power output, battery capacity and energy storage system charging and discharging constraints;

[0139] Step (5): Improve the particle swarm algorithm and improve the particle swarm algorithm process as follows: Figure 2 As shown; and the multi-objective optimization model of the distribution network is solved to obtain the capacity of the energy storage system in the distribution network.

[0140] Figure 1 This is the operation strategy of the energy storage system in the new distribution network. When the difference between the power generation and load of the distributed power generation in the distribution network is greater than zero and the SOC value of the energy storage system is less than the maximum SOC value, the energy storage system starts to charge to prevent the distribution network from exceeding the voltage limit. When the power generation and load difference of the distributed power generation in the distribution network is less than zero and the SOC value of the energy storage system is greater than the minimum SOC value, the energy storage system discharges to prevent the distribution network from having low voltage problems.

[0141] Node voltage stability index α i+1By comprehensively considering the electrical parameters on the line, it can effectively reflect the voltage fluctuation of the node under different conditions. For this index, α i+1 The smaller it is, the more unstable the node is. α i+1 As shown in formula (1):

[0142]

[0143] Introduce the dynamic thermal rating technology into the line overload rate. This technology can real-time monitor parameters such as the temperature of the line, environmental conditions, and the current load situation. It can more accurately evaluate the current overload rate. Under the dynamic thermal rating, the value of the line current is as shown in formula (2):

[0144]

[0145] The current overload rate is an important indicator for measuring whether the current in the distribution network line or electrical equipment exceeds its rated carrying capacity. When the current overload rate in the line is relatively high, it indicates that the stability of the line is poor, which also means that the risk faced by this line is greater compared to other lines. The current overload rate η m As shown in formula (3):

[0146]

[0147] To avoid the limitations of a single index, the present invention will adopt the weighted average method to combine the above two indexes into a comprehensive index. Since the performance of the distribution network is affected by various factors, doing so can not only consider the possible low voltage problems in the distribution network, but also consider the possible impacts of on-site environmental factors on the line, thus having a more comprehensive assessment of the condition of the distribution network. To conduct a comprehensive assessment, first, α i+1 and η m are normalized to obtain α norm,i+1 and η norm,m . Since the smaller the value of α norm,i+1 indicates that the distribution network is more unstable, and the larger the value of η norm,m indicates that the distribution network is more unstable. Since it needs to be unified in the stability evaluation index. So by equating η norm,m to 1 / η norm,m . Therefore, the comprehensive stability evaluation index of the distribution network is as shown in formula (4):

[0148]

[0149] To determine the capacity of the energy storage device under the optimal economic operation of the distribution network, it is necessary to establish the objective function of the multi-objective optimization model of the distribution network. This objective function mainly considers minimizing the cost while ensuring that the power system provides a stable power supply to meet the needs of users. Therefore, the selection of the objective function will be described from aspects such as the investment and operation costs of distributed power sources and energy storage systems, annual power losses, and the power generation income of distributed power sources. As shown in formula (5):

[0150] F = min(F1 + F2 + F3)(5);

[0151] The investment cost of distributed power sources and energy storage systems is related to their installed capacity. Therefore, the average annual investment cost F1 of the system is as shown in formula (6):

[0152]

[0153] Where: P e represents the installed capacity of the energy storage system; C i e represents the installation cost of the energy storage system; n is the service life of the energy storage system (unit: year); r z is the discount rate; C OM (t) is the operation cost in the t-th year.

[0154] Since there are operation costs generated by the charging and discharging of the energy storage system, it is necessary to consider the charging and discharging conditions of the energy storage system of the distribution network. Therefore, the working maintenance cost F2 of the system in one year is as shown in formula (7):

[0155]

[0156] Where: k ch represents the maintenance cost coefficient when the energy storage system is charging; k dis represents the maintenance cost coefficient when the energy storage system is discharging. η e represents the annual maintenance cost coefficient of the energy storage system according to its capacity.

[0157] Calculating the power loss of the distribution network can effectively evaluate the energy efficiency and loss degree of the distribution system. By reducing the power loss, the system efficiency can be effectively improved and the economic loss can be reduced. Considerations will be made from the network loss cost and the loss cost caused by the load valley difference. Network loss refers to the power loss within the power system. In the present invention, the network loss of the distribution network is reflected in the form of power loss, and the economy lost due to the network loss is calculated accordingly.

[0158] During peak load periods, more power supply is required, resulting in higher costs. During low load periods, although the required power supply is smaller, there is a difference in power generation costs compared to peak times. Therefore, the present invention introduces the concept of the peak-valley difference in load. Reducing the peak-valley difference in the distribution network can not only reduce losses but also improve the stability of the system. Thus, the annual power loss F3 of the distribution network is as shown in Equation (8):

[0159]

[0160] Where: π L represents the average electricity price; P peak represents the maximum load value of the distribution network; P valley represents the minimum load value of the distribution network; π p represents the grid electricity price during peak load; π v represents the grid electricity price during valley load.

[0161] The multi-objective optimization model proposed in the present invention needs to be restricted by some constraint conditions. The purpose of setting the constraint conditions is to make it operate within the specified range. The specific constraint conditions are as follows.

[0162] 1) Node voltage constraint:

[0163] To ensure that the voltage fluctuates within the specified standard range, it needs to be constrained. The specific constraint formula is as shown in Equation (9):

[0164] 0.95U n ≤U i ≤1.05U n (9);

[0165] Where: U n represents the rated value of the node voltage of the distribution network; U i represents the voltage value of node i.

[0166] 2) Power flow equation constraint:

[0167] The specific constraint formula is as shown in Equation (10):

[0168]

[0169] Where: P G,i represents the active power output at node i; P L,i represents the active power load at node i; Q G,i represents the reactive power output at node i; Q L,i represents the reactive power load at node i; G i 、B i represent the conductance and susceptance between node i and i + 1.

[0170] 3) Power output constraint:

[0171] To ensure that the charging and discharging power of the distributed generation and energy storage system operates within a specific range at time t. The specific constraint formula is shown in formula (11):

[0172]

[0173] 4) Battery capacity constraint:

[0174] To extend the service life of the device to achieve the optimal economic constraint. It is necessary to limit its charging and discharging to extend the service life of the device. The specific constraint formula is shown in formula (12):

[0175] SOC min <SOC t <SOC max (12);

[0176] 5) Energy storage system charging and discharging constraint:

[0177] Since the energy storage system cannot perform charging and discharging operations simultaneously, it is necessary to constrain it. For example, when the energy storage system discharges, at this time P dis (t) = 0. Therefore, at any time t, the sum of the charging and discharging power is 0. The specific constraint formula is shown in formula (13):

[0178] P dis (t) × P ch (t) = 0(13);

[0179] As Figure 2 shown, by improving the cross-genetic algorithm based on the traditional particle swarm optimization algorithm, the algorithm can better handle multi-objective problems and effectively improve the solution efficiency and quality. Thus, it can better solve the capacity of the energy storage system installed in the distribution network. In the traditional particle swarm optimization algorithm, the problems existing in the algorithm itself may lead to premature convergence during the iteration process. Therefore, the present invention improves the cross-genetic operation of the algorithm on this basis.

[0180] The optimal cost is analyzed using the traditional particle swarm optimization algorithm and the improved particle swarm optimization algorithm respectively, and the iteration curve is as Figure 3 shown. Through data comparison, the optimal cost obtained by the improved particle swarm optimization algorithm is reduced by 47.37% compared with the traditional algorithm. This result shows the superiority of the improved multi-objective particle swarm optimization algorithm.

[0181] To verify the effectiveness of the site selection method mentioned above, the present invention will select the following three scenarios for comparison:

[0182] Scenario 1: Without connecting the energy storage device.

[0183] Scenario 2: Select the nodes corresponding to the lines with medium stability in the distribution network to connect the energy storage device. For example, Node 22.

[0184] Scenario 3: Select the nodes corresponding to the lines with the worst stability in the distribution network to connect the energy storage device. For example, Node 17.

[0185] Figures 4(a) to 4(c) It reflects the variation of the voltage of each node with time under different scenarios. By comparing Scenario 1 with Scenarios 2 and 3, it can be found that after connecting the energy storage device, the voltage fluctuations of each node are reduced. And the voltage fluctuations of each node in Scenario 3 are smaller. This shows that the distribution network is more stable and also shows the feasibility of the site selection method.

Claims

1. The location and capacity configuration strategy of the distribution network energy storage system based on multiple factors is characterized by The following steps are involved: Step 1: Considering the access of photovoltaic power generation and wind power generation, as well as the operating limitations of the energy storage system, a distribution network system model is established; Step 2: The node voltage stability index and line overload rate index are introduced to comprehensively evaluate the stability of the distribution network, and nodes with poor stability are selected as candidate nodes for energy storage system installation; Step 3: Considering the investment and operation costs, maintenance costs, and annual power loss of the distribution network, a multi-objective optimization model for the distribution network is established; Step 4: Set node voltage, power flow balance, power output, battery capacity, and energy storage system charging and discharging constraints; Step 5: Improve the particle swarm algorithm and solve the multi-objective optimization model of the distribution network to obtain the capacity of the energy storage system in the distribution network.

2. According to the distribution network energy storage system site selection and capacity configuration strategy based on multiple factors as described in claim 1, it is characterized by : In step 1, the distribution network system model includes an energy storage system operation strategy: when the difference between the power generation and load of the distributed power sources in the distribution network is greater than zero and the SOC value of the energy storage system is less than the maximum SOC value, the energy storage system starts to charge; when the difference between the power generation and load of the distributed power sources in the distribution network is less than zero and the SOC value of the energy storage system is greater than the minimum SOC value, the energy storage system discharges.

3. According to the distribution network energy storage system site selection and capacity configuration strategy based on multiple factors as described in claim 1, it is characterized in that In step 2, the node voltage stability index α i+1 It can effectively reflect the voltage fluctuation of nodes under different conditions. For the node voltage stability index, α i+1 The smaller the node, the more unstable it is. i+1 The expression is shown in formula (1): In formula (1), represents the voltage of node i; P i+1 represents the active power of node i+1; R i represents the equivalent resistance of line i; Q i+1 represents the reactive power of node i+1; X i represents the equivalent reactance of line i; The dynamic thermal set value is introduced into the line overload rate. Under the dynamic thermal set value, the value of the line current is shown in formula (2): In formula (2), I DTR Indicates the actual current size under dynamic thermal setting value; q c represents the convective heat dissipation of wind; q r represents radiation heat dissipation; q s Indicates the amount of heat currently absorbed by the line due to solar radiation; R indicates the resistance of the line; When the current overload rate in the line is high, it means that the stability of the line is poor, which means that the risk faced by the line is greater than that of other lines; the current overload rate η m As shown in formula (3): In formula (3), I n Indicates the rated current of the line; In order to avoid the limitation of a single indicator, the node voltage stability index and line overload rate index will be combined into a comprehensive index using the weighted average method; First, we need to i+1 and η m After normalization, we get α norm,i+1 and η norm,m ; α norm,i+1 represents the node voltage stability index of node i after normalization; η norm,m represents the node voltage stability index of branch m after normalization; due to α norm,i+1 The smaller the value of is, the more unstable the distribution network is. norm,m The larger the value of indicates, the more unstable the distribution network is. Since it needs to be unified in the stability evaluation index, η norm,m Equivalent to 1 / η norm,m ; The comprehensive stability evaluation index of the distribution network is shown in formula (4): In formula (4), λ i represents the comprehensive stability evaluation index of the distribution network; ω1 and ω2 represent weight coefficients.

4. The distribution network energy storage system site selection and capacity configuration strategy based on multiple factors according to claim 3 is characterized by: Select nodes with poor stability as candidate nodes for energy storage system installation. i The smaller it is, the less stable the line is; that is, the stability is the worst.

5. The distribution network energy storage system site selection and capacity configuration strategy based on multiple factors according to claim 1 is characterized by: In step 3, in order to determine the energy storage system capacity of the distribution network under the optimal economic operation condition, it is necessary to establish the objective function of the distribution network multi-objective optimization model. The objective function takes into account the minimum cost while ensuring that the power system provides a stable power supply to meet the needs of users. Therefore, the selection of the objective function includes: the investment and operating costs of the distributed power source and the energy storage system, the annual power loss of the distribution network, and the distributed power generation income, as shown in formula (5): F = min (F1 + F2 + F3) (5); In formula (5), F represents the objective function of capacity optimization of the energy storage system; F1 represents the average investment cost of the system in one year; F2 represents the maintenance cost of the system in one year; and F3 represents the annual power loss of the distribution network.

6. The distribution network energy storage system site selection and capacity configuration strategy based on multiple factors according to claim 5 is characterized by: The investment cost of distributed power generation and energy storage system is related to its installation capacity. The average investment cost F1 of the system in one year is shown in formula (6): In formula (6), P e Represents the installed capacity of the energy storage system; represents the installation cost of the energy storage system; n is the service life of the energy storage system; r z is the discount rate; C OM (t) is the operating cost in year t; t represents the year; Since the energy storage system has operating costs generated by charging and discharging, it is necessary to consider the charging and discharging of the distribution network energy storage system. The maintenance cost F2 of the system within one year is shown in formula (7): In formula (7), k ch k represents the maintenance cost coefficient when the energy storage system is charged; dis Represents the maintenance cost coefficient of the energy storage system when discharging; η e P represents the annual maintenance cost coefficient of the energy storage system according to capacity; ch (t) represents the charging power of the energy storage system; P dis (t) represents the charging and discharging power of the energy storage system; T represents the total time in a day; The annual power loss F3 of the distribution network is shown in formula (8): In formula (8), π L represents the average electricity price; P peak Indicates the maximum load value of the distribution network; P valley Represents the minimum load value of the distribution network; π p Represents the grid electricity price during peak load period; π v Represents the grid electricity price during the low load period; R i represents the equivalent resistance of node i; P i represents the active power of node i; Q i represents the reactive power of node i; U i represents the voltage at node i.

7. The distribution network energy storage system site selection and capacity configuration strategy based on multiple factors according to claim 1 is characterized by: In step 4, 1) Node voltage constraints: 0.95U n ≤U i ≤1.05U n (9); In formula (9), U n Indicates the rated value of the distribution network node voltage; U i represents the voltage value of node i; 2) Power flow equation constraints: In formula (10), P G,i represents the active power output at node i; P L,i represents the active load at node i; Q G,i represents the reactive power output at node i; Q L,i represents the reactive load at node i; G i , B i Respectively represent the conductance and susceptance between nodes i and i+1; U i represents the voltage of node i; U i+1 represents the voltage of node i+1; n represents the number of nodes; 3) Power output constraints: In order to ensure that the charging and discharging power of distributed generation and energy storage system operates within a specific range at time t, the specific constraint formula of power output constraint is shown in formula (11): In formula (11), P pv (t) represents the power generation of the photovoltaic system; P wt (t) represents the power P of wind power generation ch (t) represents the charging power of the energy storage system; P dis (t) represents the discharge power of the energy storage system They represent the minimum and maximum power generation of the photovoltaic system respectively; Respectively represent the minimum and maximum values ​​of wind power generation; Respectively represent the minimum and maximum charging power of the energy storage system; Respectively represent the minimum and maximum discharge power of the energy storage system; 4) Battery capacity constraints: SOC min <SOC t <SOC max (12); In formula (12), SOC min Indicates the minimum value of the battery state of charge in the energy storage system; SOC t Indicates the current state of charge of the battery in the energy storage system; SOC max Indicates the maximum value of the battery state of charge in the energy storage system; 5) Energy storage system charging and discharging constraints: Since the energy storage system cannot be charged and discharged at the same time, it needs to be constrained. When the energy storage system is discharging, P dis (t) = 0; at any time t, the sum of the charge and discharge power is 0; its specific constraint formula is shown in formula (13): P dis (t)×P ch (t)=0(13); In formula (13), P dis (t) represents the discharge power of the energy storage system; P ch (t) represents the charging power of the energy storage system.

8. The distribution network energy storage system site selection and capacity configuration strategy based on multiple factors according to claim 1 is characterized by: In step 5, the crossover genetic algorithm is improved on the basis of the traditional particle swarm algorithm; By introducing the particle similarity measurement method, the differences in the values ​​of multiple objective functions of particles are comprehensively considered. For the i-th particle and the j-th particle, their similarity S ij The calculation formula is as follows: Where: M represents the number of objective functions; f m (x i ) represents the value of the ith particle on the mth objective function; β1 and β2 represent weight coefficients; f m (x j ) represents the value of the jth particle on the mth objective function; x id represents the d-dimensional component of the ith particle in the solution space; x jd represents the component of the dth dimension of the jth particle in the solution space; D represents the total number of dimensions of the solution space; d represents the dimension number in the solution space; m represents the objective function number; Among them, the selection of the crossover point is judged according to the crossover degree of the particles. For particles with high similarity, crossover operations are performed on them. Specifically, it includes: The first is the selection of intersection points. In the improved particle swarm algorithm, for particle pairs with high similarity, the dimension of the solution space is set to D, and the sum of the absolute values ​​of the parameter differences of the two particles in each dimension is calculated. Find the intersection point near the dimension with the largest difference; After the intersection is determined, some parameter values ​​of the selected particle pair are exchanged at the intersection to generate new particles; assuming that particles i and j are two similar particle pairs, the intersection point is in dimension k, so in x ik and x jk Because the dimension of the solution space is D, particle i is divided by x i1 Arrange to x id , so particle i is represented by [x i1 ,x i2 ,....x ik ,...,x iD ], similarly, particle j is represented by [x j1 ,x j2 ,....x jk ,...,x jD ] After the crossover operation, the new particle i' is generated as [x i1 ,x i2 ,....x jk ,...,x iD ]The new particle j' is [x j1 ,x j2 ,....x ik ,...,x jD ], of the two new particles x ik and x jk Produce exchange; During the operation of the improved particle swarm optimization algorithm, by calculating the fitness of the particles, it includes: Mean value μ of fitness: The mean value of fitness is the average value of the fitness of all particles in the population, which reflects the overall fitness level of the population; assuming that there are N particles in the population, the fitness of the i-th particle is f i , then the calculation formula of the fitness mean μ is Standard deviation σ: The standard deviation measures the degree to which the fitness of particles in the population deviates from the mean, reflecting the diversity of the population; its calculation formula is When σ is large, it means that the particle fitness is highly dispersed, the population diversity is rich, and there are many different potential solutions; on the contrary, when σ is small, the particle fitness is more concentrated, the population diversity is low, and it may be close to the convergence state; The extreme difference of fitness R: The extreme difference of fitness is the difference between the maximum fitness and the minimum fitness in the population, that is, R = f max -f min , where: f max is the maximum fitness of particles in the population, f min is the minimum fitness; the larger R is, the greater the fitness difference of particles in the population is, and the search space still has great exploration potential; the smaller R is, the closer the fitness of particles is, and the algorithm may tend to converge; thus reflecting the diversity and convergence degree of the population; Then the crossover probability P c Make corresponding adjustments. For particles whose fitness is close to the mean and whose population diversity is high, give them a lower crossover probability; for particles whose fitness is far from the mean or whose population diversity is low, increase their crossover probability. The calculation formula is as follows: Where: P cmin represents the minimum value of the crossover probability; P cmax represents the maximum value of the crossover probability; f i represents the fitness of the i-th particle; σ0 and R0 represent the preset thresholds; P c (i) represents the dynamic adaptive crossover probability of the i-th particle.

9. The distribution network energy storage system site selection and capacity configuration strategy based on multiple factors according to claim 8 is characterized by: The improved particle swarm optimization algorithm solves the multi-objective optimization model of the distribution network, including the following steps: S1: Initialize the particle swarm: Randomly generate a certain number of particles in the feasible solution space. Each particle represents a potential solution to the multi-objective optimization problem of the distribution network. Each particle has two attributes, position and velocity. The position represents the parameter value of the current solution, and the velocity determines the moving direction and step size of the particle in the solution space. S2: Calculate the objective function value of each particle: According to the specific objective function of the multi-objective optimization model of the distribution network, use formula (5) to calculate the objective function value corresponding to each particle. S3: Evaluate the particle fitness: According to the objective function value F calculated by formula (5), the goal is to minimize the cost. Therefore, the smaller the objective function value F of the particle, the higher the fitness of the particle, which means that the energy storage system configuration scheme represented by the particle is better in the multi-objective optimization problem. S4: Compare the current fitness of each particle with its historical best fitness. If the current is better, update its historical best solution. At the same time, find the global best solution among the historical best solutions of all particles. Specifically as follows: In the improved particle swarm algorithm, the fitness of a particle is measured by the value of the objective function; assuming that the fitness of the i-th particle at the current iteration t is f i (t), whose historical optimal fitness is f i,best If f i (t) <f i,best , then update the historical optimal solution; that is, record the current particle position and objective function value as the historical optimal solution; then find the global optimal solution. After all particles have updated their historical optimal solutions, combine the historical optimal solutions of all particles and find the solution with the smallest objective function value as the global optimal solution; S5: Calculate the particle similarity: Use formula (16) to calculate the similarity between particles to measure the similarity degree of particles in the solution space. S6: Select the crossover particle pair and crossover point according to the similarity. The first is the selection of intersection points. In the improved particle swarm algorithm, for a pair of particles with high similarity, assuming that the dimension of the solution space is D, the sum of the absolute values ​​of the parameter differences of the two particles in each dimension is calculated. Find the intersection point near the dimension with the largest difference; S7: Calculate the dynamic adaptive crossover probability according to formula (17): Dynamically adjust the crossover probability according to information such as the diversity of the population and the fitness of the particles. S8: Judge whether to perform crossover: According to the calculated crossover probability P c (i) Generate a random number r between (0,1), r∈(0,1); if r <P c (i) Then perform a crossover operation on the selected particle pair; if r ≥ P c (i), no crossover operation is performed; S9: Execute the crossover operation: If the intersection is determined, some parameter values ​​of the selected particle pairs are exchanged at the intersection to generate new particles; after the intersection is determined, some parameter values ​​of the selected particle pairs are exchanged at the intersection to generate new particles; suppose particles i and j are two similar particle pairs, and the intersection point is in dimension k; particle i is represented by [x i1 ,x i2 ,....x ik ,...,x iD ], particle j is represented by [x j1 ,x j2 ,....x jk ,...,x jD ], after the crossover operation, the new particle i' is [x i1 ,x i2 ,....x jk ,...,x iD ], the new particle j' is [x j1 ,x j2 ,....x ik ,...,x jD ]; S10: Local search optimization after crossover: Perform local search on the particles after crossover to further optimize their positions; first determine the neighborhood of the current position of the particle, and set the particle position vector X = [x1, x2, ..., x n ],x1,x2,…,x n is the number of particles in the vector X; the neighborhood is defined as a series of positions after a small perturbation in each dimension; in the i-th dimension, the new position x i ′ can be expressed as x i ′=x i +δ, where δ is a small perturbation value and n represents the number of dimensions of the solution space; Compare the objective function value obtained by the current particle position in formula (5) with the objective function value of the neighborhood position. If there is an objective function value of a certain neighborhood position that is better, update the particle position to that neighborhood position. Assume that the objective function value corresponding to the current particle position X is F(X), and the objective function value corresponding to a certain neighborhood position X′ is F(X′). If F(X′) < F(X), then update the particle position to X′. Continuously repeat this process until no better position can be found in the current neighborhood. At this time, it is considered that a local search optimization has been completed and the particle position has been further optimized. S11: Update the particle velocity and position: For each particle, first calculate its new velocity in each dimension according to the velocity update formula This new speed comprehensively considers the particle's current speed, the gap between the particle's own historical optimal position and the current position, and the gap between the population's global optimal position and the current position; then, the particle's current position is updated according to the position update formula. With new speed Add them together to get the new position of the particle in each dimension This enables the particle's position in the solution space to be updated, guiding the particle to search for a better solution space area; S12: Recalculate the particle objective function value and fitness: Based on the updated particle position, recalculate the objective function value F and the fitness evaluation according to formula (5). The smaller the objective function value F of the particle, the higher the fitness of the particle. S13: Update the particle historical best solution and non-dominated solution set: Compare the current fitness of the particle with the historical optimal fitness and update the historical optimal solution. In each iteration, compare the current fitness of the particle with the historical optimal fitness. Assume that the fitness of the i-th particle at the current iteration t is f i (t), whose historical optimal fitness is f i,best If f i (t) <f i,best , then update the historical optimal solution; at this time, record the current particle position and objective function value as the historical optimal solution; at the same time, add the non-dominated solution to the non-inferior solution set. In the multi-objective optimization problem, for two solutions A and B, if solution A is not worse than solution B in all objective functions, and is better than solution B in at least one objective function, then solution A is said to dominate solution B; if a solution is not dominated by any other solution, then this solution is a non-dominated solution; in each iteration, for the solution corresponding to each particle, it must be compared with the solutions of other particles to determine whether it is a non-dominated solution; the judgment formula can be abstractly expressed as: for solution X i (corresponding to the i-th particle) and X j (corresponding to the jth particle), if for all objective functions m = 1, 2, ..., M, M represents the number of objective functions; there are f m (X i ) <f m (X j ), f m (X i ) represents the solution X corresponding to the i-th particle i The value of the mth objective function; f m (X j ) represents the solution X corresponding to the jth particle j The value of the mth objective function; And there exists at least one objective function m′ such that f m′ (X i ) <f m′ (X j ), f m′ (X i ) represents the solution X corresponding to the i-th particle i The value of the m′th objective function; f m′ (X j ) represents the solution X corresponding to the jth particle j The value of the m′th objective function; then X i DominateX j ; If there is no such X i DominateX j The solution is X i It is a non-dominated solution. When the solution corresponding to a particle is determined to be a non-dominated solution, it is added to the non-inferior solution set. The non-inferior solution set is the set of all non-dominated solutions. As the iteration proceeds, the non-inferior solution set is continuously updated, and the final optimal Pareto solution set is selected from the non-inferior solution set. S14: Select the global best solution of the population based on the crowding distance: Calculate the crowding distance of the particles in the non-dominated solution set, and select the particle with a larger crowding distance as the global best solution. The density calculation formula is as follows: Where: I i represents the dense distance of the i-th particle in the non-inferior solution set; f m (i+1) indicates the value of the i+1th particle on the objective function m after sorting the values ​​of the objective function m from small to large in the non-inferior solution set; f m (i-1) represents the value of the i-1th particle on the objective function m after sorting the non-inferior solution set from small to large according to the value of a certain objective function m; and is the maximum and minimum value of the mth objective function on the non-inferior solution set; S15: Determine whether the termination conditions are met: Check whether the preset termination conditions are met, such as the maximum number of iterations, convergence of the objective function, etc.; if satisfied, output the optimal Pareto solution set; otherwise, return to continue iteration.

10. The distribution network energy storage system site selection and capacity configuration strategy based on multiple factors according to claim 6, characterized in that: In step 5, according to the objective function F1, the capacity of the energy storage system in the distribution network can be obtained.

Images