Electric vehicle photovoltaic charging station multi-dimensional target collaborative optimization scheduling method fused with improved particle swarm intelligent optimization mechanism

Abstract

The invention relates to the technical field of electric vehicle charging and renewable energy source integration. The invention discloses an electric vehicle photovoltaic charging station multi-dimensional target collaborative optimization scheduling method fused with an improved particle swarm intelligent optimization mechanism. The method is characterized by comprising the following steps: 1) constructing a multi-objective optimization model: taking electricity purchase cost minimization and storage battery cycle electric quantity minimization as objective functions, and taking a charging time constraint, an energy storage constraint, a power balance constraint and an equipment capacity constraint as constraint conditions; the method comprises the steps of (1) selecting a Pareto optimal solution set, (2) improving the design of a particle swarm algorithm, (3) outputting a result: outputting a Pareto optimal solution set after iteration is ended, and (4) selecting a decision: selecting an optimal scheme from the solution set through a fuzzy membership function or a weight preference method (for example, the optimal scheme focuses on reducing the electricity purchase cost when the optimal scheme is equal to 0.6). The method achieves the minimization of the electricity purchase cost and the minimization of the cycle electric quantity of the storage battery, prolongs the energy storage life and reduces the operation cost.

Description

Technical Field

[0001] This invention relates to the field of electric vehicle charging and renewable energy integration technology, specifically to a multi-objective optimization scheduling method for electric vehicle photovoltaic charging stations that integrates an improved particle swarm optimization (IPSO) algorithm, applicable to efficient energy management and low-carbon operation of charging infrastructure in a smart microgrid environment. Background Technology

[0002] The widespread adoption of electric vehicles (EVs) highlights their environmental potential, but traditional charging methods rely on coal-fired power, resulting in limited carbon reduction. To fully leverage the role of EVs in the energy transition, it is necessary to explore low-carbon and efficient charging solutions. Combining EVs with renewable energy sources such as wind and solar power is considered a key pathway. By constructing renewable energy bases and directly connecting them to EV charging networks, the cleanliness and low-carbon level of the charging process can be improved. With the development of smart grid technology, intelligent scheduling and coordinated optimization of renewable energy generation and EV charging will be possible in the future, improving energy efficiency and promoting energy system flexibility. Therefore, EV charging models that combine renewable energy will be an important way to promote carbon reduction in the transportation sector and achieve low-carbon goals in the future.

[0003] Currently, photovoltaic charging stations integrate photovoltaic power generation, energy storage, and charging facilities, enabling local consumption of renewable energy and representing a key focus of current demonstration projects. Their main components include photovoltaic cell arrays, energy storage battery arrays, a central control unit, DC / DC converters, AC / DC converters, and charging piles. In the main application method, the objective functions are set as minimizing electricity purchase costs and minimizing battery cycle capacity. The four constraints are charging time constraints, energy storage constraints, power balance constraints, and equipment capacity constraints. The solution algorithm typically employs the NSGA-II algorithm (Non-Dominated Sorting Genetic Algorithm).

[0004] However, the traditional NSGA-II algorithm has shortcomings in handling photovoltaic fluctuations and multi-objective conflicts. Existing solutions do not consider dimensions such as carbon emission reduction and user experience. Conventional charging station optimization scheduling methods do not consider the volatility of photovoltaic power generation and the lifespan of energy storage systems. Photovoltaic charging stations need to coordinate the relationship between photovoltaic, energy storage and charging loads. Existing research focuses on capacity configuration and lacks multi-objective optimization scheduling methods. These issues remain key technical challenges. Summary of the Invention

[0005] The purpose of this invention is to provide a multi-dimensional objective collaborative optimization scheduling method for electric vehicle photovoltaic charging stations that integrates an improved particle swarm intelligent optimization mechanism. This method minimizes electricity purchase costs and battery cycle power, extends energy storage life, and reduces operating costs.

[0006] To achieve the above objectives, the technical solution adopted by this invention is: a multi-dimensional objective collaborative optimization scheduling method for electric vehicle photovoltaic charging stations that integrates an improved particle swarm intelligent optimization mechanism, characterized by the following steps: 1) Construction of multi-objective optimization model: The objective functions are minimizing electricity purchase cost and minimizing battery cycle power, and the constraints are charging time constraint, energy storage constraint, power balance constraint and equipment capacity constraint. 2) Improved Particle Swarm Optimization Algorithm Design: Particle encoding, where each particle represents a set of decision variables. [ ] in: t sk1 Indicates the start time of charging the first electric vehicle. t skN This indicates the start time of charging for the Nth electric vehicle. t Pk This represents the arrival time of the k-th electric vehicle. t sk This indicates the start time of charging for the kth electric vehicle. t 1k This represents the departure time of the kth electric vehicle. P b,1 This represents the charging and discharging power of the energy storage system at the first moment. P b,T This represents the charging and discharging power of the energy storage system at time T. Eevk This represents the electricity demand of the k-th electric vehicle. P 0 This indicates the rated charging power of the charging station; : Energy storage charging and discharging power in the i-th time period (charging is positive, discharging is negative), range [-200kW, 200kW]). Dynamic inertia weight, In the formula, w represents the inertia weight, w max w represents the upper limit of the weight. min T represents the lower bound of the weight. max Indicates the maximum number of iterations. t Indicates the current iteration number; in, =0.9, =0.4, = 600 iterations; The results show that it quickly explores the solution space in the early stage and accurately approximates the Pareto front in the later stage, with a 30% improvement in convergence speed; Adaptive learning factor In the formula, c1 represents the dynamic factor by which a particle learns from its own historical best. max c1 represents the initial maximum value of c1. min T represents the final minimum value of c1. max This indicates the maximum number of iterations of the algorithm. t c2 represents the current iteration step, and c2 represents the dynamic factor by which the particle learns from the global historical optimum. max c2 represents the final maximum value of c2. min This represents the initial minimum value of c2; Fitness function and Pareto solution set; for dual-objective scenarios, weighted summation method is used to normalize the dual objectives. In the formula, C represents the actual electricity purchase cost under the current dispatch strategy. max E represents the theoretical maximum cost of electricity purchase, and E represents the target energy efficiency index under the current dispatch strategy. max This represents the theoretical maximum value of the energy efficiency index. Preference coefficient ( For three-object scenarios, crowding distance sorting is introduced to generate a three-dimensional Pareto solution set; Optimized process: The input parameter during data input and initialization is the predicted photovoltaic power value. Vehicle parking time , Power demand Time-of-use electricity pricing The particle swarm initialization randomly generates Np = 600 particles, with charging start times evenly distributed during vehicle parking periods, and energy storage power randomly selected between -200kW and 200kW. Iterative optimization includes objective function calculation, particle update, and optimal solution update, calculating the optimal solution for each time period based on particle encoding. The objective functions are designed by C and E, and the particle update formula is as follows: . in, For particle velocity, Let c be the particle's position (decision variable value), c1 represent the dynamic factor for the particle to learn from its own historical best, and c2 represent the dynamic factor for the particle to learn from the global historical best. id Let represent the best historical position of the i-th particle in the d-th dimension. d This represents the globally optimal position of the entire population in the d-th dimension. , Use [0,1] random numbers to avoid particles getting stuck in a fixed search pattern; Optimal update solution: compare with the individual's historical optimal solution ( ) and the global optimal solution ( ), preserving the better non-dominated solution; 3) Output Results: After the iteration terminates, output the Pareto optimal solution set; 4) Decision selection: Select the optimal solution from the solution set using fuzzy membership functions or weighted preference methods (e.g., When the value is 0.6, the focus is on reducing electricity purchase costs.

[0007] The inventive point of this invention: 1. Replace the traditional Non-Dominated Sorting Genetic Algorithm (NSGA-II) with the Improved Particle Swarm Optimization Algorithm (IPSO). Regarding dynamic inertia weight and adaptive factor, a linearly decreasing inertia weight is introduced. It achieves a smooth transition from global search to local development, with a convergence speed that is more than 30% faster than the traditional NSGA-II algorithm. Furthermore, it adaptively adjusts the learning factors c1 and c2, enhancing individual exploration capabilities in the early stages and strengthening social learning in the later stages, thus avoiding premature convergence of particles and improving the diversity of solution sets.

[0008] Regarding the multi-objective processing mechanism, the weighted summation method is used to normalize the two objectives into a single fitness function, which supports the generation of Pareto solutions with different preferences by adjusting the weight coefficients. The congestion calculation and non-dominated sorting are introduced to ensure that the Pareto solution set is evenly distributed in the objective space, avoid clustering, and cover the full range of strategies from "high cost-low loss" to "low cost-high loss".

[0009] In terms of particle encoding and constraint processing, the charging start time is directly encoded by particles. With energy storage power Continuous variables are used to adapt to the real-time scheduling needs of photovoltaic charging stations. For infeasible solutions such as charging time exceeding the limit or energy storage SOC (state of charge of energy storage system) exceeding the limit, the fitness is dynamically adjusted through the penalty function method to ensure the physical feasibility of the solution.

[0010] 2. Achieve collaborative modeling of dual objective functions Objective function 1 is to minimize electricity purchase costs. This invention employs a time-of-use pricing mechanism to quantify grid electricity purchase costs, guiding electricity purchases during off-peak hours and energy storage discharges during peak hours, thereby reducing overall costs. Objective function 2 is to minimize battery cycle capacity. This invention establishes a quantitative relationship between "scheduling strategy and equipment lifespan" by statistically analyzing the absolute value of discharge power and directly linking it to battery lifespan loss. It also supports the introduction of new objectives such as carbon emissions and user waiting time, forming a three-objective optimization model to improve the overall system efficiency.

[0011] 3. Coordinated scheduling strategy for photovoltaic charging station systems This invention dynamically coordinates photovoltaic (PV) and energy storage. Firstly, PV power is prioritized for charging: PV power is used to meet the charging load first, and any remaining energy is used for energy storage charging, reducing the phenomenon of wasted solar power (e.g., when sunlight is strong, the energy storage charging power is increased). ,in, P b,i This represents the charging and discharging power of the energy storage system at time i. P pv,i This represents the output power of the photovoltaic power generation system at time i. P ev,i This represents the power demand of the electric vehicle at time i.

[0012] Secondly, regarding the energy storage charging and discharging strategy: based on time-of-use pricing and energy storage SOC, charging or direct power supply is implemented during off-peak hours, while discharging during peak hours replaces high-cost electricity purchases, thus smoothing load fluctuations. In terms of the orderly management of large-scale charging loads, this invention generates a charging load prediction curve by statistically analyzing the normal distribution characteristics of vehicle parking time and electricity demand, thereby optimizing the charging start time. To avoid grid impact caused by centralized charging, support flexible adjustment of charging priority (such as priority charging for ambulances), and achieve differentiated scheduling through constraints or penalty terms of objective function; This invention achieves an optimal balance between economic efficiency, reliability, and sustainability of photovoltaic charging stations through a three-dimensional technological breakthrough involving algorithm innovation, model construction, and system collaboration, providing a new technological path for the efficient operation of smart charging infrastructure.

[0013] The beneficial effects of this invention are: 1. Significantly improved efficiency: The convergence speed of the improved particle swarm optimization algorithm is more than 30% faster than NSGA-II, and it can quickly approach the Pareto front within 600 iterations, reducing computation time.

[0014] 2. Outstanding multi-objective optimization results: Reduced electricity purchase cost: In scenarios with strong sunlight / SOC=0.4, it is up to 54% lower than the "instant charging" solution; Reduced cycle power: In scenarios with weak sunlight / SOC=0.8, it is up to 34% lower than the traditional solution; Energy storage life is effectively extended by 15%-20%.

[0015] 3. Enhanced photovoltaic absorption capacity: By dynamically adjusting the charging start time, the utilization rate of excess photovoltaic energy is increased to over 92%, reducing curtailment of solar power.

[0016] 4. Improved system robustness: Adapts to different light intensities and initial energy storage conditions, the generated Pareto solution set covers multiple scheduling strategies, supporting flexible responses to actual scenario requirements.

[0017] 5. This invention minimizes electricity purchase costs and battery cycle charge by coordinating photovoltaic power, energy storage charging and discharging, and charging load, thereby extending energy storage life and reducing operating costs. This invention relates to the field of electric vehicle charging and renewable energy integration technology, specifically a multi-objective optimization scheduling method for electric vehicle photovoltaic charging stations that integrates an improved particle swarm optimization (IPSO) algorithm. This method is suitable for efficient energy management and low-carbon operation of charging infrastructure in smart microgrid environments. Attached Figure Description

[0018] Figure 1 This is a flowchart of the present invention.

[0019] Figure 2 This is a flowchart of the improved particle swarm optimization (IPSO) algorithm of this invention.

[0020] Figure 3 This is a structural diagram of the photovoltaic power generation electric vehicle charging station system of the present invention.

[0021] Figure 4 This is the photovoltaic power generation prediction diagram of the present invention.

[0022] Figure 5a , Figure 5b , Figure 5c , Figure 5d The graph shows the optimization results of the two objective functions under different scenarios. Detailed Implementation

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] The following description, in conjunction with the accompanying drawings of the embodiments of the present invention, will provide a clear and complete explanation of the technical solutions of the present invention. It should be understood that the described embodiments represent only a part of the present invention, and not all of it. Based on these embodiments, all other embodiments that can be conceived by those skilled in the art without creative effort should be considered within the scope of the present invention.

[0025] The complete steps of the method of this invention are as follows: In one embodiment, the present invention provides a multi-dimensional objective collaborative optimization scheduling method for electric vehicle photovoltaic charging stations that integrates an improved particle swarm intelligent optimization mechanism.

[0026] like Figure 1As shown, a multi-dimensional objective collaborative optimization scheduling method for electric vehicle photovoltaic charging stations, integrating an improved particle swarm intelligence optimization mechanism, is illustrated in the flowchart below. Figure 1 The method includes the following specific steps: 1. Step S101: Photovoltaic power generation and charging demand forecasting.

[0027] like Figure 3 The diagram shows the structure of the photovoltaic power generation electric vehicle charging station system of this invention. The photovoltaic power generation forecast determines the available photovoltaic power, which serves as the "clean energy input" in the optimization model, prioritizing the charging load demand, with any remaining power used for energy storage charging. If the forecast value is higher than the charging load, an energy storage charging strategy is triggered; if it is lower than the load, an energy storage discharge or grid power purchase strategy is triggered. The charging demand forecast provides the charging power for each time period. As the "base load" of the optimization model, it drives the charging start time. and energy storage power The optimization process involves generating a large number of vehicle charging scenarios based on statistical patterns to simulate large-scale charging demand and ensure the universality of the optimization results.

[0028] Photovoltaic power generation and charging demand forecasting are fundamental inputs for optimized scheduling. This step first introduces the specifics of photovoltaic power generation forecasting, the objective of which is to obtain the photovoltaic output power for each period within a future timeframe. This reflects the impact of environmental factors such as solar radiation intensity and temperature on photovoltaic power generation. The data is based on historical meteorological data and real-time weather forecasts. Its output is in time-period units, generating a 24-hour photovoltaic power curve, such as... Figure 4 As shown. The physical model in the prediction model is based on the photoelectric conversion characteristics of photovoltaic cells, combined with the irradiance-power conversion formula: in: Photovoltaic output power at different times within a certain period in the future; Rated power of the photovoltaic system; : Actual irradiance (W / m²) in the i-th time period; = 1000W / m²: Standard irradiance; Temperature coefficient of photovoltaic cells (typically -0.3% / ℃); : Ambient temperature (°C) during the i-th time period. = 25 ℃: Standard temperature.

[0029] The objective of charging demand forecasting is to determine the total charging power of electric vehicles in each time period within a certain future period. This includes charging start time, charging duration, and power demand. The data is based on two aspects: firstly, statistical patterns of vehicle behavior, mainly divided into parking time distribution (the arrival and departure times of private cars at charging stations follow a normal distribution), and power demand distribution (the remaining battery capacity (SOC) of electric vehicles follows a normal distribution). Based on battery capacity and target SOC, real-time data such as vehicle parking status and charging progress at charging stations are calculated. In the prediction method based on statistical modeling, the parking time prediction specifically involves assuming the arrival time of the kth vehicle. departure time Power demand forecast includes the amount of electricity waiting to be charged. Among them, t pk Let t represent the arrival time of the kth electric vehicle. 1k This represents the departure time of the kth electric vehicle. This represents the average arrival time of electric vehicles. This represents the variance of the arrival time of electric vehicles. This represents the average time the electric vehicle leaves the vehicle. E represents the variance of the time the electric vehicle leaves. evk This represents the charging energy required for the k-th electric vehicle. For the battery's rated capacity, SOC target Indicates the target state of charge of the electric vehicle. The target SOC is set to 0.9.

[0030] Charging power calculation is set to the rated power of a single pile. Charging time Then the charging power of the k-th vehicle in time period i is: in, P k,i This represents the charging power of the k-th electric vehicle at time i. P 0 This indicates the rated charging power of the charging station. T ch,k Let t represent the charging duration of the k-th electric vehicle, where i represents the time variable and t represents the charging duration of the k-th electric vehicle. sk This indicates the start time of charging for the kth electric vehicle.

[0031] The total charging power of the entire station is (N is the total number of vehicles, such as 100 vehicles). Its output format includes the total charging power curve for each time period, reflecting the distribution characteristics of the charging load throughout the day.

[0032] Photovoltaic power generation and charging demand forecasting are crucial components of the optimization process. Photovoltaic power generation forecasting determines the available photovoltaic power, serving as the "clean energy input" to the optimization model. Priority is given to meeting charging load demand, with surplus power used for energy storage charging. If the forecast value exceeds the charging load, an energy storage charging strategy is triggered; if it falls below the load, an energy storage discharging strategy or a grid power purchase strategy is triggered. Charging demand forecasting provides charging power for different time periods. As the "base load" of the optimization model, it drives the charging start time. and energy storage power The optimization process generates a large number of vehicle charging scenarios through statistical regularities to simulate large-scale charging demand and ensure the universality of the optimization results.

[0033] 2. Step S102: Initial SOC setting for energy storage Initial State of Charge (SOC) of Energy Storage This refers to the percentage of remaining battery capacity relative to the rated capacity at the start of the optimized scheduling cycle. The calculation formula is: in, This represents the initial state of charge of the electric vehicle. Remaining electricity (kWh) This refers to the rated capacity of the battery pack.

[0034] The initial SOC range for energy storage is limited by the battery's maximum depth of discharge (DOD). The initial SOC must meet the following requirements: in, , This represents the initial state of charge (SOC) of the electric vehicle. In this invention, the maximum depth of discharge (DOD) is set to 70%, therefore the initial SOC range is... .

[0035] In this step, the typical scenario settings for the initial SOC of energy storage are divided into two types: one is low initial SOC ( =0.4), the scenario is characterized by low remaining battery power, requiring priority to utilize photovoltaic charging or purchase electricity during off-peak hours to supplement the power. Optimization strategies tend to include charging the energy storage through photovoltaics as much as possible during the day when photovoltaic power is sufficient, increasing the SOC to a safe range, and charging the energy storage or directly powering the vehicle during off-peak hours at night through grid purchase. Another approach is high initial SOC ( =0.8), the scenario is characterized by sufficient remaining battery power, which can release more power to smooth grid electricity demand. Optimization strategies tend to include reducing the amount of solar power charged to energy storage to avoid exceeding the SOC limit and causing "curtailment" of solar power, and prioritizing the use of energy storage to power vehicles during peak grid periods to reduce electricity purchase costs.

[0036] For low initial SOC scenarios, the impact on electricity purchase costs is that increased grid purchases are needed to supplement energy storage capacity or for direct power supply, potentially leading to higher purchase costs. However, purchasing electricity during off-peak hours can partially offset these costs. The impact on battery cycle capacity is that energy storage requires frequent charging and discharging to maintain capacity balance, resulting in a higher cycle capacity. The impact on photovoltaic utilization is that surplus photovoltaic power is prioritized for energy storage charging, reducing curtailment and leading to higher photovoltaic utilization. For high initial SOC scenarios, the impact on electricity purchase costs is that energy storage can provide more discharge capacity, reducing peak-hour electricity purchases and significantly lowering costs. The impact on battery cycle capacity is that reduced energy storage charging and discharging demand results in a lower cycle capacity, which helps extend battery life. The impact on photovoltaic utilization is that limited energy storage charging demand may lead to a surplus of photovoltaic power that cannot be fully absorbed.

[0037] The initial SOC is used as an input parameter in the optimization model. Before the optimization scheduling begins, the initial SOC of the energy storage is obtained in real time through the battery management system (BMS), or estimated based on historical charge and discharge data (such as the SOC at the end of the previous day's scheduling), and must meet dynamic balance constraints. in, Let SOC be the value at the end of time period i. This refers to the charging and discharging power (positive for charging, negative for discharging). SOC i-1 This represents the energy storage state of charge at time i-1. E bn Indicates the rated capacity of the energy storage system. △t i This represents the duration of the i-th time period (in hours, typically 1 hour or (0.25) hours). The initial SOC is the boundary condition ( ), ensuring that the SOC does not exceed the limit at any time period ( ).

[0038] Finally, by setting different initial SOC values, multiple Pareto solution sets are generated, covering strategies such as "high cycle power - low power purchase cost" and "low cycle power - high power purchase cost", for dispatchers to choose according to actual needs.

[0039] 3. Step S103: Calculation of the objective function Objective function calculation is the core of multi-objective optimization scheduling. Specific content includes mathematical modeling, variable definition, and calculation logic for minimizing electricity purchase costs and minimizing battery cycle capacity. The following is a detailed explanation: Objective function 1 is to minimize the cost of electricity purchase, and its mathematical expression is: Where minC represents minimizing the electricity purchase cost. : Optimize the total number of time periods (e.g., 24 hours, T=24); : Duration of the i-th time period (in hours, usually 1 hour or (0.25) hours); The electricity purchase price (yuan / kWh) for the i-th time period adopts a time-of-use pricing mechanism. Peak hours (08:00-12:00, 17:00-21:00): (0.869) yuan / kWh; Off-peak hours (00:00-08:00): (0.365) yuan / kWh; During normal hours (12:00-17:00, 21:00-00:00): (0.687) yuan / kWh.

[0040] The power supply (kW) of the distribution network in the i-th time period is determined by the system power balance relationship. in: : Total charging power of electric vehicles in the i-th time period (kW); : Battery charging and discharging power (kW, positive for charging, negative for discharging) in the i-th time period; Photovoltaic power generation (kW) in the i-th time period.

[0041] During the calculation process, the power balance judgment criteria of the calculation logic are as follows: If photovoltaic power Insufficient to meet charging load With the demand for energy storage charging and discharging (Right now If photovoltaic power is excessive, then electricity must be purchased from the grid, and the purchased power will be the difference between the purchased and purchased power; If the grid purchases power is 0, the excess photovoltaic power will be prioritized for energy storage charging. ).

[0042] In addition, time-of-use pricing aims to sum the power purchased in each time period according to the corresponding electricity price. The higher the proportion of electricity purchased during off-peak hours, the lower the total cost.

[0043] Objective function 2 is to minimize the battery cycle cost, and its mathematical expression is: Where minE represents minimizing the battery cycle cost. To optimize the total number of time periods; The duration of the i-th time period (in hours, usually 1 hour or (0.25) hours); The battery discharge power (kW) in the i-th time period is... (Only when) <0 is included in the cycle charge); This is the absolute value of the discharge power, reflecting the energy loss during a single discharge.

[0044] The calculation logic of objective function 2 includes distinguishing between charging and discharging states: during charging ( >0) is not included in the cycle charge; during discharge ( <0) Accumulate the absolute values ​​of discharge power to obtain the relationship between total cycle capacity and lifespan loss: According to the characteristics of lead-acid batteries, the number of cycles L is related to the discharge capacity E. d satisfy Where a and b are the fitting constants for the lifetime model (a = 0.003, b = 1.0), E d E represents the cumulative depth of charge and discharge of the energy storage system. bn This indicates the rated capacity of the energy storage system. The smaller the cycle charge E, the lower the battery life loss.

[0045] However, the two objective functions also have conflicting objectives. One conflict is that reducing electricity purchase costs (min C) tends to involve purchasing large amounts of electricity during off-peak hours or utilizing energy storage for peak discharge, which may increase the battery's cycle capacity (E increases). The other conflict is that reducing cycle capacity (min E) requires reducing the number of energy storage charge-discharge cycles, which may lead to increased electricity purchases during peak hours (C increases). The solution is to generate a non-dominated solution set through a multi-objective optimization algorithm, covering scheduling strategies with different preferences. Focus on cost optimization: Select a solution with high cycle capacity and low electricity purchase cost; Focus on lifespan optimization: Select a solution with low cycle power consumption and high electricity purchase cost.

[0046] Regarding weight factor adjustment, a weight factor α is introduced into the fitness function. Different priorities are generated by adjusting the value of α, as shown in the formula: In the formula, Fitness represents the fitness function value, α is the weighting factor (0 < α < 1), and C represents the actual cost.max E represents the maximum cost, and E represents the actual energy efficiency. max Indicates the maximum energy efficiency; In terms of integrating the objective function and constraints, there are three main constraints. The first constraint is the charging time constraint: the calculation of electricity purchase cost and cycle power must ensure the charging start time. During the vehicle parking period ( Otherwise, the target value is increased through a penalty function; the second constraint is the energy storage SOC constraint: the energy storage SOC in each time period must meet ( If the SOC limit is exceeded, an infeasible solution will be obtained, and readjustment is required. The third constraint is the power balance constraint: the objective function calculation must satisfy... This ensures the conservation of energy.

[0047] The following example uses a time period i: 1. Given = 150kW, = 100kW, = -50kW.

[0048] 2. Calculate the power consumption: Therefore = 0, electricity purchase cost = 0.

[0049] 3. Calculate the cycle charge: = -50kW, = 50kW, cycle power = 50×∆ .

[0050] By calculating and accumulating the values ​​in each time period, the global objective function value is obtained, providing an iterative basis for the optimization algorithm.

[0051] 4. Step S104: Generation of Pareto optimal solution set The generation of the Pareto optimal solution set is the core technology of this invention for achieving multi-objective optimization. Through algorithmic innovation and scenario segmentation, it provides a flexible and efficient set of scheduling strategies for photovoltaic charging stations, ensuring optimal overall performance under different operating conditions. The generation of the Pareto optimal solution set is a core achievement of multi-objective optimization, and its specific content involves the definition of the solution set, generation method, characteristic analysis, and practical application. The following is a detailed explanation: The Pareto optimal solution set is defined as follows: For a bi-objective optimization problem involving electricity purchase cost (C) and battery cycle capacity (E), if any two solutions A and B in a set of solutions satisfy: 1. If solution A has no worse C and E than solution B, and at least one of its objectives is strictly better, then A is said to dominate B. A and B both represent "scheduling strategy solutions".

[0052] 2. Pareto optimal solution: refers to a solution that has no other dominant solution, that is, it is impossible to improve one objective without worsening at least one objective.

[0053] 3. Pareto optimal solution set: The set of all Pareto optimal solutions constitutes the Pareto front.

[0054] The generation method and algorithm used in this invention are the Improved Particle Swarm Optimization (IPSO) algorithm. Its multi-objective processing mechanism divides particles into different levels through non-dominated sorting; particles of higher levels are closer to the Pareto front. Figure 2 The following are the specific implementation steps; crowding calculation is introduced to ensure that the solution set is evenly distributed in the target space and to avoid clustering. The key steps are: 1) Initialize the population: Generate 600 particles, each particle representing a set of charging start time and energy storage power (decision variables).

[0055] 2) Iterative optimization: Calculate the objective function value (C and E) and constraint violation status for each particle; update the particle position through dynamic inertia weight and adaptive learning factor to guide the population to converge toward the Pareto front.

[0056] 3) Solution set selection: After the iteration is completed, all non-dominated particles are retained to form a Pareto optimal solution set.

[0057] Regarding variable parameter settings: Different solar irradiance: Affecting photovoltaic power forecast values .

[0058] Initial State of Charge (SOC): Changes the initial state of energy storage charging and discharging.

[0059] The solution set diversity and robustness of the improved particle swarm optimization (IPSO) algorithm are as follows: Uniformity of distribution: Through the crowding maintenance mechanism, the solution set is uniformly distributed on the Pareto front, avoiding local clustering and ensuring the diversity of scheduling strategies.

[0060] Robustness: The solution set covers scenarios such as sudden changes in light intensity and load fluctuations. For example, when the solar intensity suddenly drops, there is an intermediate solution that balances the cost of purchasing electricity and the discharge of energy storage.

[0061] There are two main methods for selecting the Pareto optimal solution set: one is the weighted preference method, which sets target weights based on actual needs (e.g., α=0.6 emphasizing electricity purchase costs), calculates the weighted fitness of each solution, and selects the optimal solution; the other is the fuzzy decision method, which evaluates the comprehensive performance of the solutions through fuzzy membership functions and selects the "most satisfactory solution." For example: in, and These are the satisfaction functions for electricity purchase costs and circulating electricity, respectively.

[0062] In practical applications, there are three main scenarios: daily scheduling, electricity price response, and equipment maintenance.

[0063] In daily dispatch scenarios, a high cycle power scheme is selected on sunny days to make full use of photovoltaic charging; a low cycle power scheme is selected on cloudy days to reduce energy storage losses; in electricity price response scenarios, if the electricity price drops significantly during off-peak hours the next day, the off-peak electricity purchase scheme is given priority; in equipment maintenance scenarios, when energy storage needs maintenance, the scheme with the lowest cycle power is selected to reduce battery losses.

[0064] In the generation of Pareto optimal solution sets, the differences between this invention and existing technologies are mainly reflected in three aspects. First, the efficiency of solution set generation: the improved particle swarm optimization algorithm achieves a 30% faster convergence speed than the traditional NSGA-II algorithm, and can generate a denser Pareto front within the same time frame. Second, multi-dimensional scene coverage: existing technologies typically only consider a single illumination or energy storage state, while this invention generates cross-condition solution sets through multiple scene inputs, making it more adaptable. Third, dynamic update capability: it supports real-time data access and the rolling generation of new solution sets, adapting to the dynamic changes in the scheduling environment.

[0065] After implementing the solutions from steps S101 to S104, a Pareto optimal solution set will be output. The optimal solution can then be selected from the solution set using a fuzzy membership function or a weighted preference method. Details are as follows: After the iteration terminates, the Pareto optimal solution set is output. For the bi-objective optimization problem of electricity purchase cost (C) and battery cycle capacity (E), the Pareto optimal solution is the solution that has no other solution dominating it, that is, it cannot improve the other objective without worsening at least one objective. The set of all Pareto optimal solutions constitutes the Pareto Front.

[0066] The optimal solution is selected from the solution set using either fuzzy membership functions or weighted preference methods (e.g., when α=0.6, the focus is on reducing electricity purchase costs). The weighted preference method sets target weights based on actual needs (e.g., when α=0.6, the focus is on electricity purchase costs), calculates the weighted fitness of each solution, and selects the optimal solution. The fuzzy decision method evaluates the comprehensive performance of the solutions using fuzzy membership functions and selects the "most satisfactory solution."

[0067] In one embodiment, the data includes four different scenarios. The program sets the population size Np to 600, the maximum number of iterations to 600, the crossover rate to 0.9, and the mutation rate to 0.1. Pareto optimal solutions for the total cycle capacity and electricity purchase cost of the battery pack are obtained under different solar irradiance and initial SOC conditions. As shown in Figure 5, the improved particle swarm optimization algorithm is used to optimize energy storage charging and discharging under initial SOC = 0.4 and SOC = 0.8 conditions. Taking initial SOC = 0.4 as an example, the population is initialized to generate 600 charging and discharging strategies. After 600 iterations, a crossover rate of 0.9 promotes information exchange between strategies, and a mutation rate of 0.1 avoids premature convergence. After optimization, regardless of whether the solar irradiance is strong or weak, the electricity purchase cost gradually decreases as the cycle capacity increases. Figure 5a , Figure 5b , Figure 5c , Figure 5d The algorithm aligns with the trend of "relationship between electricity purchase cost and circulating power when the initial SOC is 0.4 under strong sunlight," validating its effectiveness in reducing electricity purchase costs in this scenario. When the initial SOC is 0.8, the algorithm searches for the optimal strategy based on set parameters, considering the low photovoltaic output during periods of weak sunlight, and rationally scheduling energy storage discharge to supplement load and charging to utilize low-cost electricity. After iteration, regardless of whether sunlight is strong or weak, the electricity purchase cost gradually decreases as the circulating power increases, and the change in electricity purchase cost with circulating power conforms to... Figure 5a , Figure 5b , Figure 5c , Figure 5d The statement "Relationship between electricity purchase cost and cycle power when the sunshine is weak and the initial SOC = 0.8" indicates that the algorithm can adapt to scenarios with weak sunshine and the optimization effect is stable.

[0068] The various embodiments in this specification are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. Especially for system embodiments, since they are basically similar to method embodiments, the description is relatively simple; relevant parts can be found in the description section of the method embodiments.

[0069] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are included within the scope of protection of the present invention.

Claims

1. A multi-dimensional target collaborative optimization scheduling method for a photovoltaic charging station of an electric vehicle, which is characterized by fusing an improved particle swarm intelligence optimization mechanism. Includes the following steps: 1) Construction of multi-objective optimization model: The objective functions are minimizing electricity purchase cost and minimizing battery cycle power, and the constraints are charging time constraint, energy storage constraint, power balance constraint and equipment capacity constraint. 2) Improved Particle Swarm Optimization Algorithm Design: Particle encoding, where each particle represents a set of decision variables. [ ] wherein: t sk1 denotes the charging start time of the 1st electric vehicle, t skN denotes the charging start time of the Nth electric vehicle, t Pk denotes the arrival time of the kth electric vehicle, t sk denotes the charging start time of the kth electric vehicle, t 1k denotes the departure time of the kth electric vehicle, P b,1 denotes the charge and discharge power of the energy storage system at the 1st time, P b,T denotes the charge and discharge power of the energy storage system at the Tth time, Eevk denotes the demand electric quantity of the kth electric vehicle, P 0 denotes the rated charging power of the charging pile; : the energy storage charging and discharging power in the i th period, positive for charging and negative for discharging, range: -200kW~200kW; dynamic inertia weight, where w represents an inertial weight, w max represents an upper weight limit, w min represents a lower weight limit, T max represents a maximum number of iterations, t represents a current number of iterations; wherein, = 0.9, = 0.4, = 600 iterations; the effect is to explore the solution space quickly in the early stage, and to approach the Pareto front accurately in the later stage, with a convergence speed improvement of 30%. Adaptive learning factor, wherein c1 represents a dynamic factor of the particle learning to its own history optimum, c1 max represents an initial maximum value of c1, c1 min represents a final minimum value of c1, T max represents a maximum iteration number of the algorithm, t represents a current iteration step number, c2 represents a dynamic factor of the particle learning to the global history optimum, c2 max represents a final maximum value of c2, c2 min represents an initial minimum value of c2; Fitness function and Pareto solution set; for dual-objective scenarios, weighted summation method is used to normalize the dual objectives. In the formula, C represents an actual power purchase cost under a current scheduling strategy, C max represents a theoretical maximum value of the power purchase cost, E represents a target energy efficiency index under the current scheduling strategy, E max represents a theoretical maximum value of the energy efficiency index, is a preference coefficient, And for the three-target scene, a crowding distance ranking is introduced to generate a three-dimensional Pareto solution set. Optimization process: input parameters in data input and initialization are photovoltaic power prediction values , vehicle parking time , , power demand , time-of-use electricity price ; particle swarm initialization randomly generates Np = 600 particles, the charging start time is uniformly distributed in the vehicle parking period, and the energy storage power is randomly selected from [-200kW~ 200kW]; iteration optimization includes target function calculation, particle update and optimal solution update, and each period , C, E design objective function, particle update formula is . where, is the particle velocity, is the particle position, c1 represents the dynamic factor of the particle learning from its own historical optimum, c2 represents the dynamic factor of the particle learning from the global historical optimum, id represents the ith particle, the d-dimensional position of its own historical optimum, d represents the entire population, the d-dimensional position of the global optimum, , is a [0, 1] random number to avoid the particle from falling into a fixed search mode; the optimal update solution: compare the individual historical optimal solution with the global optimal solution , keep the better non-dominated solution; 3) Output Results: After the iteration terminates, the Pareto optimal solution set is output; 4) Decision selection: Select the optimal solution from the solution set using fuzzy membership function or weight preference method.

2. The multi-dimensional objective collaborative optimization scheduling method for electric vehicle photovoltaic charging stations based on an improved particle swarm intelligent optimization mechanism as described in claim 1, characterized in that, Step 4) Select the optimal solution from the solution set: = 0.6, the focus is on reducing electricity costs. 3.The method of claim 1, wherein, The specific steps are as follows: Step S101: Photovoltaic power generation and charging demand forecasting For photovoltaic power generation prediction, the available photovoltaic power is determined as the "clean energy input" of the optimization model, which prioritizes meeting the charging load demand, and the remaining power is used for energy storage charging; if the predicted value is higher than the charging load, the energy storage charging strategy is triggered. If the load is below the threshold, the energy storage discharge or grid power purchase strategy will be triggered. The charging demand prediction provides charging power for each period , as the "base load" of the optimization model, to drive the optimization of charging start time and energy storage power ; a large number of charging scenarios of vehicles are generated by statistical rules to simulate large-scale charging demand and ensure the universality of the optimization results; The specific content of the photovoltaic power generation prediction is to obtain the photovoltaic output power of each period in a future period , reflecting the influence of environmental factors such as sunshine intensity and temperature on photovoltaic power generation, and the data basis comes from historical meteorological data and real-time meteorological forecast; the output form is in units of time period, generating a 24-hour photovoltaic power curve, and the physical model in the prediction model is based on the photoelectric conversion characteristics of the photovoltaic cell, combined with the irradiance-power conversion formula: wherein: photovoltaic output power for each time period within a future certainty period; : rated power of the photovoltaic system; : Actual irradiance (W / m2) for the ith period; = 1000 W / m2: standard irradiance; : temperature coefficient of the photovoltaic cell, typically -0.3% / °C; : Ambient temperature of the ith period (°C), = 25 °C: Standard temperature; The prediction target of the charging demand prediction is to determine the total power of electric vehicle charging in each time period in the future , including charging start time, charging duration and power demand; the data basis comes from two aspects, the first aspect is the statistical law of vehicle behavior, mainly including parking time distribution: the arrival time and the leaving time of private cars to charging stations meet normal distribution, and the power demand distribution: the remaining power SOC of electric vehicles obeys normal distribution, and the amount to be charged According to the calculation of battery capacity and target SOC, the real-time data of the real-time parking state of the vehicle accessing the charging station, the charging progress and the like; in the prediction method based on the statistical law modeling, the parking time prediction specific content is to assume that the kth vehicle arrives at time , the leaving time , and the power demand prediction includes the amount to be charged where t pk denotes the arrival time of the kth electric vehicle, t 1k denotes the departure time of the kth electric vehicle, denotes the mean of the electric vehicle arrival times, denotes the variance of the electric vehicle arrival times, denotes the mean of the electric vehicle departure times, denotes the variance of the electric vehicle departure times, E evk denotes the required charging energy of the kth electric vehicle, is the battery rated capacity, SOC target denotes the target state of charge of the electric vehicle, The target SOC is taken as 0.

9. The charging power calculation is set to a single-pile rated power , charging duration The charging power of the kth vehicle in period i is: wherein, P k,i Pki(i) denotes the charging power of the kth electric vehicle at time instant i, P 0 Pmax denotes the rated charging power of the charging station, T ch,k tki(i) denotes the charging duration of the kth electric vehicle, i denotes the time variable, t sk tki(i) denotes the charging start time of the kth electric vehicle; The total charging power of the whole station is N is the total number of vehicles, and the output forms include a total charging power curve of each period, reflecting the distribution characteristics of the charging load in a day; Photovoltaic power generation and charging demand forecasting are crucial components of the optimization process. Photovoltaic power generation forecasting determines the available photovoltaic power, serving as the "clean energy input" for the optimization model. Priority is given to meeting charging load demand, with surplus power used for energy storage charging. If the forecast value exceeds the charging load, an energy storage charging strategy is triggered; if it falls below the load, an energy storage discharging or grid power purchase strategy is triggered. Charging demand forecasting provides the charging power for different time periods. As the "base load" of the optimization model, it drives the charging start time. and energy storage power The optimization process generates a large number of vehicle charging scenarios through statistical regularities to simulate large-scale charging demand and ensure the universality of the optimization results. Step S102: Initial SOC setting for energy storage Initial State of Charge (SOC) of energy storage refers to the percentage of remaining charge of the battery pack relative to its rated capacity at the start of the optimized scheduling cycle. The calculation formula is as follows: wherein, SoC is the initial state of charge of the electric vehicle, SoR is the remaining amount of electricity (kWh), SoC is the rated capacity of the battery pack; The initial SOC range for energy storage is limited by the battery's maximum depth of discharge (DOD). The initial SOC must meet the following requirements: wherein, , is the initial state of charge of the electric vehicle; taking the maximum depth of discharge DOD = 70%, the initial SOC range is therefore ; In this step, the typical scenario setting of the initial SOC of energy storage is divided into two kinds, one is low initial SOC, = 0.4, the scene feature is that the remaining battery power is less, and it is necessary to preferentially use photovoltaic charging or purchase power during the low valley period to supplement the power; the optimization strategy tends to include that during the day when the photovoltaic is sufficient, the energy storage is charged as much as possible through photovoltaic to improve the SOC to the safe range, and at night during the low valley period, the energy storage is charged through the power grid or directly powers the vehicle; the other is high initial SOC, = 0.8, the scene feature is that the remaining battery power is sufficient, and more power can be released to suppress the power grid purchase demand; the optimization strategy tends to include reducing the charging amount of photovoltaic to the energy storage, avoiding the SOC exceeding the upper limit, and appearing "abandoned light" phenomenon, and during the peak period of the power grid, preferentially using the energy storage to discharge to power the vehicle to reduce the purchase cost; For low initial SOC scenarios, the impact on electricity purchase costs is that increased grid purchases are needed to supplement energy storage capacity or for direct power supply, potentially leading to higher purchase costs. However, purchasing electricity during off-peak hours can partially offset these costs. The impact on battery cycle capacity is that energy storage requires frequent charging and discharging to maintain capacity balance, resulting in a higher cycle capacity. The impact on photovoltaic utilization is that surplus photovoltaic power is prioritized for energy storage charging, reducing curtailment and leading to higher photovoltaic utilization. For high initial SOC scenarios, the impact on electricity purchase costs is that energy storage can provide more discharge capacity, reducing peak-hour electricity purchases and significantly lowering costs. The impact on battery cycle capacity is that reduced energy storage charging and discharging demand results in a lower cycle capacity, which helps extend battery life. The impact on photovoltaic utilization is that limited energy storage charging demand may lead to a situation where photovoltaic power is surplus but cannot be fully absorbed. The initial SOC is used as an input parameter in the optimization model. Before the optimization scheduling begins, the initial SOC of the energy storage is obtained in real time through the battery management system, or estimated based on historical charge and discharge data, such as the SOC at the end of the previous day's scheduling, and must meet dynamic balance constraints. wherein, is the SOC at the end of the i-th period, is the charge-discharge power, positive for charging, negative for discharging, SOC i-1 is the state of charge of the energy storage at the i-1-th time instant, E bn is the rated capacity of the energy storage system, △t i is the length of the i-th period, in hours, typically 1 hour or 0.25 hours; the initial SOC is a boundary condition, , ensuring that the SOC does not go out of limits at each period, ; Finally, by setting different initial SOC values, multiple Pareto solution sets are generated, covering strategies such as "high cycle power - low power purchase cost" and "low cycle power - high power purchase cost", for dispatchers to choose according to actual needs; Step S103: Calculation of the objective function Objective function calculation is the core of multi-objective optimization scheduling. Specific content includes mathematical modeling, variable definition, and calculation logic for minimizing electricity purchase costs and minimizing battery cycle capacity. The following is a detailed explanation: Objective function 1 is to minimize the cost of electricity purchase, and its mathematical expression is: Where minC represents minimizing the electricity purchase cost. : number of optimization periods, e.g. 24 hours, T=24; : the length of the i-th period, hours, usually taken as 1 hour or 0.25 hour; : the power purchase price of the i-th period distribution network (yuan / kWh), adopting time-of-use pricing mechanism: Peak hours (08:00-12:00, 17:00-21:00): (0.869) yuan / kWh; Off-peak hours (00:00-08:00): (0.365) yuan / kWh; During normal hours (12:00-17:00, 21:00-00:00): (0.687) yuan / kWh; : Power supplied by the distribution network in the ith period (kW), determined from the system power balance wherein: : Total electric vehicle charging power (kW) for the ith period; : battery charge and discharge power in the i-th period, kW, positive for charging, negative for discharging; : Photovoltaic power (kW) for the ith period; During the calculation process, the power balance judgment criteria of the calculation logic are as follows: If the photovoltaic power is insufficient to meet the charging load and the energy storage charging and discharging demand , that is , electricity needs to be purchased from the power grid, and the purchased power is the difference; if the photovoltaic power is excessive, , the power grid purchase power is 0, and the excess photovoltaic power is preferentially used for energy storage charging, ; In addition, time-of-use pricing aims to sum the power purchased in each time period according to the corresponding electricity price. The higher the proportion of electricity purchased during off-peak hours, the lower the total cost. Objective function 2 is to minimize the battery cycle cost, and its mathematical expression is: Where minE represents minimizing the battery cycle cost. To optimize the total number of time periods; The duration of the i-th time period is in hours, usually 1 hour or 0.25 hours. The battery discharge power (kW) in the i-th time period is... Only when When the value is less than 0, the circulating charge is included. This is the absolute value of the discharge power, reflecting the energy loss during a single discharge. The calculation logic of objective function 2 includes distinguishing between charging and discharging states: during charging... >0 is not included in the cycle charge; during discharge <0 The absolute values ​​of the discharge power are summed to obtain the relationship between the total cycle capacity and lifespan loss: According to the characteristics of lead-acid batteries, the number of cycles L is related to the discharge capacity E. d satisfy wherein a, b are life model fitting constants, a = 0.003, b = 1.0, E d represents the cumulative charge and discharge depth of the energy storage system, E bn represents the rated capacity of the energy storage system, the smaller the cycle electric quantity E, the lower the battery life loss; However, the two objective functions also have conflicting objectives. One conflict is that reducing electricity purchase costs (min C) tends to involve purchasing large amounts of electricity during off-peak hours or utilizing energy storage for peak discharge, which may increase the battery's cycle capacity, thus increasing E. The other conflict is that reducing cycle capacity (min E) requires reducing the number of energy storage charge-discharge cycles, which may lead to increased electricity purchases during peak hours, thus increasing C. The solution is to generate a non-dominated solution set through a multi-objective optimization algorithm, covering scheduling strategies with different preferences. Focus on cost optimization: Select a solution with high cycle capacity and low electricity purchase cost; Focus on lifespan optimization: Choose a solution with low cycle power consumption and high electricity purchase cost; Regarding weight factor adjustment, a weight factor α is introduced into the fitness function. Different priorities are generated by adjusting the value of α, as shown in the formula: In the formula, Fitness represents a fitness function value, a is a weight factor, 0 < a < 1, C represents an actual cost, C max represents a cost maximum value, E represents an actual energy efficiency, E max represents an energy efficiency maximum value; In terms of integrating the objective function and constraints, there are three main constraints; the first constraint is the charging time constraint: the calculation of electricity purchase cost and cycle power must ensure the charging start time. During the vehicle parking period, Otherwise, the target value is increased through a penalty function; the second constraint is the energy storage SOC constraint: the energy storage SOC must meet certain conditions in each time period. SOC exceeding the limit will result in an infeasible solution, requiring readjustment. The third constraint is the power balance constraint: the objective function calculation must satisfy... This ensures the conservation of energy. The following example uses a time period i:

1. Given = 150kW, = 100kW, = -50kW; 2. Calculate the power consumption: Therefore = 0, electricity purchase cost = 0; 3. Calculate the cycle charge: = -50kW, = 50kW, cycle power = 50×∆ ; By calculating and accumulating the values ​​in each time period, the global objective function value is obtained, providing an iterative basis for the optimization algorithm; Step S104: Generation of Pareto optimal solution set The generation of the Pareto optimal solution set is the core technology of this invention for achieving multi-objective optimization. Through algorithmic innovation and scenario segmentation, it provides a flexible and efficient set of scheduling strategies for photovoltaic charging stations, ensuring optimal overall performance under different operating conditions. The generation of the Pareto optimal solution set is the core achievement of multi-objective optimization, and its specific content involves the definition of the solution set, generation method, characteristic analysis, and practical application. The following is a detailed explanation: The Pareto optimal solution set is defined as follows: For a bi-objective optimization problem involving electricity purchase cost C and battery cycle capacity E, if any two solutions A and B in a set of solutions satisfy:

1. If solution A has both C and E that are no worse than solution B, and at least one of its objectives is strictly better, then A is said to dominate B; A and B both represent "scheduling strategy solutions"; 2. Pareto optimal solution: This is a solution that has no other dominant solution, meaning it cannot improve one objective without worsening at least one objective.

3. Pareto optimal solution set: The set of all Pareto optimal solutions constitutes the Pareto front; The applied generation method and algorithm is an improved particle swarm optimization algorithm. Its multi-objective processing mechanism divides particles into different levels through non-dominated sorting. Particles with higher levels are closer to the Pareto front. The specific implementation steps include: introducing crowding degree calculation to ensure that the solution set is evenly distributed in the target space and avoids aggregation. The key steps are: 1) Initialize the population: Generate 600 particles, each particle representing a set of charging start time and energy storage power; 2) Iterative optimization: Calculate the objective function value and constraint violation status for each particle, and the objective function values ​​C and E; update the particle positions through dynamic inertia weights and adaptive learning factors to guide the population to converge toward the Pareto front; 3) Solution set selection: After the iteration, all non-dominated particles are retained to form a Pareto optimal solution set; Regarding variable parameter settings: Different solar irradiance: Affecting photovoltaic power forecast values ; Initial State of Charge (SOC): This refers to the initial state of energy storage during charging and discharging. The solution set diversity and robustness of the improved particle swarm optimization algorithm are as follows: Uniformity of distribution: Through the crowding maintenance mechanism, the solution set is uniformly distributed on the Pareto front, avoiding local clustering and ensuring the diversity of scheduling strategies; Robustness: The solution set covers scenarios such as sudden changes in light intensity and load fluctuations. For example, when the solar intensity suddenly drops, there is an intermediate solution that balances the cost of purchasing electricity and the discharge of energy storage. There are two main methods for selecting the Pareto optimal solution set: one is the weighted preference method, which sets target weights based on actual needs, with α=0.6 emphasizing electricity purchase costs, calculates the weighted fitness of each solution, and selects the optimal solution; the other is the fuzzy decision method, which evaluates the comprehensive performance of the solutions through fuzzy membership functions and selects the "most satisfactory solution." For example: in, and These are the satisfaction functions for electricity purchase costs and circulating electricity, respectively; In practical applications, there are three main scenarios: daily dispatching, electricity price response, and equipment maintenance. In daily scheduling scenarios, a high cycle power scheme is selected on sunny days to make full use of photovoltaic charging; a low cycle power scheme is selected on cloudy days to reduce energy storage losses. In the electricity price response scenario, if the electricity price drops significantly during off-peak hours the next day, the off-peak electricity purchase option should be prioritized; in the equipment maintenance scenario, when the energy storage needs maintenance, the option with the lowest cycle charge should be selected to reduce battery wear. After implementing the schemes from steps S101 to S104, a Pareto optimal solution set will be output. The optimal solution can then be selected from the solution set using a fuzzy membership function or a weighted preference method; the details are as follows: After the iteration terminates, the Pareto optimal solution set is output. For the bi-objective optimization problem of electricity purchase cost C and battery cycle capacity E, the Pareto optimal solution is the solution that has no other solution that dominates it, that is, it is impossible to improve the other objective without worsening at least one objective. The set of all Pareto optimal solutions constitutes the Pareto front. The optimal solution is selected from the solution set by using fuzzy membership functions or weighted preference methods. For example, when α=0.6, the focus is on reducing electricity purchase costs. The weighted preference method sets the target weight according to actual needs. For example, when α=0.6, the focus is on electricity purchase costs. The weighted fitness of each solution is calculated and the optimal solution is selected. The fuzzy decision method evaluates the comprehensive performance of the solution through fuzzy membership functions and selects the "most satisfactory solution". The data includes four different scenarios. The program sets the population size Np to 600, the maximum number of iterations to 600, the crossover rate to 0.9, and the mutation rate to 0.

1. Pareto optimal solutions for the total cycle capacity and electricity purchase cost of the battery pack are obtained under different solar radiation intensities and initial SOC conditions. The improved particle swarm optimization algorithm is used to optimize energy storage charging and discharging under initial SOC = 0.4 and SOC = 0.8 conditions. Taking initial SOC = 0.4 as an example, 600 charging and discharging strategies are generated during population initialization. After 600 iterations, a crossover rate of 0.9 promotes information exchange between strategies, and a mutation rate of 0.1 avoids premature convergence. After optimization, regardless of whether the solar radiation is strong or weak, the electricity purchase cost gradually decreases as the cycle capacity increases. At 0.8, the algorithm searches for the optimal strategy based on the set parameters, taking into account the low output of photovoltaic power when the sunlight is weak, and reasonably scheduling energy storage discharge to supplement the load and charging to utilize low-priced electricity; after the iteration is completed, regardless of whether the sunlight is strong or weak, as the cycle electricity increases, the electricity purchase cost also gradually decreases.

Images