Energy storage power station business model collaborative optimization method and system based on multi-objective programming

Abstract

The application provides a method and system for collaborative optimization of a commercial model of an energy storage power station based on multi-objective programming, relates to the technical field of energy storage power station operation, and comprises the following steps: obtaining a commercial model feature set by performing space-time dimension decoupling on operation data of the energy storage power station; establishing a multi-objective programming constraint system; obtaining a commercial model combination scheme based on a Pareto front search strategy; generating a dispatch execution instruction according to the commercial model combination scheme; and identifying and adjusting the capacity allocation of the commercial model in conflict. The application realizes dynamic collaborative optimization of multiple commercial models of the energy storage power station, and improves economic benefits and operation safety.

Description

Technical Field

[0001] This invention relates to the field of energy storage power station operation technology, and in particular to a collaborative optimization method and system for energy storage power station business models based on multi-objective planning. Background Technology

[0002] With the large-scale grid integration of renewable energy and the transformation of power system structure, energy storage power stations, as an important grid regulation resource, are playing an increasingly important role in peak shaving, frequency regulation, reserve capacity, and demand response. Energy storage power stations can participate in multiple business models simultaneously, including the electricity ancillary services market, the energy market, and the capacity market, to improve their economic efficiency and social value. Optimizing the business model of energy storage power stations is a key issue in the current operation of the power system and the energy transition process, and is of great significance for promoting the consumption of new energy and ensuring the stable operation of the power grid.

[0003] Traditional optimization methods fail to effectively decouple the spatiotemporal characteristics of energy storage power station business models, making it difficult to accurately capture the mutual influence between different business models and their differentiated demands for energy storage resources. This results in discrepancies between optimization results and actual operating conditions. Existing business model optimization methods are typically oriented towards a single economic objective, lacking the ability to dynamically adjust to risk constraints and failing to achieve a flexible balance between returns and risks. This is particularly problematic under volatile market conditions, making it difficult to guarantee the stable operation and long-term profitability of energy storage power stations. When multiple business models operate simultaneously, energy storage power stations often encounter capacity competition and timing conflicts. Existing technologies lack effective real-time collaborative optimization mechanisms, failing to dynamically adjust resource allocation among business models based on actual operating conditions. This hinders seamless integration and global optimization between business models, reducing the utilization efficiency of energy storage resources. Summary of the Invention

[0004] This invention provides a method and system for collaborative optimization of energy storage power station business models based on multi-objective programming, which can solve the problems in the prior art.

[0005] A first aspect of this invention provides a collaborative optimization method for energy storage power station business models based on multi-objective programming, comprising:

[0006] The spatiotemporal dimensions of the energy storage power station operation data are decoupled to obtain a business model feature set. The business model feature set is then combined with the economic optimization objective and the risk constraint optimization objective to establish a multi-objective programming constraint system.

[0007] The multi-objective programming constraint system is solved iteratively using a Pareto front search strategy to obtain a non-dominated solution set. During the solution process, the weight coefficients of each schedulable feature vector in the business model feature set are dynamically adjusted according to the risk constraint optimization objective. Under the condition of satisfying the risk constraint optimization objective, the optimal solution set of the economic optimization objective is obtained, and the business model combination scheme is obtained.

[0008] Based on the time allocation ratio of each business model in the aforementioned business model combination scheme, and combined with the capacity status data of the energy storage power station, a scheduling execution instruction is generated and executed.

[0009] During instruction execution, business models with overlapping timelines and exceeding capacity limits in the business model combination scheme are identified. By calculating the capacity status data usage and release time of each business model, a timeline priority ranking is established. Based on the timeline priority ranking, capacity allocation is adjusted for conflicting business models, and the adjusted capacity allocation results are fed back to the schedulable feature vector corresponding to the business model feature set, thereby realizing dynamic collaboration between business models.

[0010] Decoupling the spatiotemporal dimensions of energy storage power station operation data yields a business model feature set. This business model feature set is then combined with economic optimization objectives and risk-constrained optimization objectives to establish a multi-objective programming constraint system, including:

[0011] The energy storage power station operation data includes charging and discharging power time-series data, state of charge time-series data, and load demand time-series data.

[0012] The charging and discharging power time series data is decomposed on a time scale to extract peak and valley period features and periodic features; the state of charge time series data is spatially mapped to identify state transition paths and extract the capacity utilization depth features and state duration features corresponding to the state transition paths.

[0013] The coupling relationship between the peak and valley time period characteristics and the capacity utilization depth characteristics is calculated to generate a time-shiftable schedulable feature vector. The coupling relationship between the periodicity characteristics and the state duration characteristics is calculated to generate an auxiliary service schedulable feature vector. The matching relationship between the load demand time series data and the state transition path is calculated to generate a demand management schedulable feature vector. The time-shiftable schedulable feature vector, the auxiliary service schedulable feature vector, and the demand management schedulable feature vector constitute the business model feature set.

[0014] The economic optimization objective quantifies the revenue of each business model, and the risk constraint optimization objective quantifies the degree of coupling conflict between business models. The economic optimization objective and the risk constraint optimization objective are used as dual optimization objectives, and the business model feature set is used as decision variables to establish the multi-objective programming constraint system.

[0015] The multi-objective programming constraint system is solved iteratively using a Pareto front search strategy to obtain a non-dominated solution set. During the solution process, the weight coefficients of each schedulable feature vector in the business model feature set are dynamically adjusted according to the risk constraint optimization objective, including:

[0016] An initial solution set is generated within the decision space corresponding to the business model feature set; the coupling conflict degree between the revenue value of the economic optimization objective and the risk constraint optimization objective is calculated, a dual-objective fitness space is constructed, and the non-dominated solutions of the initial solution set are identified in the dual-objective fitness space and stored in the Pareto front set;

[0017] Based on the coupling conflict degree value corresponding to each non-dominated solution in the Pareto front set, the conflict contribution degree of the business model feature set in terms of time-series occupancy and capacity occupancy is calculated. According to the conflict contribution degree and the non-dominated solution with the largest economic optimization objective benefit value, the weight coefficients between the time-shift schedulable feature vector, the auxiliary service schedulable feature vector, and the demand management schedulable feature vector are adaptively adjusted to generate a candidate solution set. The candidate solution set is then compared with the Pareto front set in terms of non-dominated relationship, and the Pareto front set is iteratively updated based on the result of the non-dominated relationship comparison.

[0018] Under the condition of satisfying the risk constraint optimization objective, the optimal solution set of the economic optimization objective is obtained, resulting in business model combination schemes including:

[0019] Calculate the total capacity utilization corresponding to each non-dominated solution in the Pareto front set, and identify the time period when the total capacity utilization exceeds the rated capacity of the energy storage power station as a capacity over-limit risk indicator; calculate the number of overlapping time periods in the time series dimension of the business model feature set, and use the number of overlapping time periods as a time series conflict risk indicator.

[0020] Determine whether the capacity over-limit risk index and the temporal conflict risk index corresponding to each non-dominated solution in the Pareto front set meet the threshold condition of the risk constraint optimization objective. From the Pareto front set that meets the threshold condition, select the non-dominated solution with the largest economic optimization objective benefit value, and take the weight coefficients of each schedulable feature vector corresponding to the non-dominated solution as the optimal solution set.

[0021] Based on the optimal solution set, the time allocation ratios of the power time-shifting business model, the ancillary service response business model, and the demand management business model are adjusted respectively within the planning period. The business model combination scheme is generated by combining the peak and valley time characteristics, periodic characteristics, and the matching relationship in the demand management schedulable feature vector.

[0022] Based on the time-series allocation ratio of each business model in the aforementioned business model combination scheme, and combined with the capacity status data of the energy storage power station, a scheduling execution instruction is generated and executed, including:

[0023] Acquire capacity status data of the energy storage power station, including the current state of charge value, available charging capacity, and available discharging capacity;

[0024] The planning period is divided into multiple scheduling time slots. The duration ratio of the power time-shifting business model, the ancillary service response business model, and the demand management business model in each scheduling time slot is determined, and a time-series resource allocation matrix is ​​generated. For each scheduling time slot, the charging interval, discharging interval, or standby interval of the energy storage power station is determined based on the current state of charge value, and the available charging capacity and the available discharging capacity are allocated to generate a capacity resource allocation matrix.

[0025] Based on the time-series resource allocation matrix and the capacity resource allocation matrix, the capacity share and duration allocated to the corresponding business model are calculated to obtain the charging and discharging power instruction value. Combined with the business model execution sequence of each scheduling time slot, the scheduling execution instruction is generated.

[0026] The scheduling execution command is sent to the power control unit of the energy storage power station, and the power control unit performs charging or discharging operations.

[0027] Identify business models with overlapping timelines and capacity limitations in the aforementioned business model combination schemes. Establish a timeline priority ranking by calculating the capacity status data usage and release time for each business model, including:

[0028] The overlap duration of actual execution periods between various business models is statistically analyzed, and combinations of business models with an overlap duration greater than zero are marked as time-overlapping business model combinations.

[0029] For each business model in the time-overlapping business model combination, based on the charging and discharging power command value corresponding to each business model and the duration of the actual execution period, the capacity occupancy requirement of each business model on the capacity status data is obtained and summed. If the summation result exceeds the available charging capacity or available discharging capacity of the energy storage power station, the time-overlapping business model combination is marked as a capacity-over-limit business model combination.

[0030] For each business model in the capacity over-limit business model combination, calculate the release time of the energy storage power station capacity for each business model. The release time is the end time of the actual execution period of each business model. Based on the release time, calculate the capacity occupancy duration of each business model.

[0031] Based on the capacity occupancy demand and the duration of capacity occupancy, a time-series priority ranking rule is established. The business model with the shortest duration of capacity occupancy and the smallest capacity occupancy demand is assigned the highest time-series priority and sorted in descending order to generate the time-series priority ranking.

[0032] Based on the time-series priority ranking, capacity allocation is adjusted for conflicting business models, and the adjusted capacity allocation results are fed back to the schedulable feature vector corresponding to the business model feature set, realizing dynamic collaboration between business models, including:

[0033] According to the time priority from high to low, the available charging capacity or available discharging capacity of the energy storage station is allocated to each business model in sequence, and the full amount is allocated according to the capacity occupancy demand of the business model; when the remaining capacity of the energy storage station is less than the capacity occupancy demand of the current business model, the remaining capacity is allocated to the current business model, and the capacity deficit of the current business model is calculated; and the adjusted capacity allocation value of each business model is generated.

[0034] The adjusted capacity allocation value is fed back to the business model feature set. Based on the updated capacity allocation parameters in the business model feature set, the actual total capacity occupancy of the time-shiftable schedulable feature vector, the ancillary service schedulable feature vector, and the demand management schedulable feature vector in the current scheduling time slot is recalculated. It is then determined whether the actual total capacity occupancy meets the capacity constraint conditions of the energy storage power station. For business models with capacity deficits, compensation instructions are generated and executed in the current scheduling time slot to compensate for the capacity deficits. The scheduling execution instructions are updated to achieve dynamic coordination between business models.

[0035] A second aspect of the present invention provides a collaborative optimization system for energy storage power station business models based on multi-objective programming, comprising:

[0036] The first unit is used to decouple the operation data of the energy storage power station in the spatiotemporal dimension to obtain the business model feature set. The business model feature set is combined with the economic optimization objective and the risk constraint optimization objective to establish a multi-objective programming constraint system.

[0037] The second unit is used to iteratively solve the multi-objective programming constraint system for non-dominated solution set based on the Pareto front search strategy. During the solution process, the weight coefficients of each schedulable feature vector in the business model feature set are dynamically adjusted according to the risk constraint optimization objective. Under the condition of satisfying the risk constraint optimization objective, the optimal solution set of the economic optimization objective is obtained, and the business model combination scheme is obtained.

[0038] The third unit is used to generate and execute scheduling execution instructions based on the time allocation ratio of each business model in the business model combination scheme and the capacity status data of the energy storage power station.

[0039] The fourth unit is used to identify business models with overlapping timelines and exceeding capacity limits in the business model combination scheme during instruction execution. By calculating the capacity status data usage and release time of each business model, a time priority ranking is established. Based on the time priority ranking, capacity allocation is adjusted for conflicting business models, and the adjusted capacity allocation result is fed back to the schedulable feature vector corresponding to the business model feature set to achieve dynamic collaboration between business models.

[0040] A third aspect of the embodiments of the present invention,

[0041] An electronic device is provided, comprising:

[0042] processor;

[0043] Memory used to store processor-executable instructions;

[0044] The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0045] Fourth aspect of the embodiments of the present invention,

[0046] A computer-readable storage medium is provided, having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0047] The beneficial effects of this application are as follows:

[0048] By decoupling the spatiotemporal dimensions of energy storage power station operation data and establishing a multi-objective planning constraint system, the precise extraction of business model characteristics and the quantitative expression of optimization objectives are achieved, enabling energy storage power stations to achieve a scientific balance between economic benefits and risk control, thereby improving the systematicness and reliability of commercial operation.

[0049] By employing a Pareto front search strategy to iteratively solve the non-dominated solution set, combined with a dynamic weight adjustment mechanism, the economically optimal solution can be found while satisfying risk constraints. This effectively solves the problem of limited returns and concentrated risks under the traditional single business model, and enhances the profitability and risk resistance of energy storage power stations.

[0050] By establishing a time-priority ranking and dynamic coordination mechanism among business models, the resource conflict problem in the parallel operation of multiple business models is solved, and the efficient allocation and real-time adjustment of energy storage capacity are realized. This significantly improves the resource utilization and scheduling flexibility of energy storage power stations, and provides more intelligent decision support for power station operation. Attached Figure Description

[0051] Figure 1 This is a flowchart illustrating the collaborative optimization method for energy storage power station business models based on multi-objective programming, as described in an embodiment of the present invention.

[0052] Figure 2 This is a schematic diagram of the Pareto front search strategy process. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. 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.

[0054] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0055] Figure 1 This is a flowchart illustrating the collaborative optimization method for energy storage power station business models based on multi-objective programming, as described in an embodiment of the present invention. Figure 1 As shown, the method includes:

[0056] The spatiotemporal dimensions of the energy storage power station operation data are decoupled to obtain a business model feature set. The business model feature set is then combined with the economic optimization objective and the risk constraint optimization objective to establish a multi-objective programming constraint system.

[0057] The multi-objective programming constraint system is solved iteratively using a Pareto front search strategy to obtain a non-dominated solution set. During the solution process, the weight coefficients of each schedulable feature vector in the business model feature set are dynamically adjusted according to the risk constraint optimization objective. Under the condition of satisfying the risk constraint optimization objective, the optimal solution set of the economic optimization objective is obtained, and the business model combination scheme is obtained.

[0058] Based on the time allocation ratio of each business model in the aforementioned business model combination scheme, and combined with the capacity status data of the energy storage power station, a scheduling execution instruction is generated and executed.

[0059] During instruction execution, business models with overlapping timelines and exceeding capacity limits in the business model combination scheme are identified. By calculating the capacity status data usage and release time of each business model, a timeline priority ranking is established. Based on the timeline priority ranking, capacity allocation is adjusted for conflicting business models, and the adjusted capacity allocation results are fed back to the schedulable feature vector corresponding to the business model feature set, thereby realizing dynamic collaboration between business models.

[0060] In one optional implementation, the spatiotemporal dimensions of the energy storage power station operation data are decoupled to obtain a business model feature set. This business model feature set is then combined with economic optimization objectives and risk-constrained optimization objectives to establish a multi-objective programming constraint system, including:

[0061] The energy storage power station operation data includes charging and discharging power time-series data, state of charge time-series data, and load demand time-series data.

[0062] The charging and discharging power time series data is decomposed on a time scale to extract peak and valley period features and periodic features; the state of charge time series data is spatially mapped to identify state transition paths and extract the capacity utilization depth features and state duration features corresponding to the state transition paths.

[0063] The coupling relationship between the peak and valley time period characteristics and the capacity utilization depth characteristics is calculated to generate a time-shiftable schedulable feature vector. The coupling relationship between the periodicity characteristics and the state duration characteristics is calculated to generate an auxiliary service schedulable feature vector. The matching relationship between the load demand time series data and the state transition path is calculated to generate a demand management schedulable feature vector. The time-shiftable schedulable feature vector, the auxiliary service schedulable feature vector, and the demand management schedulable feature vector constitute the business model feature set.

[0064] The economic optimization objective quantifies the revenue of each business model, and the risk constraint optimization objective quantifies the degree of coupling conflict between business models. The economic optimization objective and the risk constraint optimization objective are used as dual optimization objectives, and the business model feature set is used as decision variables to establish the multi-objective programming constraint system.

[0065] The acquisition of operational data for energy storage power stations includes time-series data on charge and discharge power, state of charge (SCC) data, and load demand data. Charge and discharge power time-series data records the charging and discharging power values ​​of the energy storage station at each point in time, typically in kilowatts (kW) or megawatts (MW). For example, the charge and discharge power data for a certain energy storage station over 24 hours might be: [2.5, 3.2, -1.8, -4.5, 2.1...], where positive values ​​represent charging power and negative values ​​represent discharging power. SCC time-series data represents the percentage of battery charge at each point in time, such as [85%, 92%, 88%, 76%...]. Load demand time-series data records user electricity demand, such as [6.8, 5.4, 7.2, 8.9...] megawatts.

[0066] The charging and discharging power time-series data is decomposed into high-frequency and low-frequency components using wavelet transform. The high-frequency components reflect short-term fluctuations, while the low-frequency components reflect long-term trends. By analyzing these components, peak and valley period characteristics are extracted, such as the peak occurrence time period (e.g., 8:00-12:00 and 17:00-21:00), peak amplitude (e.g., 5.6MW), valley duration period (e.g., 1:00-5:00), and valley amplitude (e.g., 0.8MW). Periodic features, including daily, weekly, and seasonal periods, are extracted, and the amplitude and phase information of each period are recorded. For example, the daily periodic feature shows that the peak power on weekdays is 30% higher than that on weekends, and the seasonal feature shows that the power fluctuation in summer is 50% greater than that in spring.

[0067] Spatial state mapping is performed on the state-of-charge (POC) time-series data, converting the POC data into trajectories in state space. Different POC intervals are defined, such as high POC (80%-100%), medium POC (40%-80%), and low POC (0%-40%). By analyzing the POC transitions between these intervals, state transition paths are identified, such as from high POC to medium POC, and from medium POC to low POC. Capacity is extracted using depth features, calculating the depth of each charge-discharge cycle, such as full charge-discharge (80% depth) or shallow charge-discharge (20% depth). State duration features are extracted, recording the duration of the battery in each POC interval, such as an average duration of 4 hours in the high POC interval and an average duration of 8 hours in the medium POC interval.

[0068] The coupling relationship between peak and valley time characteristics and capacity utilization depth characteristics is calculated, and the correlation between charging and discharging peaks and valleys and battery utilization depth is analyzed. For example, when the power grid is in peak electricity demand (e.g., 18:00-20:00), the battery undergoes deep discharge (utilization depth of 75%); during off-peak electricity demand (e.g., 2:00-5:00), it undergoes full charging. This coupling relationship generates a time-shifted dispatchable feature vector, which includes dispatchability indicators for time-shifted peak and valley filling, such as peak-valley electricity price difference (0.4 yuan / kWh), mobile load ratio (65%), and time-shifted economic benefits (0.25 yuan per kilowatt-hour).

[0069] The coupling relationship between periodic characteristics and state duration characteristics is calculated to analyze the matching degree between battery charge / discharge cycles and state duration. For example, when the grid frequency regulation demand exhibits a 4-hour cycle, the battery's mid-charge zone duration is also approximately 4 hours, indicating suitability for providing frequency regulation services. This coupling relationship generates an ancillary service schedulable feature vector, which includes ancillary service capability indicators such as frequency response rate (1 MW / min), continuous power supply time (2.5 hours), and voltage support capability (±3%).

[0070] The matching relationship between load demand time-series data and state transition paths is calculated to analyze whether battery state changes can match load demand changes. For example, before the load peak (e.g., 16:00), the battery is in a high-charge state; during the load peak (e.g., 18:00), the battery transitions to a medium-low charge state, indicating that battery discharge supports peak load. This matching relationship generates a demand management schedulable feature vector, which includes demand response indicators such as load peak shaving rate (25%), demand response reliability (92%), and demand management economic value (0.6 yuan per kW).

[0071] The business model feature set is composed of time-shiftable schedulable feature vectors (including peak-valley price difference utilization rate, time-shift capacity utilization rate, etc.), ancillary service schedulable feature vectors (including frequency regulation capability, voltage support capability, etc.), and demand management schedulable feature vectors (including load peak shaving capability, demand response flexibility, etc.), which comprehensively describe the business value of energy storage power stations.

[0072] The revenue of each business model is calculated to obtain the economic optimization target, including peak shaving and valley filling revenue (5000 yuan per day), ancillary service revenue (3500 yuan per day), and demand management revenue (4200 yuan per day). The total revenue is calculated as the sum of the revenue of each model, while also considering the costs of each model, including battery wear and tear costs, operation and maintenance costs, etc., to obtain the net revenue.

[0073] Risk constraints optimize the degree of coupling conflict between business models. For example, the conflict degree between time-shifting mode and ancillary service mode is 0.65, meaning that these two modes cannot operate simultaneously 65% ​​of the time; the conflict degree between ancillary service and demand management is 0.4; and the conflict degree between time-shifting and demand management is 0.3. The total conflict degree is calculated as the weighted sum of the conflict degrees between each pair of modes.

[0074] The multi-objective programming constraint system takes economic optimization objective and risk constraint optimization objective as dual optimization objectives, and takes business model characteristic set as decision variables. The constraints include energy storage capacity limit (e.g., total capacity 10MWh), charging and discharging power limit (e.g., maximum charging and discharging power 5MW), state of charge limit (e.g., maintaining within the range of 20%-95%), and technical requirements of each business model (e.g., frequency response requires at least 2 hours of continuous power supply). By solving this multi-objective programming problem, the optimal configuration ratio of each business model is obtained, achieving a balance between maximizing economic benefits and minimizing risks.

[0075] In one optional implementation, the multi-objective programming constraint system is solved iteratively using a Pareto front search strategy to obtain a non-dominated solution set. During the solution process, the weight coefficients of each schedulable feature vector in the business model feature set are dynamically adjusted according to the risk constraint optimization objective, including:

[0076] An initial solution set is generated within the decision space corresponding to the business model feature set; the coupling conflict degree between the revenue value of the economic optimization objective and the risk constraint optimization objective is calculated, a dual-objective fitness space is constructed, and the non-dominated solutions of the initial solution set are identified in the dual-objective fitness space and stored in the Pareto front set;

[0077] Based on the coupling conflict degree value corresponding to each non-dominated solution in the Pareto front set, the conflict contribution degree of the business model feature set in terms of time-series occupancy and capacity occupancy is calculated. According to the conflict contribution degree and the non-dominated solution with the largest economic optimization objective benefit value, the weight coefficients between the time-shift schedulable feature vector, the auxiliary service schedulable feature vector, and the demand management schedulable feature vector are adaptively adjusted to generate a candidate solution set. The candidate solution set is then compared with the Pareto front set in terms of non-dominated relationship, and the Pareto front set is iteratively updated based on the result of the non-dominated relationship comparison.

[0078] An initial solution set is generated within the decision space corresponding to the business model feature set. This initial solution set contains multiple potential solutions, each representing a combination of schedulable feature vectors from the business model feature set, including time-shifted schedulable feature vectors, ancillary service schedulable feature vectors, and demand management schedulable feature vectors. The initial solution set can be generated by random sampling in the decision space. For example, for a business model feature set containing 10 time-shifted schedulable features, 8 ancillary service schedulable features, and 12 demand management schedulable features, 100 initial solutions can be randomly generated, each representing a different combination of these features.

[0079] like Figure 2 As shown, the method includes:

[0080] For each solution in the generated initial solution set, calculate the coupling conflict degree between the economic optimization objective's revenue value and the risk constraint optimization objective's revenue value. The economic optimization objective's revenue value can be obtained by calculating the expected revenue of the business model corresponding to the solution. For example, the revenue value of a certain solution is 5 million yuan. The coupling conflict degree value reflects the degree of conflict between the solution in satisfying the economic objective and the risk constraint. It can be calculated by analyzing the resource competition situation of the business model corresponding to the solution in terms of time occupancy and capacity occupancy. For example, the coupling conflict degree value of a certain solution is 0.35 (ranging from 0 to 1, with a larger value indicating a more severe conflict).

[0081] Based on the calculated payoff and coupling conflict values, a dual-objective fitness space is constructed. In this two-dimensional space, the horizontal axis represents the payoff value, and the vertical axis represents the coupling conflict value. All solutions in the initial solution set are mapped to this fitness space, forming a series of points. By performing non-dominated relation analysis on these points, non-dominated solutions in the initial solution set are identified. A non-dominated solution is one that is not dominated by any other solution on any objective. For example, if solution A has a payoff of 6 million yuan and a coupling conflict degree of 0.4, and solution B has a payoff of 5.5 million yuan and a coupling conflict degree of 0.45, then solution A dominates solution B because A is superior to B on both objectives. In this way, approximately 20 non-dominated solutions can be selected from the initial 100 solutions and stored in the Pareto front set.

[0082] The conflict contribution of the business model feature set to time occupancy and capacity occupancy is calculated. For each non-dominated solution, the influence of time-shift schedulable features, ancillary service schedulable features, and demand management schedulable features on time occupancy and capacity occupancy is analyzed. For example, it is found that in a certain non-dominated solution, the conflict contribution of time-shift schedulable features to time occupancy is 0.5, the conflict contribution of ancillary service schedulable features to time occupancy is 0.3, and the conflict contribution of demand management schedulable features to time occupancy is 0.2; while for capacity occupancy, the conflict contributions of these three types of features are 0.4, 0.4, and 0.2, respectively.

[0083] Based on the conflict contribution analysis results and the non-dominated solution that maximizes the economic optimization objective benefit, the weight coefficients of the time-shiftable schedulable feature vector, the ancillary service schedulable feature vector, and the demand management schedulable feature vector are adaptively adjusted. For example, if the time-shiftable feature contributes the most to conflict, its weight coefficient can be reduced from 0.4 to 0.3; conversely, if the demand management schedulable feature has a smaller contribution to conflict but a larger contribution to benefit, its weight coefficient can be increased from 0.3 to 0.4, while keeping the weight coefficient of the ancillary service schedulable feature unchanged at 0.3. This weight adjustment mechanism generates a new set of candidate solutions. Each candidate solution is generated based on the non-dominated solutions in the Pareto front set, with the weight coefficients adjusted according to the conflict contribution analysis results.

[0084] After generating the candidate solution set, it is compared with the current Pareto front set for non-domination. For each solution in the candidate solution set, its economic optimization objective's payoff value and risk constraint optimization objective's coupling conflict degree value are calculated, and then compared with all solutions in the Pareto front set for non-domination. If a candidate solution is not dominated by any solution in the Pareto front set, it is added to the Pareto front set; conversely, if any solution in the Pareto front set is dominated by a newly added candidate solution, these dominated solutions are removed from the Pareto front set. The Pareto front set is iteratively updated in this way.

[0085] After multiple iterations, the Pareto front set will converge to a relatively stable state, which contains a series of non-dominated solutions that achieve a good balance between economic optimization objectives and risk-constrained optimization objectives.

[0086] In one optional implementation, obtaining the optimal solution set for the economic optimization objective under the condition of satisfying the risk constraint optimization objective, and thus obtaining the business model combination scheme includes:

[0087] Calculate the total capacity utilization corresponding to each non-dominated solution in the Pareto front set, and identify the time period when the total capacity utilization exceeds the rated capacity of the energy storage power station as a capacity over-limit risk indicator; calculate the number of overlapping time periods in the time series dimension of the business model feature set, and use the number of overlapping time periods as a time series conflict risk indicator.

[0088] Determine whether the capacity over-limit risk index and the temporal conflict risk index corresponding to each non-dominated solution in the Pareto front set meet the threshold condition of the risk constraint optimization objective. From the Pareto front set that meets the threshold condition, select the non-dominated solution with the largest economic optimization objective benefit value, and take the weight coefficients of each schedulable feature vector corresponding to the non-dominated solution as the optimal solution set.

[0089] Based on the optimal solution set, the time allocation ratios of the power time-shifting business model, the ancillary service response business model, and the demand management business model are adjusted respectively within the planning period. The business model combination scheme is generated by combining the peak and valley time characteristics, periodic characteristics, and the matching relationship in the demand management schedulable feature vector.

[0090] Under the condition of satisfying the risk constraint optimization objective, the optimal solution set of the economic optimization objective is obtained. The total capacity occupancy corresponding to each non-dominated solution in the Pareto front set is calculated, and the time period when the total capacity occupancy exceeds the rated capacity of the energy storage power station is identified and used as the capacity over-limit risk indicator. In specific implementation, taking an energy storage power station with a capacity of 100MWh as an example, within a 24-hour planning cycle, the hourly capacity occupancy is calculated based on the non-dominated solution. When the calculated capacity occupancy is 105MWh during a certain period (such as from 12:00 to 14:00), which exceeds the rated capacity of 100MWh, this period is marked as a capacity over-limit risk period with an over-limit of 5MWh. By counting the number of over-limit periods in the entire planning cycle, the capacity over-limit risk indicator value of the non-dominated solution is obtained.

[0091] The number of overlapping time periods in the business model feature set along the temporal dimension is calculated and used as a temporal conflict risk indicator. In specific implementation, for each time period within the planning period, the number of different business models existing simultaneously is counted. For example, if both power time-shifting and ancillary service response business models exist simultaneously between 9:00 and 11:00, the business model overlap number for that time period is 2. If two or more business models overlap in 5 out of the 24 time periods throughout the day, the temporal conflict risk indicator value for this non-dominated solution is 5.

[0092] Determine whether the capacity overrun risk index and the temporal conflict risk index corresponding to each non-dominated solution in the Pareto front set meet the threshold conditions of the risk constraint optimization objective. Assuming that the capacity overrun risk threshold is set to 3 time periods and the temporal conflict risk threshold is set to 6 time periods, select a subset of non-dominated solutions that meet both threshold conditions. For example, from the original 10 non-dominated solutions, select 6 solutions that simultaneously satisfy the condition that the number of capacity overrun time periods is ≤3 and the number of temporal conflict time periods is ≤6.

[0093] From the Pareto frontier set that meets the threshold condition, the non-dominated solution with the largest economic optimization objective return value is selected. The weight coefficients of each schedulable feature vector corresponding to the non-dominated solution are taken as the optimal solution set. Taking the above six non-dominated solutions that meet the risk constraints as examples, assuming that their returns are 1000 yuan, 1200 yuan, 950 yuan, 1100 yuan, 1350 yuan and 1050 yuan respectively, the non-dominated solution with a return value of 1350 yuan is selected as the optimal solution. The weight coefficients of the three business models of power time shifting, ancillary service response and demand management corresponding to the optimal solution are [0.4, 0.3, 0.3], which represent the degree of preference for the three models in resource allocation.

[0094] Based on the obtained optimal solution set, the time allocation ratios of the power time-shifting business model, the ancillary service response business model, and the demand management business model are adjusted within the planning period. Combining the peak-valley time characteristics, periodic characteristics, and the matching relationship in the demand management schedulable feature vector, a business model combination scheme is generated. In specific implementation, according to the weight coefficients [0.4, 0.3, 0.3], the time periods are allocated within the 24-hour planning period as follows: the off-peak period from 0:00 to 6:00 and the peak period from 18:00 to 22:00 are mainly allocated to the power time-shifting business model, accounting for 40% of the total capacity; the daytime period from 10:00 to 16:00 is mainly allocated to the ancillary service response business model, accounting for 30% of the total capacity; and the morning peak period from 7:00 to 9:00 and the evening period from 22:00 to 24:00 are mainly allocated to the demand management business model, accounting for 30% of the total capacity.

[0095] In the actual allocation process, the characteristic matching degree of each time period is further considered. For example, for the electricity time-shifting business model, the peak and valley periods with large electricity price differences are given priority; for ancillary service response, the periods with high grid frequency regulation demand are considered; and for demand management, the periods with high user electricity load are considered. Through this matching, the final business model combination scheme not only meets the risk constraints, but also maximizes economic benefits.

[0096] In one optional implementation, based on the time-series allocation ratio of each business model in the business model combination scheme, and combined with the capacity status data of the energy storage power station, a scheduling execution instruction is generated and executed, including:

[0097] Acquire capacity status data of the energy storage power station, including the current state of charge value, available charging capacity, and available discharging capacity;

[0098] The planning period is divided into multiple scheduling time slots. The duration ratio of the power time-shifting business model, the ancillary service response business model, and the demand management business model in each scheduling time slot is determined, and a time-series resource allocation matrix is ​​generated. For each scheduling time slot, the charging interval, discharging interval, or standby interval of the energy storage power station is determined based on the current state of charge value, and the available charging capacity and the available discharging capacity are allocated to generate a capacity resource allocation matrix.

[0099] Based on the time-series resource allocation matrix and the capacity resource allocation matrix, the capacity share and duration allocated to the corresponding business model are calculated to obtain the charging and discharging power instruction value. Combined with the business model execution sequence of each scheduling time slot, the scheduling execution instruction is generated.

[0100] The scheduling execution command is sent to the power control unit of the energy storage power station, and the power control unit performs charging or discharging operations.

[0101] The dispatch system of an energy storage power station acquires the capacity status data of the power station, including the current state of charge (SOC), available charging capacity, and available discharging capacity. For example, a 100MWh energy storage power station has a current SOC of 50%, meaning it has 50MWh of stored energy, a maximum charging power of 20MW, a maximum discharging power of 25MW, an available charging capacity of 50MWh, and an available discharging capacity of 50MWh.

[0102] The dispatching system divides the planning cycle into multiple dispatch slots. Taking day-ahead dispatching as an example, the 24-hour planning cycle is divided into 96 dispatch slots of 15 minutes each. For each dispatch slot, the time allocation for the power time-shifting business model, the ancillary service response business model, and the demand management business model is determined based on historical data and market forecasts. For example, during peak electricity price periods (e.g., 10:00-15:00), 70% of the time is allocated to the power time-shifting model, 20% to the ancillary service response model, and 10% to the demand management model. During off-peak electricity price periods (e.g., 01:00-05:00), 85% of the time is allocated to the power time-shifting model, 15% to the ancillary service response model, and 0% to the demand management model.

[0103] By determining the duration percentage of each business model in each scheduling time slot, the scheduling system generates a time-series resource allocation matrix. This matrix has a dimension of 96×3, representing the duration percentage allocation of the three business models across the 96 time slots. For example, the element value of the 45th time slot (corresponding to 11:15-11:30) in the time-series resource allocation matrix is ​​[0.7, 0.2, 0.1], indicating that in this time slot, the power time-shifting mode accounts for 70% of the duration, the ancillary service response mode accounts for 20% of the duration, and the demand management mode accounts for 10% of the duration.

[0104] For each scheduling time slot, the scheduling system determines the operating range of the energy storage power station based on the current state of charge (SOC) value and sets SOC thresholds, such as SOC < 30% for the charging range, SOC > 70% for the discharging range, and 30% ≤ SOC ≤ 70% for the standby range. Based on different ranges, available charging capacity and available discharging capacity are allocated. In the charging range, more available charging capacity is allocated first; in the discharging range, more available discharging capacity is allocated first; and in the standby range, capacity resources are flexibly allocated based on time period characteristics and price signals.

[0105] By determining the capacity resource allocation for each scheduling time slot, the scheduling system generates a capacity resource allocation matrix with dimensions of 96×6, representing the charging and discharging capacities allocated to the three business models across the 96 time slots. For example, the element value of the 45th time slot in the capacity resource allocation matrix is ​​[30MWh, 10MWh, 5MWh, 35MWh, 10MWh, 5MWh], indicating that in this time slot, the power time-shifting mode allocates 30MWh of charging capacity and 35MWh of discharging capacity, the ancillary service response mode allocates 10MWh of charging capacity and 10MWh of discharging capacity, and the demand management mode allocates 5MWh of charging capacity and 5MWh of discharging capacity.

[0106] Based on the time-series resource allocation matrix and the capacity resource allocation matrix, the scheduling system calculates the capacity share and duration allocated to the corresponding business model to obtain the charging / discharging power command value. For each scheduling time slot, the allocated capacity is divided by the allocated duration to obtain the charging / discharging power. For example, for the power time-shifting mode in the 45th time slot, the discharge capacity is 35MWh, the duration share is 0.7, and the time slot length is 15 minutes. Then the charging / discharging power command value is 35÷(0.7×0.25)=200MW. However, since the maximum discharge power is limited to 25MW, the actual command value is 25MW.

[0107] In addition to calculating power command values, the scheduling system also needs to determine the execution sequence of business models within each scheduling time slot. The determination of the execution sequence is based on the priority and time urgency of the business models. Typically, demand management mode has the highest priority, followed by ancillary service response mode, and finally power time shift mode. In the 45th time slot, the execution sequence is: demand management (first 1.5 minutes) → ancillary service response (middle 3 minutes) → power time shift (last 10.5 minutes).

[0108] By combining the charging and discharging power command values ​​and the business model execution sequence, the scheduling system generates complete scheduling execution instructions. These instructions include the operating mode, power output, and duration for different time periods within each time slot. For example, the scheduling execution instructions for the 45th time slot are: 11:15-11:16:30, demand management mode, discharging power 20MW; 11:16:30-11:19:30, ancillary service response mode, discharging power 15MW; 11:19:30-11:30:00, power time-shifting mode, discharging power 25MW.

[0109] The dispatch system sends the generated dispatch execution instructions to the power control unit (PCU) of the energy storage power station. After receiving the instructions, the PCU controls the power conversion system (PCS) to perform charging or discharging operations according to the instructions. For example, between 11:15 and 11:16:30, the PCU controls the PCS to discharge from the battery to the grid at a power of 20MW; between 11:16:30 and 11:19:30, the PCU controls the PCS to discharge from the battery to the grid at a power of 15MW; and between 11:19:30 and 11:30:00, the PCU controls the PCS to discharge from the battery to the grid at a power of 25MW.

[0110] Using the above methods, energy storage power stations can generate and execute optimal scheduling instructions based on the time allocation ratio of different business models and the current capacity status, thereby achieving efficient utilization of energy storage resources and maximizing economic benefits.

[0111] In one optional implementation, identifying business models with overlapping timelines and overcapacity in the business model combination scheme, and establishing a timeline priority ranking by calculating the capacity status data usage and release time of each business model, including:

[0112] The overlap duration of actual execution periods between various business models is statistically analyzed, and combinations of business models with an overlap duration greater than zero are marked as time-overlapping business model combinations.

[0113] For each business model in the time-overlapping business model combination, based on the charging and discharging power command value corresponding to each business model and the duration of the actual execution period, the capacity occupancy requirement of each business model on the capacity status data is obtained and summed. If the summation result exceeds the available charging capacity or available discharging capacity of the energy storage power station, the time-overlapping business model combination is marked as a capacity-over-limit business model combination.

[0114] For each business model in the capacity over-limit business model combination, calculate the release time of the energy storage power station capacity for each business model. The release time is the end time of the actual execution period of each business model. Based on the release time, calculate the capacity occupancy duration of each business model.

[0115] Based on the capacity occupancy demand and the duration of capacity occupancy, a time-series priority ranking rule is established. The business model with the shortest duration of capacity occupancy and the smallest capacity occupancy demand is assigned the highest time-series priority and sorted in descending order to generate the time-series priority ranking.

[0116] Energy storage power stations can typically execute multiple business models simultaneously, such as peak-valley arbitrage, demand response, and frequency regulation ancillary services. Each business model has its specific execution period and charging / discharging power commands. When the execution periods of multiple business models overlap, and the total charging / discharging demand exceeds the capacity limit of the energy storage power station, these business models need to be prioritized to determine their execution order.

[0117] In practical applications, the overlap of actual execution periods between various business models is statistically analyzed. Assuming an energy storage power station simultaneously executes three business models A, B, and C, with model A executing from 8:00-10:00, model B from 9:00-11:00, and model C from 10:30-12:30, the time intersection between models is calculated. This reveals that the overlap between model A and model B is 9:00-10:00 (1 hour); the overlap between model B and model C is 10:30-11:00 (0.5 hours); and there is no time overlap between model A and model C. Therefore, combinations of model A and model B, and combinations of model B and model C, can be labeled as time-overlapping business model combinations.

[0118] For the identified overlapping business model combinations, further analysis is conducted on their capacity utilization of energy storage power stations. For example, during the overlapping period of 9:00-10:00 between Mode A and Mode B, assuming Mode A requires the energy storage power station to charge at a power of 2MW and Mode B requires to discharge at a power of 1.5MW, then during this overlapping period, the capacity utilization requirement of Mode A is 2MW × 1 hour = 2MWh of charging capacity, and the capacity utilization requirement of Mode B is 1.5MW × 1 hour = 1.5MWh of discharging capacity. If the available charging capacity of the energy storage power station is 1.8MWh, then the charging capacity requirement of the combination of Mode A and Mode B exceeds the available charging capacity of the energy storage power station. Therefore, this combination is marked as a capacity-over-limit business model combination.

[0119] The release time of the energy storage power station capacity for each business model is calculated, which is the end time of the actual execution period of each business model. In the example above, the release time of model A is 10:00, and the release time of model B is 11:00. Based on the release time, the duration of capacity occupation for each business model can be calculated. Calculated from the current time 9:00, the duration of capacity occupation for model A is 10:00 - 9:00 = 1 hour, and the duration of capacity occupation for model B is 11:00 - 9:00 = 2 hours.

[0120] Based on capacity utilization demand and capacity utilization duration, a time-series priority ranking rule is established. In this implementation, business models with shorter capacity utilization durations are given priority, followed by those with smaller capacity utilization demands. This allows for more efficient utilization of the energy storage power station's capacity resources. In the example above, Model A has a capacity utilization duration of 1 hour and a capacity utilization demand of 2 MWh; Model B has a capacity utilization duration of 2 hours and a capacity utilization demand of 1.5 MWh. Although Model B's capacity utilization demand is less than that of Model A, Model A has a shorter capacity utilization duration; therefore, Model A has a higher time-series priority than Model B.

[0121] Assuming the total capacity of the energy storage power station is 10MWh, the current State of Charge (SOC) is 50%, meaning the current electricity consumption is 5MWh, the available charging capacity is 5MWh, and the available discharging capacity is 5MWh. Four business models, D, E, F, and G, are running simultaneously. Model D operates from 14:00 to 16:00 with a charging power of 2MW; Model E operates from 15:00 to 17:00 with a discharging power of 3MW; Model F operates from 15:30 to 18:00 with a charging power of 1.5MW; and Model G operates from 16:30 to 19:00 with a discharging power of 2.5MW.

[0122] By calculating the time intersection, the following overlapping combinations can be determined: the overlapping period of pattern D and pattern E is 15:00-16:00; the overlapping period of pattern D and pattern F is 15:30-16:00; the overlapping period of pattern E and pattern F is 15:30-17:00; the overlapping period of pattern E and pattern G is 16:30-17:00; and the overlapping period of pattern F and pattern G is 16:30-18:00.

[0123] For these overlapping combinations, the capacity utilization demand is calculated. Taking the overlapping period of Mode E and Mode F (15:30-17:00) as an example, Mode E requires discharging at 3MW power for 1.5 hours, with a capacity utilization demand of 4.5MWh; Mode F requires charging at 1.5MW power for 1.5 hours, with a capacity utilization demand of 2.25MWh. Since the available discharge capacity of the energy storage station is 5MWh, which is greater than the demand of Mode E, and the available charging capacity is 5MWh, which is greater than the demand of Mode F, this combination does not belong to the capacity over-limit business model combination.

[0124] In one optional implementation, capacity allocation is adjusted for conflicting business models based on the time priority ranking, and the adjusted capacity allocation result is fed back to the schedulable feature vector corresponding to the business model feature set, thereby realizing dynamic collaboration between business models, including:

[0125] According to the time priority from high to low, the available charging capacity or available discharging capacity of the energy storage station is allocated to each business model in sequence, and the full amount is allocated according to the capacity occupancy demand of the business model; when the remaining capacity of the energy storage station is less than the capacity occupancy demand of the current business model, the remaining capacity is allocated to the current business model, and the capacity deficit of the current business model is calculated; and the adjusted capacity allocation value of each business model is generated.

[0126] The adjusted capacity allocation value is fed back to the business model feature set. Based on the updated capacity allocation parameters in the business model feature set, the actual total capacity occupancy of the time-shiftable schedulable feature vector, the ancillary service schedulable feature vector, and the demand management schedulable feature vector in the current scheduling time slot is recalculated. It is then determined whether the actual total capacity occupancy meets the capacity constraint conditions of the energy storage power station. For business models with capacity deficits, compensation instructions are generated and executed in the current scheduling time slot to compensate for the capacity deficits. The scheduling execution instructions are updated to achieve dynamic coordination between business models.

[0127] Obtain the capacity requirements of each business model in the current time slot. For example, in a certain scheduling time slot, the ancillary service model requires 15MW of discharge capacity, the demand management model requires 10MW of discharge capacity, and the time-shifting model requires 5MW of charging capacity. Assume the total capacity of the energy storage power station is 25MW, the currently available discharge capacity is 20MW, and the available charging capacity is 20MW.

[0128] The capacity allocation process proceeds sequentially from highest to lowest time priority. The ancillary service mode is allocated its required 15MW of discharge capacity, leaving a remaining 5MW of discharge capacity. Capacity is then allocated to the demand management mode. Since the demand management mode requires 10MW but only has 5MW remaining, all of the remaining 5MW is allocated to it, and its capacity deficit is calculated to be 5MW. For the time-shifted mode, since it requires charging capacity, and the current charging capacity is sufficient, all of its required 5MW of charging capacity is allocated to it.

[0129] After the initial allocation is completed, adjusted capacity allocation values ​​are generated: ancillary service mode receives 15MW of discharge capacity, demand management mode receives 5MW of discharge capacity, and time-shift mode receives 5MW of charging capacity. These allocation values ​​are fed back to the business model feature set to update the capacity allocation parameters for each mode.

[0130] Based on the updated parameters, the actual total capacity occupancy of each business model in the current scheduling time slot is recalculated. For example, in the discharge direction, the ancillary service mode occupies 15MW, the demand management mode occupies 5MW, totaling 20MW, which is equal to the available discharge capacity and meets the constraints. In the charging direction, the time-shift mode occupies 5MW, which is less than the available charging capacity of 20MW, and also meets the constraints.

[0131] For demand management modes with capacity deficits, compensation instructions are generated and executed in the current scheduling time slot. The compensation method can be to adjust the operating status of electrical equipment to reduce the load, purchase additional power from the grid, or appropriately reduce the requirements for demand management if permitted. For example, an instruction can be issued to temporarily reduce the non-critical load in the plant area by 5MW, or to call 5MW of power from the standby power generation equipment to compensate for the capacity deficit of the demand management mode.

[0132] Through a real-time compensation mechanism, even when capacity is insufficient, each business model can achieve its functional goals to a certain extent. At the same time, the scheduling execution instructions are updated according to the actual allocation results to ensure that the energy storage power station operates in accordance with the adjusted capacity allocation plan.

[0133] In another example, suppose that in a certain scheduling time slot, the ancillary service mode requires 8MW of charging capacity, the demand management mode requires 12MW of discharging capacity, and the time-shift mode requires 15MW of charging capacity. The total capacity of the energy storage station is still 25MW, the current available discharging capacity is 18MW, and the available charging capacity is 10MW.

[0134] An ancillary service mode is allocated 8MW of charging capacity, leaving 2MW of remaining charging capacity. Since demand management mode requires discharge capacity, which does not conflict with charging capacity, 12MW of discharge capacity is allocated to it, leaving 6MW of remaining discharge capacity. For time-shift mode, since it requires 15MW of charging capacity, but only 2MW is available, the remaining 2MW of charging capacity is allocated to time-shift mode, and the capacity deficit for time-shift mode is calculated to be 13MW.

[0135] After allocation, auxiliary services occupy 8MW in the charging direction, time-shift mode occupies 2MW, totaling 10MW, which is equal to the available charging capacity; demand management in the discharging direction occupies 12MW, which is less than the available discharging capacity of 18MW, both of which meet the constraints.

[0136] For time-shifted charging modes with a 13MW charging capacity shortfall, compensatory measures can be taken, such as partially postponing the time-shifted charging plan to the next time slot or reducing the charging amount within the allowed price range. At the same time, the dispatch execution instructions should be updated to ensure that the energy storage power station operates in accordance with the new capacity allocation scheme.

[0137] This time-priority-based capacity allocation mechanism, combined with a real-time compensation strategy, can effectively handle capacity conflicts when multiple business models are running simultaneously, ensuring the normal operation of high-priority models and providing resource support for low-priority models, thereby achieving dynamic collaboration between business models.

[0138] This invention relates to a collaborative optimization system for energy storage power station business models based on multi-objective programming, the system comprising:

[0139] The first unit is used to decouple the operation data of the energy storage power station in the spatiotemporal dimension to obtain the business model feature set. The business model feature set is combined with the economic optimization objective and the risk constraint optimization objective to establish a multi-objective programming constraint system.

[0140] The second unit is used to iteratively solve the multi-objective programming constraint system for non-dominated solution set based on the Pareto front search strategy. During the solution process, the weight coefficients of each schedulable feature vector in the business model feature set are dynamically adjusted according to the risk constraint optimization objective. Under the condition of satisfying the risk constraint optimization objective, the optimal solution set of the economic optimization objective is obtained, and the business model combination scheme is obtained.

[0141] The third unit is used to generate and execute scheduling execution instructions based on the time allocation ratio of each business model in the business model combination scheme and the capacity status data of the energy storage power station.

[0142] The fourth unit is used to identify business models with overlapping timelines and exceeding capacity limits in the business model combination scheme during instruction execution. By calculating the capacity status data usage and release time of each business model, a time priority ranking is established. Based on the time priority ranking, capacity allocation is adjusted for conflicting business models, and the adjusted capacity allocation result is fed back to the schedulable feature vector corresponding to the business model feature set to achieve dynamic collaboration between business models.

[0143] A third aspect of the present invention provides an electronic device, comprising:

[0144] processor;

[0145] Memory used to store processor-executable instructions;

[0146] The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0147] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0148] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0149] 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 them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A collaborative optimization method for energy storage power station business models based on multi-objective programming, characterized in that, include: The spatiotemporal dimensions of energy storage power station operation data are decoupled to obtain a business model feature set. This business model feature set is then combined with economic optimization objectives and risk-constrained optimization objectives to establish a multi-objective programming constraint system, including: The energy storage power station operation data includes charging and discharging power time-series data, state of charge time-series data, and load demand time-series data. The charging and discharging power time-series data is decomposed on a time scale to extract peak-valley period features and periodic features; Spatial state mapping is performed on the state-of-charge time-series data to identify state transition paths and extract capacity utilization depth features and state duration features corresponding to the state transition paths. Calculate the coupling relationship between the peak and valley time period characteristics and the capacity utilization depth characteristics to generate a time-shiftable schedulable feature vector; calculate the coupling relationship between the periodicity characteristics and the state duration characteristics to generate an auxiliary service schedulable feature vector; calculate the matching relationship between the load demand time series data and the state transition path to generate a demand management schedulable feature vector. The time-shiftable schedulable feature vector, the auxiliary service schedulable feature vector, and the demand management schedulable feature vector constitute the business model feature set; The economic optimization objective quantifies the revenue of each business model, and the risk constraint optimization objective quantifies the coupling conflict degree between business models. The economic optimization objective and the risk constraint optimization objective are used as dual optimization objectives, and the business model feature set is used as decision variables to establish the multi-objective programming constraint system. The multi-objective programming constraint system is solved iteratively using a Pareto front search strategy to obtain a non-dominated solution set. During the solution process, the weight coefficients of each schedulable feature vector in the business model feature set are dynamically adjusted according to the risk constraint optimization objective. Under the condition of satisfying the risk constraint optimization objective, the optimal solution set of the economic optimization objective is obtained, and the business model combination scheme is obtained. Based on the time allocation ratio of each business model in the aforementioned business model combination scheme, and combined with the capacity status data of the energy storage power station, a scheduling execution instruction is generated and executed. During instruction execution, business models with overlapping timelines and exceeding capacity limits in the business model combination scheme are identified. By calculating the capacity status data usage and release time of each business model, a timeline priority ranking is established. Based on the timeline priority ranking, capacity allocation is adjusted for conflicting business models, and the adjusted capacity allocation results are fed back to the schedulable feature vector corresponding to the business model feature set, thereby realizing dynamic collaboration between business models.

2. The method according to claim 1, characterized in that, The multi-objective programming constraint system is solved iteratively using a Pareto front search strategy to obtain a non-dominated solution set. During the solution process, the weight coefficients of each schedulable feature vector in the business model feature set are dynamically adjusted according to the risk constraint optimization objective, including: An initial solution set is generated within the decision space corresponding to the business model feature set; Calculate the coupling conflict degree between the profit value of the economic optimization objective and the risk constraint optimization objective, construct a dual-objective fitness space, identify the non-dominated solutions of the initial solution set in the dual-objective fitness space and store them in the Pareto front set; Based on the coupling conflict degree value corresponding to each non-dominated solution in the Pareto front set, the conflict contribution degree of the business model feature set in terms of time-series occupancy and capacity occupancy is calculated. According to the conflict contribution degree and the non-dominated solution with the largest economic optimization objective benefit value, the weight coefficients between the time-shift schedulable feature vector, the auxiliary service schedulable feature vector and the demand management schedulable feature vector are adaptively adjusted to generate a candidate solution set. The candidate solution set is compared with the Pareto front set using non-dominated relations, and the Pareto front set is iteratively updated based on the result of the non-dominated relations comparison.

3. The method according to claim 1, characterized in that, Under the condition of satisfying the risk constraint optimization objective, the optimal solution set of the economic optimization objective is obtained, resulting in business model combination schemes including: Calculate the total capacity utilization corresponding to each nondominated solution in the Pareto front set, and identify the time period when the total capacity utilization exceeds the rated capacity of the energy storage power station as a capacity over-limit risk indicator. Calculate the number of overlapping time periods in the time sequence dimension of the business model feature set, and use the number of overlapping time periods as a time sequence conflict risk indicator; Determine whether the capacity over-limit risk index and the temporal conflict risk index corresponding to each non-dominated solution in the Pareto front set meet the threshold condition of the risk constraint optimization objective. From the Pareto front set that meets the threshold condition, select the non-dominated solution with the largest economic optimization objective benefit value, and take the weight coefficients of each schedulable feature vector corresponding to the non-dominated solution as the optimal solution set. Based on the optimal solution set, the time allocation ratios of the power time-shifting business model, the ancillary service response business model, and the demand management business model are adjusted respectively within the planning period. The business model combination scheme is generated by combining the peak and valley time characteristics, periodic characteristics, and the matching relationship in the demand management schedulable feature vector.

4. The method according to claim 1, characterized in that, Based on the time-series allocation ratio of each business model in the aforementioned business model combination scheme, and combined with the capacity status data of the energy storage power station, a scheduling execution instruction is generated and executed, including: Acquire capacity status data of the energy storage power station, including the current state of charge value, available charging capacity, and available discharging capacity; The planning period is divided into multiple scheduling time slots. The duration ratio of the power allocation time-shifting business model, the ancillary service response business model, and the demand management business model in each scheduling time slot is determined, and a time-series resource allocation matrix is ​​generated. For each scheduling time slot, the charging interval, discharging interval, or standby interval of the energy storage power station is determined based on the current state of charge value, and the available charging capacity and the available discharging capacity are allocated to generate a capacity resource allocation matrix; Based on the time-series resource allocation matrix and the capacity resource allocation matrix, the capacity share and duration allocated to the corresponding business model are calculated to obtain the charging and discharging power instruction value. Combined with the business model execution sequence of each scheduling time slot, the scheduling execution instruction is generated. The scheduling execution command is sent to the power control unit of the energy storage power station, and the power control unit performs charging or discharging operations.

5. The method according to claim 1, characterized in that, Identify business models with overlapping timelines and capacity limitations in the aforementioned business model combination schemes. Establish a timeline priority ranking by calculating the capacity status data usage and release time for each business model, including: The overlap duration of actual execution periods between various business models is statistically analyzed, and combinations of business models with an overlap duration greater than zero are marked as time-overlapping business model combinations. For each business model in the time-overlapping business model combination, based on the charging and discharging power command value corresponding to each business model and the duration of the actual execution period, the capacity occupancy requirement of each business model on the capacity status data is obtained and summed. If the summation result exceeds the available charging capacity or available discharging capacity of the energy storage power station, the time-overlapping business model combination is marked as a capacity-over-limit business model combination. For each business model in the capacity over-limit business model combination, calculate the release time of the energy storage power station capacity for each business model. The release time is the end time of the actual execution period of each business model. Based on the release time, calculate the capacity occupancy duration of each business model. Based on the capacity occupancy demand and the duration of capacity occupancy, a time-series priority ranking rule is established. The business model with the shortest duration of capacity occupancy and the smallest capacity occupancy demand is assigned the highest time-series priority and sorted in descending order to generate the time-series priority ranking.

6. The method according to claim 1, characterized in that, Based on the time-series priority ranking, capacity allocation is adjusted for conflicting business models, and the adjusted capacity allocation results are fed back to the schedulable feature vector corresponding to the business model feature set, realizing dynamic collaboration between business models, including: According to the time priority from high to low, the available charging capacity or available discharging capacity of the energy storage power station is allocated to each business model in turn, and the full amount is allocated according to the capacity occupancy requirements of the business model. When the remaining capacity of the energy storage power station is less than the capacity requirement of the current business model, all the remaining capacity is allocated to the current business model, and the capacity deficit of the current business model is calculated. Generate adjusted capacity allocation values ​​for each business model; The adjusted capacity allocation value is fed back to the business model feature set. Based on the updated capacity allocation parameters in the business model feature set, the actual total capacity occupancy of the time-shift schedulable feature vector, the ancillary service schedulable feature vector, and the demand management schedulable feature vector in the current scheduling time slot is recalculated. It is then determined whether the actual total capacity occupancy meets the capacity constraint conditions of the energy storage power station. A compensation instruction is generated for the business model with the aforementioned capacity deficit and executed in the current scheduling slot to compensate for the capacity deficit; Update the scheduling execution instructions to achieve dynamic collaboration between business models.

7. A collaborative optimization system for energy storage power station business models based on multi-objective programming, used to implement the method as described in any one of claims 1-6, characterized in that, include: The first unit is used to decouple the operation data of the energy storage power station in the spatiotemporal dimension to obtain the business model feature set. The business model feature set is combined with the economic optimization objective and the risk constraint optimization objective to establish a multi-objective programming constraint system. The second unit is used to iteratively solve the multi-objective programming constraint system for non-dominated solution set based on the Pareto front search strategy. During the solution process, the weight coefficients of each schedulable feature vector in the business model feature set are dynamically adjusted according to the risk constraint optimization objective. Under the condition of satisfying the risk constraint optimization objective, the optimal solution set of the economic optimization objective is obtained, and the business model combination scheme is obtained. The third unit is used to generate and execute scheduling execution instructions based on the time allocation ratio of each business model in the business model combination scheme and the capacity status data of the energy storage power station. The fourth unit is used to identify business models with overlapping timelines and exceeding capacity limits in the business model combination scheme during instruction execution. By calculating the capacity status data usage and release time of each business model, a time priority ranking is established. Based on the time priority ranking, capacity allocation is adjusted for conflicting business models, and the adjusted capacity allocation result is fed back to the schedulable feature vector corresponding to the business model feature set to achieve dynamic collaboration between business models.

8. An electronic device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by the processor, they implement the method described in any one of claims 1 to 6.

Images