A method and system for optimal configuration of wind-solar-storage capacity based on multi-strategy coupled genetic algorithm

Abstract

The present application relates to the technical field of new energy power generation system planning and optimization, and discloses a wind-solar-storage capacity optimization configuration method and system based on a multi-strategy coupled genetic algorithm.In the method provided by the present application, a plurality of candidate energy storage charge-discharge strategy types are coded as discrete strategy genes of a chromosome, and wind-solar-storage capacity parameters are coded as real number capacity genes, and the two are connected in series to form a composite chromosome;hourly full-time sequence operation simulation is performed in a year cycle, the comprehensive electricity price is calculated based on the reverse calculation of the green electricity price as the fitness value, and the joint optimization of the strategy type and the capacity parameter is realized in the genetic evolution framework;after optimization, strategy sensitivity comparative analysis is performed based on the full population snapshot, and the sensitivity difference of each strategy type when the external parameter changes is automatically identified.The present application converts the energy storage strategy from the artificial preset external parameter into an endogenous variable of the genetic algorithm, and the strategy and the capacity are searched cooperatively under the same fitness function, thus overcoming the problem of missing the global optimal solution caused by the fixed strategy, and supporting optimization calculation under two pricing systems of the time-of-use electricity price and the single electricity price.

Description

Technical Field

[0001] This invention relates to the field of new energy power generation system planning and optimization technology, specifically to a method and system for optimizing the configuration of wind, solar and energy storage capacity based on a multi-strategy coupled genetic algorithm. Background Technology

[0002] With the introduction of national policies promoting direct green electricity connections and local consumption of new energy, the feasibility for continuous production-oriented electricity-consuming enterprises such as chemical plants and data centers to simultaneously obtain green electricity attributes and lower electricity costs by building their own or connecting to nearby wind, solar, and other new energy power generation facilities has significantly increased. These types of enterprises have stable load curves with minimal annual fluctuations, making them suitable for the construction of stable-output wind, solar, and energy storage systems. However, in the time-of-use pricing mechanisms currently implemented in most provinces, the division of peak and valley periods has undergone structural changes with the increasing penetration rate of new energy. Taking Shandong Province's 2026 time-of-use pricing as an example, due to the concentrated output of solar power at midday, 11:00 to 14:00 is classified as a deep valley or low valley period, while 16:00 to 22:00 is classified as a peak or high peak period, and there are significant differences in the division of periods between different months. Under this pricing structure, the choice of charging and discharging times for energy storage equipment—i.e., which times to charge and which times to discharge—directly determines the electricity cost savings and overall economic efficiency of wind, solar, and energy storage projects. Especially in scenarios with stable loads, the annual electricity cost deviation caused by differences in energy storage charging and discharging strategies can reach a significant proportion of the total project cost, and the choice of strategy type itself becomes the core decision variable of the optimization problem.

[0003] In the wind, solar, and energy storage capacity optimization configuration methods disclosed in related technologies, the energy storage charging and discharging strategy is usually treated as a preset external parameter before optimization. For example, the existing method CN202511526273 proposes a green electricity direct-connection energy storage optimization configuration method based on production simulation, but its energy storage strategy is a single fixed strategy—"charging when new energy is in surplus, discharging at other times, and peak shaving and valley filling when there is surplus capacity," and the strategy type cannot be changed. Another example is CN202511848116, which proposes an industrial park power planning method based on NSGA-II multi-objective optimization, which optimizes the source, grid, load, and storage in a coordinated manner, but its energy storage charging and discharging strategy is still a preset fixed rule, and the strategy type itself does not participate in the optimization process.

[0004] Treating energy storage strategies as preset external parameters before optimization has the following three technical drawbacks. First, the strategy type is selected manually based on experience and is not included in the optimization search space, posing a risk of missing the globally optimal strategy. Different projects exhibit significant differences in load curve shapes, temporal distribution of wind and solar resources, and electricity price structures. Strategies selected manually based on limited experience may not match the objective conditions of a specific project. Examples show that for projects with large peak-valley price differences, a strategy focused on peak-hour discharge is more advantageous, while for projects with concentrated daytime loads and low nighttime loads, a full-time discharge strategy may achieve lower overall electricity costs. The workload of manually pre-setting and calculating each strategy increases exponentially with the number of candidate strategies, making exhaustive calculation impossible in practice. Secondly, there is a non-linear coupling between strategy type and capacity configuration, and step-by-step optimization severs this coupling. The optimal strategy type and the optimal installed capacity are interdependent—a capacity scheme optimized under the same strategy may no longer be the economically optimal solution when switching to another strategy. Existing methods separate strategy selection and capacity optimization into two independent steps, severing the feedback loop between strategy and capacity. Even if a local optimum is found under the given strategy during the capacity optimization stage, the global optimality of the strategy-capacity combination in the joint solution space cannot be guaranteed. Third, existing methods lock the energy storage strategy into a single, fixed type before optimization, and their sensitivity analysis results only reflect the capacity response characteristics under that specific strategy. Planners, based on the sensitivity analysis conclusions of a single strategy, may conclude that "the project is highly sensitive to changes in external parameters" or "capacity configuration is insensitive to price fluctuations." However, different strategy types exhibit significantly different response magnitudes to changes in the same external parameter, and the sensitivity conclusions of a single strategy may deviate from reality. Existing methods cannot compare the sensitivity differences of different strategies during a single optimization process, and planners cannot ascertain whether the sensitivity conclusions of their selected strategy are representative. Therefore, how to endogenize energy storage strategy types into optimization variables during capacity optimization, so that strategies and capacity can be jointly optimized within the same framework, and how to compare the sensitivity differences of each strategy type to changes in external parameters after optimization, and avoid sensitivity misjudgments caused by improper strategy presets, has become a key issue in improving the quality of wind, solar and energy storage system optimization configuration. Summary of the Invention

[0005] This invention provides a method and system for optimizing the configuration of wind, solar and energy storage capacity based on a multi-strategy coupled genetic algorithm, in order to solve the problems in related technologies where the energy storage strategy type needs to be preset manually and the global optimal solution is missed due to the step-by-step optimization of strategy and capacity.

[0006] In a first aspect, the present invention provides a method for optimizing the allocation of wind, solar, and energy storage capacity based on a multi-strategy coupled genetic algorithm, comprising: Based on the load data and wind and solar resource data of the target project, the output characteristics of the wind and solar power generation system are determined, and an annual time-series operation scenario is constructed. A composite chromosome coding system is constructed, wherein the composite chromosome includes a strategy gene segment and a capacity gene segment; wherein the strategy gene segment is a discrete code, and its value corresponds to one of a variety of preset candidate energy storage charging and discharging strategies; the capacity gene segment is a real number code, and its value corresponds to the installed capacity of each wind, solar and energy storage device. An initial population containing multiple individuals is generated, and each individual's composite chromosome is randomly assigned a candidate energy storage strategy and a set of capacity parameters; Based on the strategy gene and capacity gene carried by each individual, a full-time simulation is performed under the annual time-series running scenario to calculate the fitness value of each individual; The population is iteratively evolved through selection, crossover, and mutation operations using a genetic algorithm. The crossover operation causes the offspring's strategy gene to be inherited from one parent, and the capacity gene to be generated through a hybrid recombination of both parents. The mutation operation switches the strategy gene among the multiple candidate energy storage strategies with a preset probability. When the preset termination conditions are met, the optimal individual's strategy type, wind-solar-storage capacity configuration scheme, and corresponding fitness index are output.

[0007] Through the above implementation method, the energy storage strategy type is transformed from a preset external parameter before optimization into an internal decision variable during the optimization process. Through the coupled evolution of strategy genes and capacity genes at the chromosome level, the genetic algorithm automatically searches for the globally optimal solution in the joint solution space of "strategy space × capacity space". Strategy and capacity are no longer determined sequentially, but rather, under the same fitness function, mismatched strategy-capacity combinations are eliminated through natural selection, while synergistic combinations are retained. This automatically discovers the strategy type and its corresponding optimal capacity configuration that match the project objective (lowest levelized cost of electricity or maximum net present value), solving the problem of optimal solution omission caused by manual strategy pre-setting and information fragmentation.

[0008] Secondly, the present invention provides a wind-solar-storage capacity optimization configuration system based on a multi-strategy coupled genetic algorithm, including a data acquisition module, a chromosome encoding module, a population initialization module, a simulation module, a genetic optimization module, and a result output module. Each module works together to execute the method described in the first aspect above to achieve joint optimization of strategy type and capacity parameters.

[0009] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention encodes multiple candidate energy storage charging and discharging strategy types into strategy genes on a composite chromosome, transforming strategy types from manually preset external parameters into endogenous variables of a genetic algorithm. During the iteration process, the genetic algorithm automatically explores and compares the fitness values ​​under each strategy type, avoiding the omission of globally optimal solutions caused by manually specified strategies, and improving the adaptability of the optimization results to the actual conditions of the project. 2. This invention encodes strategy genes and capacity genes together on the same composite chromosome. Strategy-capacity combinations are evaluated uniformly under the same fitness function, achieving collaborative search of strategy type and capacity parameters through a genetic evolution process. This overcomes the problem of fragmented coupling information caused by separating strategy selection and capacity optimization in existing methods, ensuring that the output is globally optimal in the joint solution space. 3. This invention establishes a full population snapshot record during the genetic algorithm iteration process. After optimization, based on the snapshot data, it extracts the fitness performance and optimal capacity configuration scheme of each strategy type under different external conditions, and compares the sensitivity differences of each strategy type to changes in external parameters. This sensitivity difference reveals that different strategy types respond drastically to changes in the same external condition—a phenomenon that cannot be detected in existing methods because the strategies are preset and fixed. Based on this sensitivity difference, it is possible to avoid sensitivity misjudgments caused by improper strategy presets in the early stages of project evaluation, thereby improving the reliability of investment decisions. 4. This invention integrates strategy selection and capacity optimization into a unified genetic evolution process—forming a closed loop from data input, chromosome encoding, simulation to genetic iteration and result output. This reduces the need for manual strategy setting and step-by-step trial calculations, and improves the adaptability of the optimization scheme to the actual conditions of the project. 5. This invention supports optimization calculations under both time-of-use (TOU) and single-price (SP) systems. Under TOU, the cost of electricity is calculated based on hourly rates, while under SP, it is calculated using a uniform price. The same optimization process can output the optimal strategy—capacity scheme—for both pricing methods, adapting to the project planning needs of regions with different electricity pricing policies. Attached Figure Description

[0010] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0011] Figure 1 This is a schematic diagram of the overall process of the wind-solar-storage capacity optimization configuration method based on a multi-strategy coupled genetic algorithm of the present invention. Figure 2 This is a schematic diagram of the structure of the composite chromosome coding system of the present invention; Figure 3 This is a schematic diagram illustrating the separate processing of strategy genes and capacity genes in the crossover operation of the genetic algorithm of this invention. Figure 4 This is a schematic diagram of the charging and discharging decision-making process in the full-time simulation of this invention; Figure 5 This is a structural block diagram of the wind-solar-storage capacity optimization configuration system based on a multi-strategy coupled genetic algorithm according to the present invention. Figure 6 This is a schematic diagram illustrating the strategy sensitivity comparison analysis performed by the present invention based on actual parameters of a sample project. Figure 7 This is a schematic diagram of the evolutionary trajectory of the optimal electricity price for each strategy in each generation of the genetic algorithm population of the same example project of this invention. Detailed Implementation

[0012] In one optional implementation, the annual time-series operation scenario constructed by this method is an annual time-series dataset with hourly resolution. The specific acquisition method is as follows: collect hourly solar irradiance, wind speed, and temperature meteorological data for the target project area for 8760 hours throughout the year; combine the power generation models of photovoltaic modules and wind turbine generators to calculate the theoretical photovoltaic output P_pv(t) and theoretical wind power output P_wind(t) for each hour; collect hourly electricity load data P_load(t) for the target project for 8760 hours throughout the year; construct a time-series array indexed by t = 1, 2, ..., 8760, where each time step t contains the current P_pv(t), P_wind(t), and P_load(t).

[0013] In one alternative implementation, the coding structure of the complex chromosome is as follows. The strategy gene segment uses discrete integer encoding. Six pre-defined candidate energy storage charging and discharging strategies are sequentially numbered, with the strategy gene value g_strategy ∈ {1, 2, 3, 4, 5, 6} corresponding to the following six strategies: Strategy 1: Prioritize charging for all electrical loads throughout the day, and allow discharging throughout the day; Strategy 2: Prioritize charging for all-day electricity loads, do not discharge during off-peak hours, and only discharge during peak and normal periods; Strategy 3: Prioritize charging for all-day electricity loads, and only allow discharging during peak hours; Strategy 4: Prioritize charging energy storage during off-peak hours and discharge it during peak and normal hours; Strategy 5: Prioritize charging of energy storage during off-peak hours, and only discharge during peak hours; Strategy 6: Prioritize charging of energy storage during off-peak hours, and only discharge during peak hours. The capacity gene segment uses real-number encoding and contains three or four real-number gene bits: photovoltaic installed capacity C_pv (MWp), wind power installed capacity C_wind (MW), energy storage power capacity C_bess_p (MW), and energy storage energy capacity C_bess_e (MWh). Energy storage power capacity and energy capacity together define the upper limit of the charging and discharging power and the upper limit of the energy storage system. In one alternative implementation, the total number of genes on the complex chromosome is 5, and its structure is [g_strategy, C_pv, C_wind, C_bess_p, C_bess_e].

[0014] In one alternative implementation, the initial population is generated as follows. First, the population size N is set (e.g., N = 60 to...). (200 individuals). Then, a strategy gene and a capacity gene are randomly assigned to each individual. The strategy gene is uniformly and randomly drawn from {1,2,3,4,5,6}, and the capacity gene is randomly generated within a preset search range. The search range is set according to the site conditions and engineering feasibility parameters of the proposed project. For example, the photovoltaic capacity range is [0, 200] MWp, the wind power capacity range is [0, 200] MW, the energy storage power capacity range is [0, 100] MW, and the energy storage energy capacity range is [0, 500] MWh.

[0015] In one alternative implementation, the fitness value is calculated based on a full-time simulation. The full-time simulation is performed step-by-step in a yearly period with an hourly time resolution, from t = 1 to t = 8760. In each time step, the following four operations are performed. S1041, read the wind power output P_wind(t), photovoltaic power output P_pv(t), and electricity load P_load(t) at the current time step, and calculate the total renewable energy output P_re(t) = P_pv(t) + P_wind(t), and the net load P_net(t) = P_load(t) - P_re(t). If P_net(t) > 0, it means that the renewable energy output is insufficient to meet the load, and there is a load deficit; if P_net(t) < 0, it means that the renewable energy output exceeds the load, and there is surplus electricity. S1042, determine the charging / discharging decision rules for the current time step based on the strategy genes carried by the current individual. The charging / discharging rules for each candidate strategy are implemented using conditional judgment logic: if the strategy requires the current time period to be a charging priority period, then the charging power is min(remaining rechargeable energy storage capacity + renewable energy output exceeding the load); if the strategy requires the current time period to be able to discharge, then the discharging power is min(remaining energy storage capacity + deficit power) when the renewable energy output is insufficient to cover the load; if the strategy prohibits discharging in the current time period, then even if there is a load deficit, no discharging will occur, and the grid will purchase electricity to supplement it. S1043, Update the State of Charge (SOC)(t) of the energy storage device. Based on the charging or discharging power at the current time step, update SOC(t) = SOC(t-1) + (η_ch × P_ch(t) - P_dis(t) / η_dis) × Δt / C_bess_e. Where η_ch is the charging efficiency and η_dis is the discharging efficiency. Δt is the time step (1 hour), and C_bess_e is the energy storage capacity. S1044 calculates the grid-supplied power, renewable energy curtailment power, and renewable energy supply power at the current time step. When renewable energy output exceeds the energy storage charging demand and there is still a surplus, the excess is recorded as curtailment power P_curt(t). When renewable energy output plus energy storage discharge power is still insufficient to meet the load, the shortfall is supplemented by grid power purchases, recorded as grid-supplied power P_grid(t). The direct renewable energy supply power is the equivalent redistribution result of P_re_to_load(t) = P_load(t) - P_grid(t) - P_dis(t) + P_curt(t). After traversing all 8760 time steps, the total annual renewable energy power generation, total abandoned power, total grid power supply, and time-of-use grid power supply for each electricity price period are summarized.

[0016] The fitness value is calculated based on the results of a full-time simulation. In one optional implementation, the fitness value is the comprehensive electricity price ¥_comprehensive (yuan / kWh). The calculation method is as follows: First, calculate the project's total lifecycle cash flow, including the initial investment, operation and maintenance costs, and annual electricity purchase expenditures for wind, solar, and energy storage equipment. Based on a preset target rate of return, calculate the green electricity price ¥_green (yuan / kWh) under the condition that the project's total lifecycle net present value (NPV) equals zero. Then, sum the green electricity cost (green electricity generation × ¥_green) and the grid supply cost (grid supply in each time period × sum of electricity prices in each time period) to obtain the total electricity cost. Divide this by the total annual electricity consumption to obtain the comprehensive electricity price. In another optional implementation, the fitness value is the total lifecycle net present value (NPV).

[0017] After the fitness value is calculated, the genetic algorithm performs iterative evolution on the population. S1051, Selection Operation. A tournament selection method is used, randomly selecting k individuals from the population (k is the tournament size, e.g., k = 3 to 5), and choosing the individual with the best fitness value to enter the next generation. This operation increases the probability of individuals with higher fitness being retained, while maintaining a certain degree of randomness to avoid premature convergence. S1052, Crossover Operation. For the selected parent pairs, crossover is performed according to the crossover probability P_c. In an optional implementation, the strategy gene crossover method is random inheritance: the offspring's strategy gene is randomly inherited from one parent's strategy gene, without mixing. The capacity gene crossover method is mixed recombination crossover: the capacity gene position of the offspring is taken as the weighted average of the corresponding gene positions of both parents, i.e., C_child = α × C_parent1 + (1 - α) × C_parent2, where α is a random number in the interval [0, 1]. The reason for using different crossover mechanisms for strategy genes and capacity genes is that strategy genes are discrete categorical variables, and weighted mixing is meaningless; capacity genes are real variables, and mixed recombination can effectively generate new capacity combinations. S1053, Mutation Operation. Mutation is performed on each gene of an individual according to the mutation probability P_m. In an optional implementation, the mutation method for the strategy gene is strategy switching: randomly switching from the other five candidate strategies with probability P_m. The mutation method for the capacity gene is Gaussian perturbation: a Gaussian random perturbation with a mean of zero and a standard deviation of a preset value (e.g., 5% to 10% of the search range) is added to the current value with probability P_m, i.e., C_new = C_current + N(0, σ^2). The mutation operation introduces random perturbation, which helps the population escape local optima. S1054, Elite Preservation. The top few individuals with the best fitness values ​​in each generation (e.g., the top 2 to 5 individuals) are directly preserved to the next generation without crossover or mutation, preventing the loss of the optimal solution during evolution.

[0018] The genetic algorithm stops iterating when the preset termination condition is met. The termination condition can be set as follows: reaching the maximum number of generations G_max (e.g., 100 to 300 generations), or the improvement of the optimal fitness value of the population in M ​​consecutive generations is less than the preset threshold ε. After optimization is terminated, the following results are output: the optimal strategy number and its corresponding strategy description, the optimal capacity configuration (PV C_pv*, wind power C_wind*, energy storage power C_bess_p*, energy storage energy C_bess_e*), the optimal fitness value (comprehensive electricity price ¥_comprehensive*), and the hourly results details of the full-time simulation (8760 rows × each electrical quantity column).

[0019] In one alternative implementation, in each generation of the genetic algorithm, the strategy type, capacity parameter, fitness value, and current external condition parameters (such as electricity price name, load level, etc.) of all individuals are recorded as a snapshot of the entire population. The snapshot data is continuously accumulated during the optimization process and can be used to trace the evolutionary trajectory of each generation and conduct subsequent comparative analysis of strategy sensitivity after optimization is completed.

[0020] Through steps S101 to S106 above, this invention achieves joint optimization of strategy type and capacity parameter. The strategy is no longer a manually preset external fixed parameter before optimization begins, but rather an endogenous variable of the genetic algorithm that co-evolves with the capacity parameter under the same fitness function. In each generation, individuals with different strategy types compete fairly under a unified simulation and fitness evaluation system. Strategy-capacity combinations with poor fitness are naturally eliminated, while combinations with good fitness are retained and continue to evolve. The optimal individual output at optimization termination simultaneously provides the optimal strategy and optimal capacity, which have been co-matched through the evolutionary process.

[0021] In one alternative implementation, a full lifecycle simulation, taking into account equipment degradation, is further performed for each year of the project's operation. S201 involves re-executing the full-time simulation for each year of the operation period after adjusting the installed capacity based on the attenuation coefficient of each device. In one alternative implementation, the photovoltaic power generation equipment adopts a linear degradation model, which decreases linearly from the rated power in the first year of operation to the end of the operation period; the wind power generation equipment is regarded as having no power degradation during the operation period; the energy storage equipment adopts a stepped degradation model, and continues to degrade after being reset to the rated capacity according to the replacement strategy in a preset year in the middle of operation. In another optional implementation, the operating period is 25 years. The photovoltaic equipment has a degradation coefficient of 1.000 in the first year, which decreases uniformly year by year thereafter, reaching a degradation coefficient of 0.903 in the 25th year. The energy storage equipment degrades in a step-like manner with increasing annual degradation for the first 10 years, and is replaced with new equipment in the 11th year to restore the degradation coefficient to 1.000, after which it continues to degrade in a step-like manner until the 25th year. For each year during the operation period, the installed capacity of photovoltaic power, wind power, and energy storage is adjusted according to the degradation coefficient of that year. The full-time operation simulation of S104 is then re-executed to obtain the renewable energy generation, abandoned power, and grid power supply for that year. Based on the operating data of each year, the electricity purchase cost for each year is calculated, and the total life cycle cash flow is obtained. Then, the total life cycle net present value or green electricity price is calculated. S202, based on the snapshot data of the entire population across generations recorded in S106, performs a comparative analysis of policy sensitivity. Specifically, it includes the following steps: S2021, Candidate Configuration Pool Construction. From the full population snapshots of each generation of the genetic algorithm, all unique policy type and capacity configuration combinations are extracted to construct a candidate configuration pool. Each candidate configuration contains policy gene values ​​and capacity gene values, and retains the full-time simulation results completed during the GA run. The candidate configurations cover all policy-capacity combinations explored during the genetic evolution process, avoiding the computational overhead of blindly enumerating the entire space. S2022, External Parameter Scanning and Rapid Evaluation. External parameters are taken in preset step sizes within a predefined range. For each parameter value, the simulation results cached in the snapshot are used to recalculate the fitness value of each candidate configuration under that parameter condition, without re-executing the full-time simulation. The evaluation method involves extracting the hourly grid supply power of each candidate configuration, weighting it with the new electricity price parameter to obtain the electricity purchase cost of each configuration under that price, and then calculating the comprehensive cost per kilowatt-hour or net present value to determine the optimal configuration scheme under that parameter value. S2023, Strategy Sensitivity Difference Analysis. Along the direction of continuous change in external parameters, the optimal capacity configuration and fitness values ​​of each strategy type at each parameter level are compared. The magnitude of change in optimal capacity configuration and fitness value for each strategy type within the parameter variation range is statistically analyzed to obtain the sensitivity differences of each strategy type to changes in external parameters. This sensitivity difference reveals that different strategy types make drastically different capacity adjustment responses to changes in the same external conditions, a phenomenon that cannot be detected in existing methods because the strategies are preset and fixed. S2024, Adaptive Supplementation of the Candidate Pool. During the above scanning process, if the capacity parameter corresponding to the optimal configuration under a certain parameter value has reached the capacity search boundary of that strategy type in the candidate configuration pool, a local genetic algorithm supplementary search is triggered: using the current parameter value as the electricity price condition and the strategy type corresponding to the optimal configuration as the fixed strategy gene, a supplementary population is initialized in the capacity gene search space for local optimization, and the newly discovered configuration schemes and simulation results are added to the candidate configuration pool. The above supplementary search can avoid the problem of missing the true optimal solution when the parameter deviates significantly from the baseline due to insufficient coverage of the candidate configuration pool under the baseline parameter. Figure 6 illustrates the strategy sensitivity comparison analysis: the horizontal axis represents external parameters (such as energy storage unit price), and the vertical axis represents the optimal energy storage capacity and the comprehensive electricity price, respectively showing the capacity adjustment trajectory and electricity price change trajectory of each strategy type within the parameter variation range. Through the above implementation method, after a single main optimization, by reusing the simulation cache, the sensitivity differences of each strategy type to changes in external parameters are obtained. This reveals that the fixed strategy method not only leads to deviations in optimization results but also to misjudgments of the project's economic sensitivity, eliminating the need to re-execute a complete genetic optimization for each parameter level.

[0022] In this embodiment, parallel computing of a multi-price electricity system is further introduced based on the above implementation method. In step S301, during the fitness value calculation phase of S104, fitness values ​​are calculated for the same set of full-time simulation results for the same individual, under both time-of-use (TOU) and single-price systems. Under the TOU system, the electricity purchase cost is calculated based on hourly grid supply power weighted according to the TOU policy; under the single-price system, the electricity price for each time period is uniformly set to the same value for calculation. The two pricing systems evaluate the economics of their respective strategy-capacity combinations independently. After optimization, the optimal strategy-capacity scheme for each pricing system is output simultaneously, providing a quantitative basis for comparing projects under different electricity price policy scenarios. Through the above implementation methods, this method is compatible with different pricing systems such as time-of-use pricing and single-price pricing, thus improving the applicability of the method.

[0023] This embodiment also provides a wind-solar-storage capacity optimization configuration system based on a multi-strategy coupled genetic algorithm. This system is used to implement the above embodiments and preferred embodiments, and details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the system described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0024] This embodiment provides a wind-solar-storage capacity optimization and allocation system based on a multi-strategy coupled genetic algorithm, as shown in Figure 5, including: The data acquisition module 510 is used to determine the output characteristics of the wind and solar power generation system and construct an annual time-series operation scenario based on the load data and wind and solar resource data of the target project. Chromosome coding module 520 is used to construct a composite chromosome coding system. The composite chromosome includes a strategy gene segment and a capacity gene segment. The strategy gene segment is a discrete code, and its value corresponds to one of a variety of preset candidate energy storage charging and discharging strategies. The capacity gene segment is a real number code, and its value corresponds to the installed capacity of each wind, solar and energy storage device. Population initialization module 530 is used to generate an initial population containing multiple individuals, wherein each individual's composite chromosome is randomly assigned a candidate energy storage strategy and a set of capacity parameters. The simulation module 540 is used to perform a full-time simulation in the said annual time-series running scenario based on the strategy genes and capacity genes carried by each individual, and to calculate the fitness value of each individual; The genetic optimization module 550 is used to iteratively evolve the population through selection, crossover and mutation operations of a genetic algorithm; wherein, the crossover operation causes the strategy gene of the offspring to be inherited from one of the parents and the capacity gene to be generated through the mixed recombination of the two parents, and the mutation operation switches the strategy gene among the multiple candidate energy storage strategies with a preset probability. The result output module 560 is used to output the strategy type, wind-solar-storage capacity configuration scheme and corresponding fitness index of the optimal individual when the preset termination conditions are met.

[0025] In an optional implementation, the above modules can be further subdivided into sub-units. The data acquisition module 510 includes a load data import unit, a wind and solar resource data import unit, and an electricity price parameter configuration unit. The operation simulation module 540 includes a charge / discharge decision unit, an energy storage status update unit, a grid power calculation unit, and a fitness calculation unit; wherein the charge / discharge decision unit determines the charging and discharging time periods and priorities based on the strategy genes carried by the current individual, and the fitness calculation unit supports both the comprehensive cost-per-kilowatt-hour (LPQE) model and the net present value (NPV) model. The genetic optimization module 550 includes a selection operation unit, a crossover operation unit, and a mutation operation unit; wherein the crossover operation unit performs inheritance-based crossover on the strategy genes and recombination-based crossover on the capacity genes, and the mutation operation unit performs candidate strategy switching on the strategy genes and Gaussian perturbation mutation on the capacity genes. The result output module 560 includes a full population snapshot storage unit, used to record all individual information of each generation of the genetic algorithm population to support strategy sensitivity comparative analysis.

[0026] This invention also provides an electronic device, including a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the wind-solar-storage capacity optimization configuration method based on a multi-strategy coupled genetic algorithm as described in any one of claims 1 to 7.

[0027] This invention also provides a computer-readable storage medium storing computer instructions, which are used to cause a computer to execute the wind-solar-storage capacity optimization configuration method based on a multi-strategy coupled genetic algorithm as described in any one of claims 1 to 7.

[0028] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Application Example

Claims

1. A method for optimizing the allocation of wind, solar, and energy storage capacity based on a multi-strategy coupled genetic algorithm, characterized in that, The method includes: Based on the load data and wind and solar resource data of the target project, the output characteristics of the wind and solar power generation system are determined, and an annual time-series operation scenario is constructed. A composite chromosome coding system is constructed, wherein the composite chromosome includes a strategy gene segment and a capacity gene segment; the strategy gene segment is a discrete code, and its value corresponds to one of a variety of preset candidate energy storage charging and discharging strategies; the capacity gene segment is a real number code, and its value corresponds to the installed capacity of each wind, solar and energy storage device. An initial population containing multiple individuals is generated, and each individual's composite chromosome is randomly assigned a candidate energy storage strategy and a set of capacity parameters; Based on the strategy gene and capacity gene carried by each individual, a full-time simulation is performed under the annual time-series running scenario to calculate the fitness value of each individual; The population is iteratively evolved through selection, crossover, and mutation operations using a genetic algorithm. In the crossover operation, the strategy gene of the offspring is inherited from one of the parents, and the capacity gene is generated by mixing and recombination of the real-number codes of both parents. In the mutation operation, the strategy gene switches between the multiple candidate energy storage strategies with a preset probability, and the capacity gene is perturbed at the current position with a preset probability. When the preset termination conditions are met, the optimal individual's strategy type, wind-solar-storage capacity configuration scheme, and fitness index are output.

2. The method according to claim 1, characterized in that, The various candidate energy storage charging and discharging strategies are organized according to two dimensions: charging priority and discharging window, and include at least: A strategy that prioritizes all-day electricity load and allows for discharge at all times; A strategy of prioritizing electricity load throughout the day and not discharging during off-peak hours; A strategy that prioritizes electricity demand throughout the day and allows discharge only during peak hours; The strategy is to prioritize charging of energy storage during off-peak hours and allow it to be discharged during peak and normal hours. The strategy is to prioritize charging of energy storage during off-peak hours and allow it to be discharged only during peak hours; The strategy is to prioritize charging of energy storage during off-peak hours and allow it to discharge only during peak hours.

3. The method according to claim 1, characterized in that, In the full-time simulation, it iterates step-by-step with a yearly period and an hourly time resolution. In each time step: Obtain the wind power output, photovoltaic power output, and electricity load at the current time step, and calculate the total output and net load of new energy sources; The charging and discharging decision rules are determined based on the strategy genes carried by the individuals, including the selection of charging priority periods and the determination of allowable discharging periods; Update the state of charge of energy storage devices; Calculate the grid-supplied power, renewable energy curtailment power, and renewable energy power supply power at the current time step.

4. The method according to claim 1, characterized in that, The fitness value is the comprehensive electricity price, which is calculated as follows: based on the preset target rate of return, the green electricity price is calculated backward with the net present value of the project's entire life cycle being zero. The green electricity fee and the grid power supply fee are added together and then divided by the total electricity consumption. The grid power supply fee is calculated according to the time-of-use pricing policy based on the hourly grid power supply.

5. The method according to claim 1, characterized in that, During the iterative evolution process, the strategy type, capacity parameter, and fitness value of all individuals in each generation of the population are recorded as a snapshot of the entire population. After optimization, a strategy sensitivity comparison analysis is performed based on the full population snapshot, including: Extract unique policy type and capacity configuration combinations from snapshots of each generation of the population to construct a candidate configuration pool; Within a preset range of external parameter variations, values ​​are taken at preset step sizes, and the fitness values ​​of each candidate configuration under the current parameters are recalculated using the simulation results cached in the snapshot. By comparing the magnitude of changes in optimal capacity configuration and fitness values ​​for different strategy types at various parameter levels along the direction of external parameter changes, we can obtain the differences in sensitivity of each strategy type to changes in external parameters.

6. The method according to claim 5, characterized in that, The sensitivity differences include the magnitude of change in optimal capacity configuration and fitness value for each strategy type when external parameters change; wherein, the change in fitness value reflects the sensitivity of the strategy type to parameter changes, and the change in capacity configuration reflects the differences in capacity adjustment strategies of different strategy types under the same parameter perturbation.

7. The method according to claim 1, characterized in that, Following the full-time simulation, a full lifecycle simulation, taking into account equipment degradation, is further performed for each year of the project's operation: Photovoltaic power generation equipment decreases year by year according to a linear decay model; Wind power generation equipment is considered to have no power degradation during the operation period; The energy storage equipment operates according to a stepped degradation model and continues to degrade after being reset to its rated capacity in a preset year during the middle of its operation. After adjusting the installed capacity according to the attenuation coefficient for each year, the full-time simulation is re-executed, and the results of each year are summarized to calculate the life-cycle fitness value.

8. A wind-solar-storage capacity optimization configuration system based on a multi-strategy coupled genetic algorithm, characterized in that, include: The data acquisition module is used to determine the output characteristics of the wind and solar power generation system and construct an annual time-series operation scenario based on the load data and wind and solar resource data of the target project. The chromosome coding module is used to construct a composite chromosome coding system. The composite chromosome includes a strategy gene segment and a capacity gene segment. The strategy gene segment is a discrete code, and its value corresponds to one of a variety of preset candidate energy storage charging and discharging strategies. The capacity gene segment is a real number code, and its value corresponds to the installed capacity of each wind, solar and energy storage device. The population initialization module is used to generate an initial population containing multiple individuals, wherein each individual's composite chromosome is randomly assigned a candidate energy storage strategy and a set of capacity parameters. The simulation module is used to perform a full-time simulation in the annual time-series running scenario based on the strategy genes and capacity genes carried by each individual, and to calculate the fitness value of each individual. The genetic optimization module is used to iteratively evolve the population through selection, crossover, and mutation operations of a genetic algorithm. In the crossover operation, the strategy gene of the offspring individual is inherited from one of the parents, and the capacity gene is generated by mixing and recombination of the real-number codes of the two parents. In the mutation operation, the strategy gene switches between the multiple candidate energy storage strategies with a preset probability. The results output module is used to output the strategy type, wind-solar-storage capacity configuration scheme and fitness index corresponding to the optimal individual when the preset termination conditions are met.

9. An electronic device, characterized in that, The system includes a memory and a processor, which are interconnected. The memory stores computer instructions, and the processor executes the computer instructions to perform the wind-solar-storage capacity optimization configuration method based on a multi-strategy coupled genetic algorithm as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to execute the wind-solar-storage capacity optimization configuration method based on a multi-strategy coupled genetic algorithm as described in any one of claims 1 to 7.

Images