Energy storage configuration optimization method

Abstract

The invention relates to the technical field of power data processing, in particular to an energy storage configuration optimization method, which comprises the following steps: acquiring new energy output time sequence data and computing power load characteristic data; generating a space-time correlation coupling evaluation result of the new energy output volatility and the computing power load volatility; inputting a result into a hybrid power supply double-layer optimization model, recursively correcting parameters through a two-stage collaborative solution algorithm, and outputting a Pareto optimal solution set; a computing power task elastic regulation and control mechanism is embedded, and the task priority is dynamically adjusted according to the energy storage charge state and the new energy output level to generate a scheduling strategy; finally, an energy storage configuration scheme and a dynamic scheduling strategy are output, and collaborative optimization of cost effectiveness and power supply reliability is achieved. The method breaks through the coupling conflict of the economic target and the robust constraint in the traditional bilevel planning, remarkably reduces the energy storage configuration cost, and improves the system stability.

Description

Technical Field

[0001] This invention relates to the field of power data processing technology, and in particular to a method for optimizing energy storage configuration. Background Technology

[0002] In hybrid power supply models, integrating renewable energy with the existing grid introduces significant supply and demand fluctuations, threatening power system reliability. Therefore, energy storage systems require robust configurations to resist uncertain disturbances and maintain power supply stability. However, overemphasizing robustness may increase initial investment costs; thus, economic optimization is crucial. A multi-objective decision-making framework is used to comprehensively evaluate robustness performance and economic indicators, such as selecting energy storage capacity and charging / discharging strategies, to minimize total operating costs in risky scenarios. Ultimately, this achieves a highly reliable and low-cost energy management solution, supporting sustainable power supply development.

[0003] Existing energy storage configurations often present a conflict between balancing economic objectives and robustness requirements. Economic objectives emphasize minimizing total lifecycle costs, including initial investment, maintenance expenses, and operational losses; while robustness requires the system to maintain stable operation under extreme uncertainties, such as continuous low output from renewable energy sources or sudden load fluctuations. These two objectives often conflict in optimization models, making it difficult to balance cost-effectiveness and reliability in configuration schemes. For example, in the real-world scenario of a hybrid power supply model for intelligent computing centers, renewable energy output is intermittently affected by wind and solar resources, leading to several consecutive days of low wind speeds or weak sunlight, causing a sharp drop in green electricity supply. Simultaneously, computing load may surge due to sudden increases in tasks, exacerbating the supply-demand imbalance. Optimizing energy storage configurations by solely pursuing minimum economic efficiency tends to reduce storage capacity to lower initial investment. However, this approach cannot buffer load fluctuations during periods of low renewable energy output, triggering a sharp increase in grid purchase costs or power curtailment penalties, resulting in a decrease in green electricity absorption and compromised operational reliability. Conversely, excessively emphasizing robustness, such as configuring ultra-large capacity energy storage to cover worst-case conditions, while capable of handling extreme scenarios, significantly increases investment costs and resource idle rates, weakening the project's economic feasibility. This conflict stems from the coupling problem between the outer-layer economic optimization objective and the inner-layer robustness constraint in the bi-level programming model. It requires dynamic trade-offs through multi-objective collaborative algorithms, but solving the model involves uncertainty quantification and multi-dimensional variable interactions, increasing implementation complexity and decision-making risks. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an energy storage configuration optimization method that solves the problem of difficulty in balancing cost-effectiveness and reliability assurance caused by the coupling problem between economic optimization objectives and robust constraints in a two-level programming model.

[0005] To solve the above-mentioned technical problems, the specific details of the present invention are as follows: This invention provides an energy storage configuration optimization method, comprising: Step 1: Collect time-series data of new energy power output and computing load characteristics of the target area; Step 2: Based on the collected new energy output time-series data and computing load characteristic data, generate a coupled evaluation result, which characterizes the spatiotemporal correlation between the volatility of new energy output and the volatility of computing load. Step 3: Input the coupling evaluation results into the hybrid power supply dual-layer optimization model. The hybrid power supply dual-layer optimization model includes an outer economic optimization model and an inner robustness verification model. The outer economic optimization model minimizes the total life cycle cost with energy storage capacity and power as variables. The inner robustness verification model uses the mismatch period information in the coupling evaluation results as the constraint boundary to verify the system reliability under extreme scenarios. Step 4: Perform a two-stage collaborative solution through the hybrid power supply dual-layer optimization model. The two-stage collaborative solution includes a first-stage solution and a second-stage solution. In the first stage, the outer layer economic optimization model is solved to generate an initial energy storage configuration scheme. In the second stage, the initial energy storage configuration scheme is input into the inner layer robustness verification model for extreme scenario simulation and verification. The solution is iteratively corrected until the Pareto optimal solution set is obtained. Step 5: Based on the energy storage parameters in the Pareto optimal solution set, embed a computing power task elastic control mechanism to generate a dynamic scheduling strategy; the computing power task elastic control mechanism dynamically adjusts task priorities according to the energy storage state of charge and the output level of new energy sources, forming the core rules of the dynamic scheduling strategy. Step 6: Execute the dynamic scheduling strategy and monitor the system operation status in real time. Deploy the energy storage system according to the energy storage configuration scheme, synchronously apply task delay rules to regulate computing load, periodically collect cost-effectiveness index and power supply reliability index data, and output the final energy storage configuration scheme and dynamic scheduling strategy when the index data meets the collaborative optimization target threshold.

[0006] Furthermore, in the energy storage configuration optimization method of the present invention, step 1 includes: Perform new energy output data collection to obtain minute-level historical data on wind speed and irradiance, as well as photovoltaic power generation curves; Simultaneously, the computing load data acquisition action is performed to obtain a joint dataset composed of the power load curves of the intelligent computing center, which have been separated into basic load, peak task and burst task segments.

[0007] Furthermore, in the energy storage configuration optimization method of the present invention, step 2 includes: The joint dataset is cleaned to remove outliers and fill in missing data, resulting in a preprocessed dataset. Based on the structured processing of the preprocessed dataset, a time-series data matrix and a load curve matrix are generated. Using the time-series data matrix and load curve matrix as input, the coupled evaluation model is triggered to perform fluctuation correlation analysis.

[0008] Furthermore, in the energy storage configuration optimization method of the present invention, step 2 further includes: Based on the time series data of new energy power output and the characteristic data of computing load, calculate and output the values ​​of new energy power output volatility and computing load volatility. The values ​​of the new energy output volatility and the computing load volatility are input into the sliding window processor, and spatiotemporal correlation analysis is performed to obtain the quantitative index of volatility correlation. Based on the anomaly detection results of the aforementioned fluctuation correlation quantitative index, the mismatch period between the low output of new energy sources and the surge in load is identified, and then a capacity matching list is generated and stored.

[0009] Furthermore, in the energy storage configuration optimization method of the present invention, step 4 includes: Configure a static constraint set for the outer layer economic optimization model, the static constraint set including the upper limit value of energy storage power and the threshold value of charge and discharge efficiency; A dynamic input boundary is constructed using a capacity adaptation bottleneck list, and then input into the inner robustness verification model. In the inner robustness verification model, the robustness compliance condition is set: the green electricity consumption rate compliance rate is not lower than the set threshold as the verification indicator.

[0010] Furthermore, the energy storage configuration optimization method of the present invention, wherein the two-stage collaborative solution is performed through the hybrid power supply dual-layer optimization model, includes: The first stage of solving the outer layer optimization involves calling the outer layer economic optimization model and using the branch and bound algorithm to solve and output the initial energy storage configuration scheme parameter set. The second stage of solving and performing robust verification involves inputting the parameter set of the initial energy storage configuration scheme into the inner robust verification model, and generating 100 sets of extreme scenario datasets through Monte Carlo random sampling for reliability verification. The parameter iteration correction mechanism updates the initial energy storage configuration scheme parameter set according to a preset incremental step size when the failure rate of the verified scenario exceeds a set threshold. The updated parameter set is re-input into the inner robustness verification model for recursive verification until the Pareto front change rate is less than 5% and the scenario failure rate meets the standard. Then the iteration is terminated and the current parameter set is stored as the Pareto optimal solution element.

[0011] Furthermore, in the energy storage configuration optimization method of the present invention, step 6 includes: Define a task priority classification benchmark and create a task priority tag group that includes real-time task tags, semi-real-time task tags, and offline task tags; Establish a state condition response rule. When it is detected in real time that the output of new energy is lower than 30% of the rated value and the state of charge of energy storage is lower than 20%, the delayed execution instruction of the tasks associated with the semi-real-time task tag and the offline task tag will be automatically triggered. A scheduling impact quantification model is generated. Based on historical task delay execution data, the numerical correspondence between the task delay ratio and the power demand reduction is statistically analyzed, and the mapping relationship table is output to the dynamic scheduling strategy.

[0012] Furthermore, in the energy storage configuration optimization method of the present invention, step 6 further includes: The improved NSGA-II optimization processor is invoked, and the mapping table in the Pareto optimal solution set and dynamic scheduling strategy is used as input to perform synchronous optimization calculations of the total life cycle cost, task delay penalty cost and green energy consumption rate, and output a multi-objective optimization scheme. The multi-objective optimization scheme is input into a dynamic programming simulator and a continuous 7-day simulation verification of the new energy off-peak operating conditions is performed. The output configuration robustness compliance signal serves as the basis for the activation of the final energy storage configuration scheme.

[0013] Furthermore, in the energy storage configuration optimization method of the present invention, step 4 includes: Define an extreme scenario type group, which includes the first extreme scenario and the second extreme scenario; The first extreme scenario is that the wind and solar power output is less than 30% of the rated value for 24 hours continuously. The second extreme scenario is a sudden increase of 200% in computing power load and a continuous 12-hour period where the output of new energy sources is below the warning value; Perform a probability distribution calibration operation, analyze the occurrence probability distribution curve of the extreme scenario type based on historical disaster early warning data, and output the calibrated probability distribution curve to the inner robustness verification model as a probability benchmark for verifying reliability.

[0014] Furthermore, the energy storage configuration optimization method of the present invention further includes: Extract the initial investment cost, equipment maintenance cost, grid power purchase cost, and power curtailment penalty cost of energy storage, and construct a full life cycle cost tree; Perform a sensitivity scan of the energy storage capacity parameters on the cost tree and output the cost and capacity response surfaces; Based on the linear relationship equation between the cost and energy storage capacity fitted by the response surface, the equation is input into the objective function calculation process of the outer layer economic optimization model.

[0015] Beneficial effects of this invention; This invention accurately quantifies the spatiotemporal correlation between the volatility of renewable energy output and the volatility of computing power load through a coupled evaluation model, providing dynamic constraint boundaries for the two-layer optimization model. A two-stage collaborative solution algorithm decouples the outer-layer economic optimization objective from the inner-layer robustness verification, outputting a Pareto optimal solution set through recursive verification and parameter iteration correction, achieving a dynamic balance between economy and reliability. An embedded elastic control mechanism for computing power tasks dynamically adjusts task priorities based on renewable energy output levels and energy storage status, generating load demand reduction strategies to alleviate pressure on energy storage configuration. Finally, through a multi-objective optimization verification process, the life-cycle cost and green electricity absorption rate are simultaneously optimized. Combined with continuous extreme operating condition simulation, the configuration scheme achieves both cost-effectiveness and power supply reliability, systematically resolving the coupling conflict between economic objectives and robust constraints in traditional two-layer planning, and improving the decision-making quality and engineering feasibility of hybrid power supply systems. Attached Figure Description

[0016] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on the drawings without creative effort.

[0017] Figure 1 This is a flowchart of an energy storage configuration optimization method provided in an embodiment of the present invention. Detailed Implementation

[0018] To make the technical solution of the present invention clearer, the present invention will be clearly and completely described below with reference to specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. The present invention provided by various embodiments will be described in detail below with reference to the accompanying drawings. To better understand the purpose of the present invention, the present invention will be described in further detail below.

[0019] Please see Figure 1 The present invention provides an energy storage configuration optimization method, comprising: Step 1: Collect time-series data of new energy power output and computing load characteristics of the target area; Step 2: Based on the collected new energy output time-series data and computing load characteristic data, generate a coupled evaluation result, which characterizes the spatiotemporal correlation between the volatility of new energy output and the volatility of computing load. Step 3: Input the coupling evaluation results into the hybrid power supply dual-layer optimization model. The hybrid power supply dual-layer optimization model includes an outer economic optimization model and an inner robustness verification model. The outer economic optimization model minimizes the total life cycle cost with energy storage capacity and power as variables. The inner robustness verification model uses the mismatch period information in the coupling evaluation results as the constraint boundary to verify the system reliability under extreme scenarios. Step 4: Perform a two-stage collaborative solution through the hybrid power supply dual-layer optimization model. The two-stage collaborative solution includes a first-stage solution and a second-stage solution. In the first stage, the outer layer economic optimization model is solved to generate an initial energy storage configuration scheme. In the second stage, the initial energy storage configuration scheme is input into the inner layer robustness verification model for extreme scenario simulation and verification. The solution is iteratively corrected until the Pareto optimal solution set is obtained. Step 5: Based on the energy storage parameters in the Pareto optimal solution set, embed a computing power task elastic control mechanism to generate a dynamic scheduling strategy; the computing power task elastic control mechanism dynamically adjusts task priorities according to the energy storage state of charge and the output level of new energy sources, forming the core rules of the dynamic scheduling strategy. Step 6: Execute the dynamic scheduling strategy and monitor the system operation status in real time. Deploy the energy storage system according to the energy storage configuration scheme, synchronously apply task delay rules to regulate computing load, periodically collect cost-effectiveness index and power supply reliability index data, and output the final energy storage configuration scheme and dynamic scheduling strategy when the index data meets the collaborative optimization target threshold.

[0020] When collecting time-series data on renewable energy output and computing load characteristics in the target area, the process specifically includes obtaining minute-level historical wind speed and irradiance data from meteorological data interfaces, and extracting actual power generation curves from photovoltaic inverter monitoring systems. Simultaneously, power load data from the intelligent computing center is collected, and the load characteristics of base load, peak tasks, and sudden task periods are separated using smart meters to form a joint dataset. This step provides the raw input foundation for subsequent analysis, ensuring data coverage of key characteristics of renewable energy generation and computing power demand.

[0021] Based on the collected time-series data of renewable energy output and computing load characteristics, data cleaning is performed to remove abnormal sensor records and fill in missing values, generating a preprocessed dataset. This dataset is then structured to form a time-series data matrix and a load curve matrix. These matrices are input to drive a coupled evaluation model, performing quantitative calculations and correlation analysis of renewable energy output volatility and computing load volatility, outputting the coupled evaluation results. These results characterize the spatiotemporal correlation of volatility, providing an analytical basis for identifying system bottlenecks and directly supporting the output of the data acquisition step.

[0022] The coupled evaluation results are input into a hybrid power supply two-layer optimization model, which includes an outer economic optimization model and an inner robustness verification model. The outer model uses energy storage capacity and power as decision variables to minimize the total life-cycle cost, encompassing initial investment, maintenance costs, and grid purchase costs. The inner model uses mismatch period information from the coupled evaluation results as dynamic constraint boundaries to verify system reliability indicators under extreme scenarios. This step constructs the optimization framework, transforms the analysis results into a solvable mathematical model, and establishes a logical connection between the evaluation results and the optimization objective.

[0023] A two-stage collaborative solution is implemented using a hybrid power supply dual-layer optimization model. The first stage calls the branch-and-bound algorithm to solve the outer economic optimization model, generating an initial parameter set for the energy storage configuration scheme. The second stage inputs this parameter set into the inner robustness verification model, generating an extreme scenario dataset through Monte Carlo random sampling for reliability verification. When the scenario failure rate exceeds a threshold, the parameter set is updated incrementally and recursively input into the inner model for re-verification until the Pareto frontier change rate is lower than a set value and the scenario failure rate meets the target, at which point the iteration terminates, and the parameter set is stored as elements of the Pareto optimal solution. This process achieves a dynamic balance between economy and robustness, relying on the model structure to complete closed-loop optimization.

[0024] Based on the energy storage parameters in the Pareto optimal solution set, an elastic control mechanism for computing power tasks is embedded. By defining priority label groups for real-time tasks, semi-real-time tasks, and offline tasks, a state response rule is established: when real-time monitoring shows that the output of new energy sources is lower than the rated threshold and the state of charge of energy storage is lower than the critical value, a non-urgent task delay instruction is automatically triggered. Then, based on the historical task delay data, the mapping relationship between the task delay ratio and the decrease in power demand is statistically analyzed to generate a dynamic scheduling strategy. This step transforms the optimization results into an executable strategy, alleviating supply and demand conflicts through priority adjustment.

[0025] After executing the dynamic scheduling strategy, the energy storage configuration scheme and the dynamic scheduling strategy are output. The configuration scheme includes the combination of energy storage types and the capacity-power ratio, and the dynamic scheduling strategy includes task delay rules. Finally, by verifying the cost-effectiveness indicators and power supply reliability indicators, the synergistic optimization goal is achieved, forming a complete technical closed loop from data acquisition to decision output.

[0026] Specifically, in the energy storage configuration optimization method of the present invention, step 1 includes: Perform new energy output data collection to obtain minute-level historical data on wind speed and irradiance, as well as photovoltaic power generation curves; Simultaneously, the computing load data acquisition action is performed to obtain a joint dataset composed of the power load curves of the intelligent computing center, which have been separated into basic load, peak task and burst task segments.

[0027] When collecting time-series data on renewable energy output, minute-level historical records of wind speed and irradiance are obtained through a meteorological data interface, while power generation curves are extracted from the photovoltaic inverter monitoring system. Simultaneously, computing load data is collected, using smart meters to separate the power load curves of the intelligent computing center, identifying characteristics of base load, peak load periods, and sudden load periods, generating a joint dataset including wind and solar power generation data and load curves. This dataset provides structured input for subsequent analysis, facilitating the connection between data sources and processing steps.

[0028] Specifically, in the energy storage configuration optimization method of the present invention, step 2 includes: The joint dataset is cleaned to remove outliers and fill in missing data, resulting in a preprocessed dataset. Based on the structured processing of the preprocessed dataset, a time-series data matrix and a load curve matrix are generated. Using the time-series data matrix and load curve matrix as input, the coupled evaluation model is triggered to perform fluctuation correlation analysis.

[0029] Data cleaning is performed on the joint dataset to remove outliers caused by sensor malfunctions, and missing data is filled in using interpolation algorithms to output a preprocessed dataset. Based on the structured transformation of this dataset, a time-series data matrix including timestamps and numerical values ​​and a load curve matrix are generated. The matrix is ​​input into the coupled evaluation model to trigger the correlation analysis between the volatility of new energy power output and the volatility of computing power load, thus establishing a technical closed loop from data cleaning to model-driven processing.

[0030] Specifically, in the energy storage configuration optimization method of the present invention, step 2 further includes: Based on the time series data of new energy power output and the characteristic data of computing load, calculate and output the values ​​of new energy power output volatility and computing load volatility. The values ​​of the new energy output volatility and the computing load volatility are input into the sliding window processor, and spatiotemporal correlation analysis is performed to obtain the quantitative index of volatility correlation. Based on the anomaly detection results of the aforementioned fluctuation correlation quantitative index, the mismatch period between the low output of new energy sources and the surge in load is identified, and then a capacity matching list is generated and stored.

[0031] Based on the preprocessed renewable energy output time-series data and computing load characteristic data, the renewable energy output volatility rate is calculated as the ratio of standard deviation to mean, and the computing load volatility rate is calculated as the ratio of peak-to-valley difference to mean. The dual volatility values ​​are input into a sliding window processor to perform spatiotemporal correlation analysis and generate quantitative indicators. Based on the abnormal threshold detection results of these indicators, the mismatch periods of output troughs and load surges are identified, and a capacity adaptation bottleneck list is output to the optimization model, completing the data transfer from volatility analysis to bottleneck identification.

[0032] Specifically, in the energy storage configuration optimization method of the present invention, step 4 includes: Configure a static constraint set for the outer layer economic optimization model, the static constraint set including the upper limit value of energy storage power and the threshold value of charge and discharge efficiency; A dynamic input boundary is constructed using a capacity adaptation bottleneck list, and then input into the inner robustness verification model. In the inner robustness verification model, the robustness compliance condition is set: the green electricity consumption rate compliance rate is not lower than the set threshold as the verification indicator.

[0033] A set of static constraints is configured for the outer-layer economic optimization model, including a preset upper limit for energy storage power and a threshold for charging and discharging efficiency; dynamic boundary conditions are constructed using a capacity adaptation bottleneck list and input into the inner-layer robustness verification model; a minimum threshold for the green electricity consumption rate is set in the model as a verification indicator, forming a two-layer verification mechanism that combines static constraints and dynamic boundaries.

[0034] Specifically, the energy storage configuration optimization method of the present invention, wherein the two-stage collaborative solution is performed through the hybrid power supply dual-layer optimization model, includes: The first stage of solving the outer layer optimization involves calling the outer layer economic optimization model and using the branch and bound algorithm to solve and output the initial energy storage configuration scheme parameter set. The second stage of solving and performing robust verification involves inputting the parameter set of the initial energy storage configuration scheme into the inner robust verification model, and generating 100 sets of extreme scenario datasets through Monte Carlo random sampling for reliability verification. The parameter iteration correction mechanism updates the initial energy storage configuration scheme parameter set according to a preset incremental step size when the failure rate of the verified scenario exceeds a set threshold. The updated parameter set is re-input into the inner robustness verification model for recursive verification until the Pareto front change rate is less than 5% and the scenario failure rate meets the standard. Then the iteration is terminated and the current parameter set is stored as the Pareto optimal solution element.

[0035] The first stage calls the branch and bound algorithm to solve the outer model and outputs the initial energy storage capacity and power parameter set. The second stage inputs the parameter set into the inner model and generates an extreme scenario dataset through Monte Carlo random sampling to perform reliability verification. When the scenario failure rate exceeds the limit, the parameter set is updated according to the preset step size and the verification is performed recursively until the Pareto front rate of change converges and the failure rate meets the standard. The optimal solution element is then output, realizing iterative feedback between economic optimization and robust verification.

[0036] Specifically, in the energy storage configuration optimization method of the present invention, step 6 includes: Define a task priority classification benchmark and create a task priority tag group that includes real-time task tags, semi-real-time task tags, and offline task tags; Establish a state condition response rule. When it is detected in real time that the output of new energy is lower than 30% of the rated value and the state of charge of energy storage is lower than 20%, the delayed execution instruction of the tasks associated with the semi-real-time task tag and the offline task tag will be automatically triggered. A scheduling impact quantification model is generated. Based on historical task delay execution data, the numerical correspondence between the task delay ratio and the power demand reduction is statistically analyzed, and the mapping relationship table is output to the dynamic scheduling strategy.

[0037] Create priority label groups for real-time tasks, semi-real-time tasks, and offline tasks, and define task classification benchmarks; construct status response rules: when real-time monitoring shows that the output of new energy sources is lower than the rated threshold and the state of charge of energy storage is lower than the critical value, automatically trigger delay instructions for semi-real-time and offline tasks; based on historical delay data, statistically analyze the mapping relationship between task delay ratio and power reduction, generate dynamic scheduling strategy rules, and form a technical link from priority control to quantitative decision-making.

[0038] Specifically, in the energy storage configuration optimization method of the present invention, step 6 further includes: The improved NSGA-II optimization processor is invoked, and the mapping table in the Pareto optimal solution set and dynamic scheduling strategy is used as input to perform synchronous optimization calculations of the total life cycle cost, task delay penalty cost and green energy consumption rate, and output a multi-objective optimization scheme. The multi-objective optimization scheme is input into a dynamic programming simulator and a continuous 7-day simulation verification of the new energy off-peak operating conditions is performed. The output configuration robustness compliance signal serves as the basis for the activation of the final energy storage configuration scheme.

[0039] The Pareto optimal solution set and mapping table are input into the improved NSGA-II algorithm to simultaneously optimize the life cycle cost, task delay penalty cost, and green electricity consumption rate, and output a multi-objective optimization scheme. The scheme is then input into a dynamic programming simulator to perform continuous multi-day simulations of low-peak renewable energy conditions, and outputs a robust compliance signal as the basis for enabling the final configuration scheme, thus completing the decision-making closed loop from multi-objective optimization to scenario verification.

[0040] Specifically, in the energy storage configuration optimization method of the present invention, step 4 includes: Define an extreme scenario type group, which includes the first extreme scenario and the second extreme scenario; The first extreme scenario is that the wind and solar power output is less than 30% of the rated value for 24 hours continuously. The second extreme scenario is a sudden increase of 200% in computing power load and a continuous 12-hour period where the output of new energy sources is below the warning value; Perform a probability distribution calibration operation, analyze the occurrence probability distribution curve of the extreme scenario type based on historical disaster early warning data, and output the calibrated probability distribution curve to the inner robustness verification model as a probability benchmark for verifying reliability.

[0041] The first type of extreme scenario is defined as wind and solar power output continuously falling below the rated threshold, and the second type of scenario is defined as a sudden increase in computing load accompanied by a continuous drop in new energy power output below the warning value. Based on historical disaster early warning data, the probability distribution curves of the two types of scenarios are fitted, and the calibrated probability distribution is output to the inner robustness verification model as a probability benchmark for Monte Carlo sampling, realizing the technical collaboration from the definition of extreme scenarios to the generation of verification benchmarks.

[0042] Specifically, the energy storage configuration optimization method of the present invention further includes: Extract the initial investment cost, equipment maintenance cost, grid power purchase cost, and power curtailment penalty cost of energy storage, and construct a full life cycle cost tree; Perform a sensitivity scan of the energy storage capacity parameters on the cost tree and output the cost and capacity response surfaces; Based on the linear relationship equation between the cost and energy storage capacity fitted by the response surface, the equation is input into the objective function calculation process of the outer layer economic optimization model.

[0043] The system extracts the initial investment cost, equipment maintenance cost, grid purchase cost, and curtailment penalty cost of energy storage; integrates them into a hierarchical life-cycle cost tree; performs a sensitivity scan of energy storage capacity parameters on the cost tree, and outputs the response surface of cost as a function of capacity; based on the surface fitting, it fits a linear substitution equation between cost and capacity, inputs it into the outer model's objective function calculation process, and supports data-driven decision-making for economic optimization objectives. This invention addresses the challenge of coupling economic optimization objectives with robust constraints in a two-layer programming model. First, it simultaneously acquires time-series data on renewable energy output and computing load characteristics at the data acquisition end, generating a coupling evaluation result that quantifies the spatiotemporal correlation between renewable energy output volatility and computing load volatility. This result is directly input into the hybrid power supply two-layer optimization model, decoupling the economic objective and robust constraints into an independently optimized two-layer structure. The outer economic optimization model minimizes the life-cycle cost using energy storage capacity and power as variables, while the inner robustness verification model uses mismatch period information from the coupling evaluation result as boundary constraints to verify reliability in extreme scenarios.

[0044] Based on this, a two-stage collaborative solution algorithm is used to achieve dynamic equilibrium: the first stage solves the outer model to generate the initial energy storage configuration scheme, and the second stage inputs it into the inner model for extreme scenario simulation verification; when the scenario failure rate exceeds the limit, the parameters are recursively adjusted in incremental steps until the Pareto frontier rate of change converges, and the Pareto optimal solution set that balances cost and reliability is output. This process transforms the static conflict between economy and robustness into a dynamic tradeoff that can be iteratively optimized.

[0045] Further embedding a computing power task elastic control mechanism, dynamically adjusting task priorities based on the energy storage charge status and new energy output level: when the new energy output is detected to be lower than the rated threshold and the energy storage charge status is lower than the critical value, the non-urgent task delay execution instruction is automatically triggered; based on historical data statistical analysis of the mapping relationship between task delay ratio and power demand reduction, a dynamic scheduling strategy is generated to actively reduce load peak and reduce the pressure on energy storage capacity configuration.

[0046] Ultimately, through a multi-objective optimization verification process, the life-cycle cost, task delay penalty cost, and green energy consumption rate were simultaneously optimized. This was combined with simulations of continuous days of low-peak renewable energy conditions to verify the configuration's robustness. This resulted in a three-tiered collaborative mechanism: data-driven coupled evaluation, two-layer model decoupling optimization, and flexible control for proactive load reduction. This mechanism, while ensuring power supply reliability, controls the life-cycle cost within the Pareto optimal range, completely overcoming the technical bottleneck of balancing economic objectives and robustness constraints.

Claims

1. A method for optimizing energy storage configuration, characterized in that, include: Step 1: Collect time-series data of new energy power output and computing load characteristics of the target area; Step 2: Based on the collected new energy output time-series data and computing load characteristic data, generate a coupled evaluation result, which characterizes the spatiotemporal correlation between the volatility of new energy output and the volatility of computing load. Step 3: Input the coupling evaluation results into the hybrid power supply dual-layer optimization model. The hybrid power supply dual-layer optimization model includes an outer economic optimization model and an inner robustness verification model. The outer economic optimization model minimizes the total life cycle cost with energy storage capacity and power as variables, and the inner robustness verification model uses the mismatch period information in the coupling evaluation results as the constraint boundary. Step 4: Perform a two-stage collaborative solution through the hybrid power supply dual-layer optimization model. The two-stage collaborative solution includes a first-stage solution and a second-stage solution. In the first stage, the outer layer economic optimization model is solved to generate an initial energy storage configuration scheme. In the second stage, the initial energy storage configuration scheme is input into the inner layer robustness verification model for extreme scenario simulation and verification. The solution is iteratively corrected until the Pareto optimal solution set is obtained. Step 5: Based on the energy storage parameters in the Pareto optimal solution set, embed a computing power task elastic control mechanism to generate a dynamic scheduling strategy. The computing power task elastic control mechanism dynamically adjusts the task priority according to the energy storage charge state and the output level of new energy sources to form a dynamic scheduling strategy. Step 6: Execute the dynamic scheduling strategy and monitor the system operation status in real time. Deploy the energy storage system according to the energy storage configuration scheme, synchronously apply task delay rules to regulate computing load, periodically collect cost-effectiveness index and power supply reliability index data, and output the final energy storage configuration scheme and dynamic scheduling strategy when the index data meets the collaborative optimization target threshold.

2. The energy storage configuration optimization method according to claim 1, characterized in that, Step 1 includes: Perform new energy output data collection to obtain minute-level historical data on wind speed and irradiance, as well as photovoltaic power generation curves; Simultaneously, the computing load data acquisition action is performed to obtain a joint dataset composed of the power load curves of the intelligent computing center, which have been separated into basic load, peak task and burst task segments.

3. The energy storage configuration optimization method according to claim 2, characterized in that, Step 2 includes: The joint dataset is cleaned to remove outliers and fill in missing data, resulting in a preprocessed dataset. Based on the structured processing of the preprocessed dataset, a time-series data matrix and a load curve matrix are generated. Using the time-series data matrix and load curve matrix as input, the coupled evaluation model is triggered to perform fluctuation correlation analysis.

4. The energy storage configuration optimization method according to claim 3, characterized in that, Step 2 also includes: Based on the time series data of new energy power output and the characteristic data of computing load, calculate and output the values ​​of new energy power output volatility and computing load volatility. The values ​​of the new energy output volatility and the computing load volatility are input into the sliding window processor, and spatiotemporal correlation analysis is performed to obtain the quantitative index of volatility correlation. Based on the anomaly detection results of the aforementioned fluctuation correlation quantitative index, the mismatch period between the low output of new energy sources and the surge in load is identified, and then a capacity matching list is generated and stored.

5. The energy storage configuration optimization method according to claim 4, characterized in that, Step 4 includes: Configure a static constraint set for the outer layer economic optimization model, the static constraint set including the upper limit value of energy storage power and the threshold value of charge and discharge efficiency; A dynamic input boundary is constructed using a capacity adaptation bottleneck list, and then input into the inner robustness verification model. In the inner robustness verification model, the robustness compliance condition is set: the green electricity consumption rate compliance rate is not lower than the set threshold as the verification indicator.

6. The energy storage configuration optimization method according to claim 5, characterized in that, The two-stage collaborative solution process using the hybrid power supply dual-layer optimization model includes: The first stage of solving the outer layer optimization involves calling the outer layer economic optimization model and using the branch and bound algorithm to solve and output the initial energy storage configuration scheme parameter set. The second stage of solving and performing robust verification involves inputting the parameter set of the initial energy storage configuration scheme into the inner robust verification model, and generating 100 sets of extreme scenario datasets through Monte Carlo random sampling for reliability verification. The parameter iteration correction mechanism updates the initial energy storage configuration scheme parameter set according to a preset incremental step size when the failure rate of the verified scenario exceeds a set threshold. The updated parameter set is re-input into the inner robustness verification model for recursive verification until the Pareto front change rate is less than 5% and the scenario failure rate meets the standard. Then the iteration is terminated and the current parameter set is stored as the Pareto optimal solution element.

7. The energy storage configuration optimization method according to claim 6, characterized in that, Step 6 includes: Define a task priority classification benchmark and create a task priority tag group that includes real-time task tags, semi-real-time task tags, and offline task tags; Establish a state condition response rule. When it is detected in real time that the output of new energy is lower than 30% of the rated value and the state of charge of energy storage is lower than 20%, the delayed execution instruction of the tasks associated with the semi-real-time task tag and the offline task tag will be automatically triggered. Generate a scheduling impact quantification model, statistically analyze the correspondence between the task delay ratio and the power demand reduction based on historical task delay execution data, and output a mapping table to the dynamic scheduling strategy.

8. The energy storage configuration optimization method according to claim 7, characterized in that, Step 6 also includes: The improved NSGA-II optimization processor is invoked, and the mapping table in the Pareto optimal solution set and dynamic scheduling strategy is used as input to perform synchronous optimization calculations of the total life cycle cost, task delay penalty cost and green energy consumption rate, and output a multi-objective optimization scheme. The multi-objective optimization scheme is input into a dynamic programming simulator and a continuous 7-day simulation verification of the new energy off-peak operating conditions is performed. The output configuration robustness compliance signal serves as the basis for the activation of the final energy storage configuration scheme.

9. The energy storage configuration optimization method according to claim 8, characterized in that, Step 4 includes: Define an extreme scenario type group, which includes the first extreme scenario and the second extreme scenario; The first extreme scenario is that the wind and solar power output is less than 30% of the rated value for 24 hours continuously. The second extreme scenario is a sudden increase of 200% in computing power load and a continuous 12-hour period where the output of new energy sources is below the warning value; Perform a probability distribution calibration operation, analyze the occurrence probability distribution curve of the extreme scenario type based on historical disaster early warning data, and output the calibrated probability distribution curve to the inner robustness verification model as a probability benchmark for verifying reliability.

10. The energy storage configuration optimization method according to claim 9, characterized in that, Also includes: Extract the initial investment cost, equipment maintenance cost, grid power purchase cost, and power curtailment penalty cost of energy storage, and construct a full life cycle cost tree; Perform a sensitivity scan of the energy storage capacity parameters on the cost tree and output the cost and capacity response surfaces; Based on the linear relationship equation between the cost and energy storage capacity fitted by the response surface, the equation is input into the objective function calculation process of the outer layer economic optimization model.

Images