Energy storage system optimization method and system based on multi-objective ant colony optimization algorithm
By constructing a heterogeneous heuristic information matrix for grouping intelligent agents using a multi-objective ant colony optimization algorithm, and realizing multi-strategy parallel search, the problem of imbalance between global exploration and local development capabilities in source-grid-load-storage optimization of metaheuristic algorithms is solved, generating a high-quality Pareto solution set and improving the economy and environmental friendliness of energy storage systems.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HUANENG HUBEI ENERGY SALES LLC
- Filing Date
- 2026-03-24
- Publication Date
- 2026-06-26
AI Technical Summary
Existing metaheuristic algorithms struggle to achieve a dynamic balance between global exploration and local exploitation in multi-objective optimization problems involving source, network, load, and storage. This leads to the algorithms getting trapped in local optima and failing to obtain high-quality Pareto optimal solution sets, thus limiting their application in practical engineering.
A multi-objective ant colony optimization algorithm is adopted. By grouping different agents into heterogeneous heuristic information matrices with differentiated optimization biases, multi-strategy parallel search is achieved. Initial charging and discharging strategies are generated by combining operational inertia, and the search results are integrated through evaluation and optimization mechanisms to generate a high-quality Pareto solution set.
It effectively expands the coverage of the solution space, avoids the algorithm getting stuck in local optima, generates higher quality Pareto solutions, provides a better coordination scheme in terms of economy and environmental protection, and improves the overall operational efficiency of the source-grid-load-storage system.
Smart Images
Figure CN122292452A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of grid regulation technology for energy storage systems, specifically relating to an optimization method and system for energy storage systems based on a multi-objective ant colony optimization algorithm. Background Technology
[0002] Renewable energy generation is characterized by significant volatility and intermittency, and its large-scale integration poses a severe challenge to the stable operation and power quality of the power system. To address this issue, the integrated generation-grid-load-storage architecture has emerged. This architecture treats distributed generation, the public grid, flexible and adjustable loads, and energy storage systems as an organic whole. Through intelligent coordination and optimized scheduling, it aims to smooth out fluctuations in renewable energy output, reduce system operating costs, and decrease carbon emissions, ultimately achieving safe, economical, and low-carbon multi-objective coordinated operation of microgrids. Therefore, source-grid-load-storage coordination and optimization technology has become a core means to address the uncertainties of high-proportion renewable energy power systems.
[0003] This technology typically relies on power prediction systems to obtain information on future renewable energy output and load demand, and uses advanced optimization algorithms to calculate the optimal scheduling strategy for each unit. These strategies are executed through hardware and software platforms such as energy management systems (EMS) and virtual power plants (VPPs), thereby achieving multiple functions such as peak shaving and valley filling, and frequency and voltage support. Currently, optimization methods in this field are mainly divided into three categories: traditional mathematical programming methods, emerging artificial intelligence methods, and mainstream heuristic methods represented by metaheuristic algorithms.
[0004] Traditional mathematical programming methods (such as linear programming and mixed-integer linear programming) construct models by making numerous simplifications and assumptions about the real-world system. While this ensures a certain level of solution speed, oversimplification can lead to optimization results that deviate from reality, making it difficult to accurately characterize the complex constraints and nonlinear characteristics of the system. Emerging artificial intelligence methods (especially deep reinforcement learning) have shown potential in handling high-dimensional and nonlinear problems, but their decision-making processes often lack interpretability, exhibiting a "black box" problem. Furthermore, they rely on large amounts of data for training, facing challenges such as high training costs and poor policy transferability.
[0005] Currently, in practical engineering applications and academic research, metaheuristic algorithms, represented by Multi-Objective Genetic Algorithm (MOGA), Particle Swarm Optimization (PSO), and Ant Colony Optimization (ACO), are the mainstream methods for solving multi-objective optimization problems involving source-grid-load-storage systems. These methods simulate natural phenomena or swarm intelligence to search in a complex solution space to find a series of non-dominated solutions (Pareto optimal solution sets), providing decision-makers with multiple trade-off options.
[0006] However, existing standard metaheuristic algorithms still have significant shortcomings when applied to high-dimensional, nonlinear, and strongly constrained multi-objective optimization problems such as source-grid-load-storage optimization. The core problem lies in the fact that these algorithms typically employ a single, fixed search strategy to guide the entire population. For example, standard genetic algorithms use fixed crossover and mutation probabilities, and standard particle swarm optimization algorithms use fixed inertia weights and social cognition coefficients. When solving problems like source-grid-load-storage optimization with complex Pareto fronts, this single strategy is ill-suited to the needs of different stages of the search process. In the early stages of the search, strong global exploration capabilities are needed to discover potential optimal solution regions; while in the later stages, strong local exploitation capabilities are required to accurately converge to the Pareto front. Fixed strategies cannot achieve this dynamic balance, leading to an imbalance between the algorithm's global exploration and local exploitation capabilities, making it prone to getting trapped in local optima and failing to obtain a uniformly distributed and sufficiently diverse Pareto optimal solution set. Ultimately, the quality and distribution of the obtained solution set are often insufficient, failing to provide operators with high-quality trade-offs covering multiple dimensions such as safety, economy, and low carbon emissions, thus limiting its application effectiveness in practical engineering.
[0007] Therefore, there is an urgent need for a new optimization method that can overcome the above limitations, coordinate the exploration and development in the search process in a dynamic and adaptive manner, effectively improve the quality, diversity and convergence of the Pareto solution set obtained by source-grid-load-storage co-optimization, and thus provide better decision support for the efficient, reliable and low-carbon operation of microgrids. Summary of the Invention
[0008] The purpose of this invention is to overcome the above-mentioned shortcomings and provide an energy storage system optimization method and system based on a multi-objective ant colony optimization algorithm.
[0009] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides an energy storage system optimization method based on a multi-objective ant colony optimization algorithm, comprising the following steps: Acquire grid data for energy storage systems; Based on the acquired grid data of the energy storage system, several heterogeneous heuristic information matrices with different optimization biases are constructed. Based on the constructed heterogeneous heuristic information matrix, a multi-strategy parallel target ant colony optimization algorithm is used for cooperative search to obtain an initial charging and discharging strategy. The initial charging and discharging strategy is then evaluated to obtain the evaluation results. Based on the evaluation results, the optimal solution is determined, and the energy storage charging and discharging power sequence corresponding to the optimal solution is output as the final coordinated optimization scheduling scheme and sent to the energy management system for execution.
[0010] In the step of acquiring grid data for the energy storage system, the grid data includes: grid time-of-use electricity price, grid carbon emission intensity, microgrid load demand, and renewable energy generation status.
[0011] The method for constructing several heterogeneous heuristic information matrices with different optimization biases based on the acquired grid data of the energy storage system is as follows: Based on the acquired grid data of the energy storage system, heuristic strategy combinations are defined according to different target bias heuristic test rates. Multidimensional guiding signals are generated based on the combination of defined heuristic strategies; Based on the multidimensional guiding signals generated by heuristic strategy combinations, a heuristic information matrix is generated for each strategy.
[0012] In the step of defining heuristic strategy combinations based on the acquired grid data of the energy storage system and according to different target bias heuristic test rates, the defined heuristic strategy combinations are as follows: Strategy 1: Environmental protection first, with a weighting combination of {cost weight: 0.35, carbon emission weight: 0.65}; Strategy 2: Balancing economic and environmental protection, with a weighted combination of {cost weight: 0.40, carbon emission weight: 0.60}; Strategy 3: Economic priority, with a weighting combination of {cost weight: 0.45, carbon emission weight: 0.55}.
[0013] The method for generating multidimensional guiding signals based on the definition-based heuristic strategy combination is as follows: For each time step t, a normalized multidimensional guiding signal is calculated based on each heuristic strategy, using the following formula:
[0014] in, Representative strategy The guiding signal at time t and These are the periodic averages of electricity price and carbon intensity, respectively. and They represent strategies respectively. Cost weights and carbon emission weights, and This represents the electricity price and carbon emission intensity at time t.
[0015] The method for generating a heuristic information matrix for each strategy in the multidimensional guiding signal based on heuristic strategy combination is as follows: Based on the guidance signal, the continuous operation space of the energy storage system at each time t is determined. Discretized into 11 action levels; For each policy k and each time t, the degree of expectation is quantified into a heuristic information matrix. ,in Indicates a discharge action; When a < 0, the desired level is proportional to the guiding signal, encouraging discharge: at this time,
[0016] When a > 0, the expected level is inversely proportional to the guiding signal to encourage charging. At this time:
[0017] in To prevent extremely small positive numbers with a denominator of zero; When a=0, It is 0 or the baseline value.
[0018] Based on the constructed heterogeneous heuristic information matrix, a multi-strategy parallel target ant colony optimization algorithm is used for cooperative search to obtain an initial charging and discharging strategy. The method for evaluating the initial charging and discharging strategy and obtaining the evaluation results is as follows: Initializing the ant colony algorithm includes: creating an agent population with multiple agents; and initializing a globally shared two-dimensional pheromone matrix. Initialize a Pareto front archive for storing nondominated solutions; Group the agents and assign heuristic information matrices, wherein multiple agents are divided into at least three groups, and an environmental priority heuristic matrix is assigned to the first group, an economic-environmental balance heuristic matrix is assigned to the second group, and an economic priority heuristic matrix is assigned to the third group. Parallel paths are constructed for the grouped agents. Each agent determines its set of possible actions based on its current state and selects the next action based on the state transition probability. After the agent completes the action selection at each moment, the initial charging and discharging strategy is generated by the inertia of operation. The generated initial charge and discharge strategy is evaluated to check whether its final state of charge meets the preset constraints, and the evaluation results are obtained.
[0019] In the step of constructing parallel paths for the grouped agents, where each agent determines its set of available actions based on its current state and selects its next action based on the state transition probability, the state transition probability is jointly determined by the heuristic information matrix of its group and the globally shared pheromone matrix.
[0020] in, Let t be the probability of taking a charging or discharging action at time t; The pheromone matrix represents the historical pheromone concentration at position (t,a), reflecting the successful experience of ants choosing this position in the past. The pheromone importance factor; Instant benefits on the platform; The importance factor for immediate returns.
[0021] The generated initial charging and discharging strategy is evaluated to check whether its final state of charge meets the preset constraints. After obtaining the evaluation results, the Pareto front archive is updated. All valid solutions generated in this iteration are compared with the solutions in the Pareto front archive. The Pareto front archive is updated using a non-dominated sorting method to retain all non-dominated solutions.
[0022] Secondly, the present invention provides an energy storage system optimization system based on a multi-objective ant colony optimization algorithm, comprising: The acquisition module is used to acquire grid data from the energy storage system. The heuristic information matrix construction module is used to construct several heterogeneous heuristic information matrices with different optimization biases based on the acquired grid data of the energy storage system. The result generation module is used to perform cooperative search based on the constructed heterogeneous heuristic information matrix and a multi-strategy parallel target ant colony optimization algorithm to obtain the initial charging and discharging strategy, evaluate the initial charging and discharging strategy, and obtain the evaluation result. The execution module is used to determine the optimal solution based on the evaluation results, output the energy storage charging and discharging power sequence corresponding to the optimal solution, and send it to the energy management system for execution as the final coordinated optimization scheduling scheme.
[0023] Compared with the prior art, the present invention has the following beneficial effects: This invention provides an energy storage system optimization method and system based on a multi-objective ant colony optimization algorithm. By constructing and assigning heterogeneous heuristic information matrices with differentiated optimization biases (such as environmental priority, economic priority, or a balance between economic and environmental protection) to different groups of intelligent agents, targeted exploration of multiple objectives such as economy and environmental protection is achieved during parallel search. This design effectively expands the coverage of the solution space, enhances the algorithm's global search capability in multi-objective trade-offs, and avoids getting trapped in local optima. The multi-strategy parallel search mechanism effectively avoids the algorithm getting trapped in local optima, generating higher-quality Pareto solutions; this enables the energy storage system to obtain a better coordinated solution in terms of economy and environmental protection, thereby maximizing its overall operational benefits.
[0024] Furthermore, a target ant colony optimization algorithm is executed in parallel by multiple groups of intelligent agents. Each group explores independently based on its heuristic bias, and the search results are integrated through an evaluation and selection mechanism, achieving synergy and complementarity among different optimization strategies. Operational inertia is introduced to generate the initial charging and discharging strategy, taking into account the continuity and stability constraints of the actual operation of the energy storage system. This makes the generated strategy not only theoretically superior but also more engineering feasible and smoother.
[0025] Furthermore, this method accelerates the search process during the initial strategy generation stage through parallel architecture and heuristic information guidance. Subsequently, it rapidly determines the optimal solution through centralized evaluation, and finally outputs a charge and discharge power sequence that can be directly issued to the energy management system for execution. The entire process achieves a closed loop from data to decision-making, ensuring that the scheduling scheme is optimized in a coordinated manner with economic and environmental goals, while also possessing good computational efficiency and engineering practicality. It is suitable for energy storage system optimization scheduling scenarios with certain real-time requirements. Attached Figure Description
[0026] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a comparison chart of total costs in Embodiment 2 of the present invention; Figure 3 This is a schematic diagram comparing carbon emissions in Embodiment 2 of the present invention; Figure 4 This is a schematic diagram of the microgrid operation in Embodiment 2 of the present invention; Figure 5 This is a schematic diagram of the peak shaving and valley filling strategy in Embodiment 2 of the present invention. Detailed Implementation
[0027] To further understand the content of this invention, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments are merely illustrative and not limiting of the invention.
[0028] Example 1 like Figure 1 As shown, an energy storage system optimization method based on a multi-objective ant colony optimization algorithm includes the following steps: S1: Obtain grid data for the energy storage system, including: grid time-of-use tariffs, grid carbon emission intensity, microgrid load demand, and renewable energy generation status; S2: Based on the acquired grid data of the energy storage system, construct several heterogeneous heuristic information matrices with different optimization biases; S3: Based on the constructed heterogeneous heuristic information matrix, a multi-strategy parallel target ant colony optimization algorithm is used for cooperative search to obtain the initial charging and discharging strategy. The initial charging and discharging strategy is evaluated to obtain the evaluation result. S4: Determine the optimal solution based on the evaluation results, output the energy storage charging and discharging power sequence corresponding to the optimal solution, and send it to the energy management system for execution as the final coordinated optimization scheduling scheme.
[0029] Specifically, in S1, the grid data of the energy storage system is acquired for 24 consecutive time steps with a time step of 1 hour, including: grid time-of-use electricity price, grid carbon emission intensity, microgrid load demand, and renewable energy generation status.
[0030] System model parameter definition defines the physical and operational constraints of the energy storage system (BESS), including: rated capacity C_max set to 3000 kWh, maximum charge / discharge power. For a power output of 1000kW, the charging efficiency η_ch is 0.9, the discharging efficiency η_dis is 0.9, and the initial state of charge... It is 0.5, and the expected state of charge at the end of the cycle. It is also 0.5.
[0031] Specifically, in S2, based on the acquired grid data from the energy storage system, several heterogeneous heuristic information sets with different optimization biases are constructed. The details are as follows: S21: Based on the acquired grid data of the energy storage system, and according to different target bias heuristic test rates, define heuristic strategy combinations as follows: Strategy 1: Environmental protection first, with a weighting combination of {cost weight: 0.35, carbon emission weight: 0.65}; Strategy 2: Balancing economic and environmental protection, with a weighted combination of {cost weight: 0.40, carbon emission weight: 0.60}; Strategy 3: Economic priority, with a weighting combination of {cost weight: 0.45, carbon emission weight: 0.55}; S22: For each time step t, generate a multidimensional guiding signal based on the defined heuristic strategy combination; For each time step t, calculate a normalized multidimensional guiding signal according to each of the above heuristic strategies. ,in The strategy number is assigned. The calculation formula is as follows:
[0032] in, Representative strategy The guiding signal at time t and These are the periodic averages of electricity price and carbon intensity, respectively. and They represent strategies respectively. Cost weights and carbon emission weights, and This represents the electricity price and carbon emission intensity at time t.
[0033] S23: Based on the multidimensional guiding signal generated by heuristic strategy combination, a heuristic information matrix is generated for each strategy.
[0034] Based on the guiding signal, for each strategy This generates a two-dimensional heuristic information matrix. This matrix represents the expected degree of action a, i.e., charging or discharging, at time t. Specifically: The continuous operating space of the energy storage system at each time t The signal is discretized into 11 action levels. For the discharge action, i.e., when a < 0, its expected level is proportional to the guiding signal. That is, when the ratio of electricity price to carbon emissions is high, discharge is encouraged.
[0035] For charging actions, i.e., when a>0, the desired level is inversely proportional to the guiding signal. When the ratio of electricity price to carbon emissions is low, charging is encouraged. In this case:
[0036] in To prevent extremely small positive numbers with a denominator of zero.
[0037] When a=0, A value of 0 or the baseline indicates a low degree of conformity to the expectation of small-amplitude movements.
[0038] At this point, three ( =1,2,3) mutually independent heuristic information matrices , , .
[0039] S3: Based on the constructed heterogeneous heuristic information matrix, a multi-strategy parallel target ant colony optimization algorithm is used for cooperative search to obtain an initial charging and discharging strategy. The initial charging and discharging strategy is then evaluated to obtain the evaluation results. Details are as follows: S31: Initialize the ant colony algorithm, including: creating an agent population containing multiple agents; initializing a globally shared two-dimensional pheromone matrix. All initial values are set to be the same, and a Pareto front archive is initialized to store non-dominated solutions. Algorithm hyperparameters: Define the control parameters of the optimization algorithm of this invention, including: ant colony size (number of agents) 120, maximum number of iterations 3000, and pheromone evaporation coefficient. The pheromone enhancement coefficient Q is 1.0, and the pheromone heuristic factor is 0.1. The value is 1.6, and the heuristic information heuristic factor The value is 2.
[0040] S32: Group the agents and assign heuristic information matrices, wherein multiple agents are divided into at least three groups, and an environmental priority heuristic matrix is assigned to the first group, an economic-environmental balance heuristic matrix is assigned to the second group, and an economic priority heuristic matrix is assigned to the third group. After initialization, the ant colony algorithm iteratively optimizes the algorithm. First, the agents are grouped and tasks are assigned. The 120 ants are divided into groups, and each group is assigned a heuristic information matrix. The first group is assigned the "environmental protection priority" heuristic matrix. The second group was assigned a "balanced economic and environmental protection" heuristic matrix. The third group's allocation of the "economic priority" heuristic matrix By having different groups explore in different directions, this diversity strategy greatly enhances the algorithm's ability to explore different regions of the solution space, avoiding premature entrapment in local optima.
[0041] S33: Construct parallel paths for the grouped agents. Each agent determines a set of possible actions based on its current state and selects the next action based on the state transition probability. After initialization and task allocation, all agents, starting from time t=0, construct a 24-hour scheduling path in parallel and independently. At each time t, each agent determines its set of available actions based on its current state (SOC(t)) and selects its next action based on the state transition probability. This probability is jointly determined by the heuristic information matrix of its group and the globally shared pheromone matrix.
[0042] in, Let be the probability of taking a charging or discharging action at time t. This is a pheromone matrix, representing the historical pheromone concentration at position (t, a), reflecting the successful experiences of ants choosing this position previously. The pheromone importance factor; Represents the position The immediate benefits are calculated based on real-time electricity prices and carbon intensity. As the importance factor for immediate returns, the probability of an ant choosing a certain action in this step is determined by both historical experience and immediate returns.
[0043] S34: After the agent completes the action selection at each moment, the initial charging and discharging strategy is generated by the inertia of operation. After the charging or discharging action at the previous moment ends, in order to make the scheduling scheme smoother, an inertia factor is introduced. The probability of selecting the same type of action at the current moment will be multiplied by an inertia coefficient greater than 1. This inertia coefficient is set to 1.8.
[0044] S35: Evaluate the generated initial charging and discharging strategy, check whether its final state of charge meets the preset constraints, if it does, determine that the scheduling path is a valid solution, and calculate the two target values of total cost and total carbon emissions corresponding to the valid solution. The constraint formula for the final state of charge is as follows:
[0045] in, This represents the absolute battery level at the end of 24 hours. The desired state of charge is set to 0.5. The maximum capacity of the battery is set to 3000 kWh, and 0.1 is the set threshold. The formula determines the absolute difference between the battery's final state of charge and the target final state. If the final SOC meets the preset constraints, the path is a valid solution, and the two target values of total cost and total carbon emissions corresponding to the valid solution are calculated. If not, the next iteration continues to generate and explore new charging and discharging paths until a valid path that meets the final power constraint is found, or the preset number of iterations is reached.
[0046] S36: Update the Pareto front archive by comparing all valid solutions generated in this iteration with the solutions in the Pareto front archive, and updating the Pareto front archive using the non-dominated sorting method to retain all non-dominated solutions; Furthermore, all pheromones in the globally shared pheromone matrix are proportionally... =0.1 is decayed, and pheromones are added only to paths corresponding to all non-dominated solutions in the current Pareto front archive.
[0047] Specifically, in S4, the optimal solution is determined based on the evaluation results, and the energy storage charging and discharging power sequence corresponding to the optimal solution is output as the final coordinated optimization scheduling scheme and sent to the energy management system for execution.
[0048] The algorithm terminates after reaching the maximum number of iterations. At this point, the Pareto front archive contains a series of optimal trade-off solutions found by this invention. Among all solutions, the cost and emissions values are normalized separately, and the solution with the smallest sum of the two normalized values is selected as the optimal balance point. Finally, the 24-hour energy storage charging and discharging power sequence corresponding to this optimal balance point is output as the final coordinated optimization scheduling scheme and sent to the energy management system for execution.
[0049] Example 2 This embodiment takes microgrid data optimization in a certain region of California, USA as an example. The data includes real-time electricity price, unmet load, carbon emission intensity, total load demand and photovoltaic power generation. This is used as the input dataset. The charging and discharging power sequence of the battery energy storage system (BESS) over 24 hours is used as the core optimization objective through a multi-heuristic ant colony algorithm. Each value in this sequence represents the action that the battery should perform in that hour. The results generated using the method described in Embodiment 1 are compared as follows.
[0050] Figure 2 The comparison of hourly costs before and after energy storage optimization is clearly shown, with an overall daily cost reduction of 2.2%. Specifically, the energy storage system performs charging operations during the early morning load curve and off-peak electricity price periods, and discharge operations during the evening peak electricity consumption and high electricity price periods, effectively avoiding the purchase of electricity at high prices. Overall, the optimization strategy effectively smooths the entire daily electricity cost curve, resulting in clear economic benefits.
[0051] Figure 3 The table compares hourly carbon emissions before and after optimization. The comparison shows that total carbon emissions increased slightly by 0.02%. Details show that the optimization strategy increased electricity purchases during the peak solar power generation periods in the early morning and midday, resulting in a slight increase in emissions. However, when grid carbon intensity is high in the evening, the optimization strategy significantly reduced electricity purchases, thereby greatly reducing emissions. The environmental benefits of prioritizing the consumption of clean electricity through energy storage and reducing electricity purchases during high-carbon emission periods are achieved through multi-objective optimization, resulting in a slight improvement in emissions.
[0052] Figure 4 This diagram illustrates the optimized operation of the microgrid. The black dashed line represents user load, and the orange line represents photovoltaic (PV) power generation. The red line, "Optimized Grid Power," is the core indicator, showing the optimized electricity demand purchased from the main grid. The blue bar chart represents the battery charging and discharging schedule: positive values indicate charging, and negative values indicate discharging. The diagram clearly shows that the batteries charge during off-peak hours in the early morning when electricity prices are low or during midday when PV power is abundant; while during peak load periods in the evening when electricity prices are high, the batteries discharge on a large scale, significantly reducing the grid's demand for electricity, successfully achieving peak shaving and valley filling, and suppressing grid power fluctuations.
[0053] Example 3 An energy storage system optimization system based on a multi-objective ant colony optimization algorithm includes: The acquisition module is used to acquire grid data from the energy storage system. The heuristic information matrix construction module is used to construct several heterogeneous heuristic information matrices with different optimization biases based on the acquired grid data of the energy storage system. The result generation module is used to perform cooperative search based on the constructed heterogeneous heuristic information matrix and a multi-strategy parallel target ant colony optimization algorithm to obtain the initial charging and discharging strategy, evaluate the initial charging and discharging strategy, and obtain the evaluation result. The execution module is used to determine the optimal solution based on the evaluation results, output the energy storage charging and discharging power sequence corresponding to the optimal solution, and send it to the energy management system for execution as the final coordinated optimization scheduling scheme.
[0054] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.
Claims
1. An optimization method for energy storage systems based on a multi-objective ant colony optimization algorithm, characterized in that, Includes the following steps: Acquire grid data from the energy storage system; Based on the acquired grid data of the energy storage system, several heterogeneous heuristic information matrices with different optimization biases are constructed. Based on the constructed heterogeneous heuristic information matrix, a multi-strategy parallel target ant colony optimization algorithm is used for cooperative search to obtain an initial charging and discharging strategy. The initial charging and discharging strategy is then evaluated to obtain the evaluation results. Based on the evaluation results, the optimal solution is determined, and the energy storage charging and discharging power sequence corresponding to the optimal solution is output as the final coordinated optimization scheduling scheme and sent to the energy management system for execution.
2. The energy storage system optimization method based on a multi-objective ant colony optimization algorithm according to claim 1, characterized in that, In the step of acquiring grid data for the energy storage system, the grid data includes: grid time-of-use electricity price, grid carbon emission intensity, microgrid load demand, and renewable energy generation status.
3. The energy storage system optimization method based on a multi-objective ant colony optimization algorithm according to claim 1, characterized in that, The method for constructing several heterogeneous heuristic information matrices with different optimization biases based on the acquired grid data of the energy storage system is as follows: Based on the acquired grid data of the energy storage system, heuristic strategy combinations are defined according to different target bias heuristic test rates. Multidimensional guiding signals are generated based on the combination of defined heuristic strategies; Based on the multidimensional guiding signals generated by heuristic strategy combinations, a heuristic information matrix is generated for each strategy.
4. The energy storage system optimization method based on multi-objective ant colony optimization algorithm according to claim 3, characterized in that, In the step of defining heuristic strategy combinations based on the acquired grid data of the energy storage system and according to different target bias heuristic test rates, the defined heuristic strategy combinations are as follows: Strategy 1: Environmental protection first, with a weighting combination of {cost weight: 0.35, carbon emission weight: 0.65}; Strategy 2: Balancing economic and environmental protection, with a weighted combination of {cost weight: 0.40, carbon emission weight: 0.60}; Strategy 3: Economic priority, with a weighting combination of {cost weight: 0.45, carbon emission weight: 0.55}.
5. The energy storage system optimization method based on a multi-objective ant colony optimization algorithm according to claim 4, characterized in that, The method for generating multidimensional guiding signals based on the definition-based heuristic strategy combination is as follows: For each time step t, a normalized multidimensional guiding signal is calculated based on each heuristic strategy, using the following formula: in, Representative strategy The guiding signal at time t and These are the periodic averages of electricity price and carbon intensity, respectively. and They represent strategies. Cost weights and carbon emission weights, and This represents the electricity price and carbon emission intensity at time t.
6. The energy storage system optimization method based on a multi-objective ant colony optimization algorithm according to claim 5, characterized in that, The method for generating a heuristic information matrix for each strategy in the multidimensional guiding signal based on heuristic strategy combination is as follows: Based on the guidance signal, the continuous operation space of the energy storage system at each time t is determined. Discretized into 11 action levels; For each policy k and each time t, the degree of expectation is quantified into a heuristic information matrix. ,in Indicates a discharge action; When a < 0, the desired level is proportional to the guiding signal, encouraging discharge: at this time, When a > 0, the expected level is inversely proportional to the guiding signal to encourage charging. At this time: in To prevent extremely small positive numbers with a denominator of zero; When a=0, It is 0 or the baseline value.
7. The energy storage system optimization method based on a multi-objective ant colony optimization algorithm according to claim 6, characterized in that, Based on the constructed heterogeneous heuristic information matrix, a multi-strategy parallel target ant colony optimization algorithm is used for cooperative search to obtain an initial charging and discharging strategy. The method for evaluating the initial charging and discharging strategy and obtaining the evaluation results is as follows: Initializing the ant colony algorithm includes: creating an agent population with multiple agents; and initializing a globally shared two-dimensional pheromone matrix. Initialize a Pareto front archive for storing nondominated solutions; Group the agents and assign heuristic information matrices, wherein multiple agents are divided into at least three groups, and an environmental priority heuristic matrix is assigned to the first group, an economic-environmental balance heuristic matrix is assigned to the second group, and an economic priority heuristic matrix is assigned to the third group. Parallel paths are constructed for the grouped agents. Each agent determines its set of possible actions based on its current state and selects the next action based on the state transition probability. After the agent completes the action selection at each moment, the initial charging and discharging strategy is generated by the inertia of operation. The generated initial charge and discharge strategy is evaluated to check whether its final state of charge meets the preset constraints, and the evaluation results are obtained.
8. The energy storage system optimization method based on a multi-objective ant colony optimization algorithm according to claim 7, characterized in that, In the step of constructing parallel paths for the grouped agents, where each agent determines its set of available actions based on its current state and selects its next action based on the state transition probability, the state transition probability is jointly determined by the heuristic information matrix of its group and the globally shared pheromone matrix. in, Let t be the probability of taking a charging or discharging action at time t; The pheromone matrix represents the historical pheromone concentration at position (t,a), reflecting the successful experience of ants choosing this position in the past. The pheromone importance factor; Instant benefits on the platform; The importance factor for immediate returns.
9. The energy storage system optimization method based on a multi-objective ant colony optimization algorithm according to claim 8, characterized in that, The generated initial charging and discharging strategy is evaluated to check whether its final state of charge meets the preset constraints. After obtaining the evaluation results, the Pareto front archive is updated. All valid solutions generated in this iteration are compared with the solutions in the Pareto front archive. The Pareto front archive is updated using a non-dominated sorting method to retain all non-dominated solutions.
10. An energy storage system optimization system based on a multi-objective ant colony optimization algorithm, characterized in that, include: The acquisition module is used to acquire grid data from the energy storage system. The heuristic information matrix construction module is used to construct several heterogeneous heuristic information matrices with different optimization biases based on the acquired grid data of the energy storage system. The result generation module is used to perform cooperative search based on the constructed heterogeneous heuristic information matrix and a multi-strategy parallel target ant colony optimization algorithm to obtain the initial charging and discharging strategy, evaluate the initial charging and discharging strategy, and obtain the evaluation result. The execution module is used to determine the optimal solution based on the evaluation results, output the energy storage charging and discharging power sequence corresponding to the optimal solution, and send it to the energy management system for execution as the final coordinated optimization scheduling scheme.