Independent energy storage multi-market coordinated optimization method and system

Abstract

A method and system for multi-market collaborative optimization of independent energy storage in the electricity market is proposed. This method acquires the statistical distribution characteristics, peak-valley structure characteristics, and extreme event characteristics of daily historical electricity price sequences, and clusters them to obtain typical days. Each typical day is divided into normal days and risk days. At the bottom layer, based on the operational data of each energy storage unit on risk days, the constraints of intraday decision parameters are adjusted to meet the constraints of safe energy storage operation. At the top layer, using the operational data of each energy storage unit on normal days, and aiming at the optimal market economic objective, the baseline values ​​of intraday decision parameters for each energy storage unit are determined under the constraints of the adjusted intraday decision parameters. At the bottom layer, based on the operational data of each energy storage unit on normal days and the baseline values ​​of intraday decision parameters, the baseline values ​​of intraday decision parameters for each energy storage unit are optimized to maximize energy storage operating efficiency and minimize energy loss, thereby obtaining a profit-maximizing control strategy under multi-market collaboration to achieve executable monthly and annual scheduling for the project.

Description

Technical Field

[0001] This invention belongs to the field of power system collaborative operation technology, specifically, it relates to a method and system for multi-market collaborative optimization of independent energy storage under the power market. Background Technology

[0002] The large-scale integration of new energy sources, such as wind and solar power, into the power grid presents significant challenges to the real-time balance of the power system due to their inherent intermittency and volatility, leading to temporal fluctuations in electricity prices in the spot market. The high proportion of new energy grid integration and the accelerated implementation of the spot market exacerbate these temporal price fluctuations. While independent energy storage possesses the potential for diversified revenue from energy, frequency regulation, and capacity leasing, its actual operation faces challenges such as capacity mutual exclusion and frequency regulation activation / recharge coupling, capacity leasing lock-in and availability constraints, inconsistencies between revenue and overall cost calculations, and the large computational burden and poor executability of full-time optimization. Therefore, an annual optimization method that balances profitability and feasibility is urgently needed. Against this backdrop, independent energy storage, as a flexible regulatory resource, needs to be developed into an efficient optimized operation strategy to maximize its resource regulation capabilities.

[0003] In existing technologies, a sequential decision-making logic first determines the type of market to participate in and then optimizes the energy storage operation strategy. However, this logic highlights the inability to achieve fully integrated global optimization within a single model. In terms of full life-cycle economic assessment, operating costs are usually simplified for ease of calculation. The optimization method for energy storage participation in the electricity market uses an economic model that linearly correlates operating costs with the charging and discharging capacity of energy storage. This method effectively assesses costs at the macro level. To make the optimization results more closely reflect actual needs, a more refined and nonlinear cost model that reflects the differentiated damage to battery life under different operating conditions should be constructed. Using heuristic algorithms to optimize the configuration, operation, and pricing clearing of energy storage provides a solution for such large-scale optimization problems. However, it is still necessary to further ensure the global optimality of the results, handle large-scale complex constraints, and improve the solution speed. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method and system for multi-market collaborative optimization of independent energy storage in the electricity market. It solves the problems of insufficient profit and unfeasible solutions caused by factors such as the lack of consideration for price fluctuations, capacity exclusivity among independent energy storage units within the market, frequency regulation activation and recharge coupling, capacity leasing lock-in and availability constraints, and inconsistencies between revenue and overall cost, making it difficult to achieve full-time collaborative optimization. By constructing intraday decision parameters and a two-layer optimization framework under a unified monthly schedule, a profit-maximizing control strategy under multi-market collaboration is obtained, enabling executable monthly and annual scheduling.

[0005] The present invention adopts the following technical solution.

[0006] This invention proposes a multi-market collaborative optimization method for independent energy storage in the electricity market, comprising:

[0007] The statistical distribution characteristics, peak-valley structure characteristics, and extreme event characteristics of the daily historical electricity price series are obtained, and typical days are obtained by clustering. Based on the extreme event characteristics, each typical day is divided into normal days and risk days.

[0008] Establish energy revenue models and ancillary service revenue models for energy storage;

[0009] Determine the intraday decision parameters for each energy storage unit and establish constraints for these parameters.

[0010] A total degradation cost model is established based on the lifespan degradation model of energy storage.

[0011] A two-layer optimization framework for energy storage operation strategies is established. At the bottom layer, based on the operational data of each energy storage unit on risk days, the constraints of intraday decision parameters are adjusted to meet the constraints of safe energy storage operation. At the top layer, using the operational data of each energy storage unit on normal days, and aiming at the optimal market economic objective, the baseline values ​​of intraday decision parameters for each energy storage unit are determined under the constraints of the adjusted intraday decision parameters. At the bottom layer, based on the operational data of each energy storage unit on normal days and the baseline values ​​of intraday decision parameters, the baseline values ​​of intraday decision parameters for each energy storage unit are optimized to maximize energy storage operating efficiency and minimize energy loss, resulting in a multi-market collaborative control strategy for energy storage.

[0012] Statistical distribution characteristics, including: the mean and standard deviation of the daily historical electricity price series;

[0013] Peak-valley structure characteristics include: peak-valley price difference, time of highest electricity price, and time of lowest electricity price;

[0014] Extreme event characteristics include: the 95th and 5th percentiles of the electricity price series;

[0015] Statistical distribution characteristics, peak-valley structure characteristics, and extreme event characteristics constitute a multidimensional feature vector.

[0016] The K-means clustering method based on feature weighted distance is used to cluster the multidimensional feature vectors of all days in a natural month, with the centroid as the typical day;

[0017] The feature-weighted distance is shown in the following formula:

[0018]

[0019] In the formula, Weighted distance for features For the first Multidimensional feature vectors of historical days For the first Cluster The centroid vector is typical of daily data; The matrix is ​​a diagonal weight matrix. In the matrix, the weights corresponding to the peak-valley price difference in the peak-valley structure feature and the 95th quantile of the electricity price sequence in the extreme event feature are all greater than 1, while the weights corresponding to other features are 1.

[0020] Obtain energy storage operation data on normal days to determine the charging and discharging amount of energy storage to meet the net power of the grid connection point under constraints, and determine the frequency regulation mileage and energy storage reserve capacity based on the actual output of energy storage under automatic generation control.

[0021] An energy revenue model is established based on the charging and discharging quantities of energy storage.

[0022] A revenue model for ancillary services is established based on frequency regulation mileage and energy storage reserve capacity.

[0023] Energy gain As shown in the following formula:

[0024]

[0025] In the formula, For time period Electricity sales settlement price, For time period The electricity purchase settlement price; For time period The discharge amount is positive; For time period The charging amount is negative.

[0026] The tracking amount of the AGC command relative to the baseline is used as the frequency modulation mileage, as shown in the following formula:

[0027]

[0028] In the formula, For time period Frequency modulation mileage; For time period The power of AGC instructions Each sample value, For time period The baseline power of the first Each sample value, For the longest time period Each sample value.

[0029] The revenue from ancillary services is shown in the following formula:

[0030]

[0031] Among them, frequency modulation revenue As shown in the following formula:

[0032]

[0033] In the formula, For time period FM mileage fee, For time period Performance coefficient;

[0034] Capacity billing As shown in the following formula:

[0035]

[0036] In the formula, For time period The spare capacity; For time period Spare capacity fee; For time period Availability coefficient.

[0037] Intraday decision-making parameters include: charging and discharging time window and power limit, frequency regulation reserved capacity and frequency regulation mileage occupancy sequence, power allocation for participation in the electric energy market and SOC trajectory;

[0038] The constraints on intraday decision parameters include: energy time-shift constraints, ancillary service capacity constraints, and resource mutual exclusion constraints.

[0039] Based on the lifetime degradation model characterizing the physical degradation properties of batteries, a nonlinear cost curve for energy storage is generated, as shown in the following equation:

[0040]

[0041] In the formula, This is the cyclic degradation cost coefficient. The equivalent full cycle number, Total degradation cost;

[0042] The nonlinear cost curve is fitted with multiple continuous straight lines, and the total degradation cost is determined by the following relationship:

[0043] , ,

[0044] In the formula, For the first The fitted power flow rate within each segmented interval For the common The cumulative electricity consumption in each segment interval The number of segmented intervals, For the first The unit degradation cost coefficient for each segmented interval, To transmit power.

[0045] The optimal market economy objective is shown in the following formula:

[0046]

[0047] In the formula, For the optimal goal of a market economy, , , The first Energy revenue, ancillary service revenue, and total degradation cost of energy storage; The amount of energy stored within the market.

[0048] The optimized intraday decision-making parameter baseline values ​​for each energy storage system are used as the intraday execution plan; the same optimized intraday decision-making parameter baseline values ​​are reused within a calendar month to obtain a unified monthly schedule.

[0049] This invention also proposes an independent energy storage multi-market collaborative optimization system under the electricity market, comprising:

[0050] The data processing module is used to obtain the statistical distribution characteristics, peak-valley structure characteristics, and extreme event characteristics of the daily historical electricity price series, and to cluster them to obtain typical days; based on the extreme event characteristics, each typical day is divided into normal days and risk days;

[0051] The data preparation module is used to establish the energy revenue model and ancillary service revenue model for energy storage; determine the intraday decision parameters for each energy storage unit and establish the constraints for the intraday decision parameters; and establish the total degradation cost model based on the lifespan degradation model of energy storage.

[0052] The collaborative optimization module is used to establish a two-layer optimization framework for energy storage operation strategies. At the bottom layer, based on the operational data of each energy storage unit on risk days, the constraints of intraday decision parameters are adjusted to meet the constraints of safe energy storage operation. At the top layer, using the operational data of each energy storage unit on normal days, the baseline values ​​of intraday decision parameters for each energy storage unit are determined under the constraints of adjusted intraday decision parameters, with the goal of optimizing market economic objectives. At the bottom layer, based on the operational data of each energy storage unit on normal days and the baseline values ​​of intraday decision parameters, the baseline values ​​of intraday decision parameters for each energy storage unit are optimized to maximize energy storage operating efficiency and minimize energy loss, resulting in a multi-market collaborative control strategy for energy storage.

[0053] The present invention is also a terminal, including a processor and a storage medium; the storage medium is used to store instructions; the processor is used to perform operations according to the instructions to execute the steps of the method.

[0054] The present invention is also a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method.

[0055] The beneficial effects of this invention are as follows, compared with the prior art, at least including the following: This invention is based on a power market energy storage economic optimization method coupled with a monthly unified schedule and a two-layer optimization algorithm. It generates typical days for each month using historical price and quantity data. The top layer determines the charging and discharging window and market capacity allocation structure, while the bottom layer performs a fine solution under constraints such as charging and discharging power, mutual exclusion, frequency regulation activation and recharge, capacity leasing lock-in / availability. This invention achieves multi-market collaborative optimization of energy storage with the goal of maximizing profits, significantly reduces the scale of daily full-time-series solutions, improves computational efficiency and solution stability, effectively avoids capacity conflicts and ensures energy closure and leasing compliance, and improves the robustness of returns and energy storage utilization under price fluctuations and assessment uncertainties. Attached Figure Description

[0056] Figure 1 This is a flowchart of the independent energy storage multi-market collaborative optimization method proposed in this invention. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this invention. The embodiments described in this application are merely some embodiments of this invention, and not all embodiments. Based on the spirit of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of this invention.

[0058] This invention proposes a multi-market collaborative optimization method for independent energy storage in the electricity market, such as... Figure 1 As shown, the method includes:

[0059] Step 1: Obtain the statistical distribution characteristics, peak-valley structure characteristics, and extreme event characteristics of the daily historical electricity price series, and cluster them to obtain typical days; based on the extreme event characteristics, divide each typical day into normal days and risk days.

[0060] Specifically, step 1 includes:

[0061] Step 1.1: Using machine learning methods, extract multi-dimensional features from the daily historical electricity price series to form a multi-dimensional feature vector;

[0062] In this embodiment, the one-dimensional electricity price time series of natural days (24-hour and above resolution) is transformed into a multi-dimensional feature vector that can be recognized by machine learning. Three types of key features are extracted for each historical day within a natural month, including:

[0063] 1) Statistical distribution characteristics, including the mean and standard deviation of daily historical electricity price series, to characterize the overall level and volatility of electricity prices;

[0064] 2) Peak-valley structure characteristics, including: peak-valley price difference, time of highest electricity price, and time of lowest electricity price; in the example, the highest and lowest prices within the day are extracted and the peak-valley price difference is calculated, and the time of extreme electricity price is recorded to characterize the time distribution of the intraday electricity price pattern;

[0065] 3) Extreme event characteristics, including the 95th and 5th percentiles of the electricity price series, to quantify the characteristics of tail-end electricity price events;

[0066] Each natural day within a calendar month corresponds to a unique structured multidimensional feature vector containing the four key features mentioned above. To reduce the computational burden of the time dimension in annual operational optimization, this invention uses machine learning algorithms to perform cluster analysis on historical electricity price and load sequences, generating a monthly typical day dataset for the electricity market, thus achieving data dimensionality reduction. This invention introduces a weighted clustering algorithm in the typical day generation stage, embedding peak-valley electricity price difference weights and extreme event difference weights into the K-means clustering objective function. This ensures that typical days retain extreme electricity price characteristics. This weighting mechanism has a technical impact on the representativeness of the model input and the robustness of subsequent optimization results.

[0067] Step 1.2: The K-means clustering method based on feature weighted distance is used to cluster the multidimensional feature vectors of all days in a natural month, with the centroid as the typical day;

[0068] Specifically, statistical distribution characteristics, peak-valley structure characteristics, and extreme event characteristics constitute a multi-dimensional feature vector; the feature vector set of all days in a natural month is used as input, the number of clusters is set, and the cluster centroid is calculated iteratively through the K-means clustering method; the typical day is the centroid, and the electricity price data of the typical day is the centroid vector, which represents the core pattern of monthly electricity price fluctuations;

[0069] This invention achieves a key information retention mechanism by performing K-means clustering on the multidimensional feature vectors of all days within a natural month. The peak-valley structure features within the cluster are explicitly retained as centroids in the form of average values, directly reflecting the monthly average statistical level. High electricity prices or low voltages that occur frequently within a natural month will raise or lower the electricity prices for the corresponding time period on typical days, thus achieving the statistical integration of extreme event features.

[0070] To ensure that the generated typical days retain extreme price information crucial to the economics of energy storage, this invention employs a K-means clustering method based on feature-weighted distance. This method modifies the distance in the standard K-means algorithm to assign higher statistical weights to specific dimensions in the feature vectors. The feature-weighted distance is shown in the following formula:

[0071]

[0072] In the formula, Weighted distance for features For the first Multidimensional feature vectors of historical days For the first Cluster The centroid vector is typical of daily data; The matrix is ​​a diagonal weight matrix. In the matrix, the weights corresponding to the peak-valley price difference in the peak-valley structure feature and the 95th quantile of the electricity price series in the extreme event feature are all greater than 1, while the weights corresponding to other features are 1.

[0073] By setting a weight value greater than 1, this invention amplifies the impact of the differences in the aforementioned key features on the objective function when calculating the weighted distance. This makes it possible for the clustering process to prioritize ensuring that samples within the cluster are closer in these high-weight features when iteratively updating the centroids (i.e., the "averaging process"). As a result, the final typical daily data can more realistically retain the peak-valley structure features and extreme event features in the original data.

[0074] Typical days obtained through clustering are a condensation of the core features of statistical distribution, suitable for formulating standardized operational strategies and understanding the main behavioral patterns of the system. After feature-weighted distance correction, they can also cover data for risk assessment or stress testing under extreme conditions. Data obtained by conventional distance cannot assess the performance of the system under extreme conditions, which may lead to an underestimation of risk. Clustering typical days is the basis for analyzing and designing peak shaving and valley filling strategies. When optimizing energy storage, the peak-valley structure based on typical days can reliably determine the charging and discharging time windows.

[0075] Step 1.3: Based on the characteristics of extreme events, each typical day is divided into normal days and risk days.

[0076] Step 2: Establish the energy revenue model and ancillary service revenue model for energy storage.

[0077] Specifically, step 2 includes:

[0078] Step 2.1: Obtain energy storage operation data for normal days to determine the charging and discharging amount of energy storage to meet the net power of the grid connection point under constraints, and determine the frequency regulation mileage and energy storage reserve capacity based on the actual output of energy storage under automatic generation control.

[0079] The net power at the grid connection point is broken down into charge and discharge quantities, as shown in the following formula:

[0080] ,

[0081] In the formula, For time period The discharge quantity is positive (unit: MW·min); For time period The charging amount is a negative value (unit: MW·min). For time period Net power at grid connection point (unit: MW), during discharge When charging, the value is positive. It is a negative value; The duration of the time period (unit: min);

[0082] The net power constraint at the grid connection point is shown in the following formula:

[0083]

[0084] In the formula, Maximum power limit for grid-connected power purchase (unit: MW, negative). Maximum power limit for grid-connected electricity sales (unit: MW, positive);

[0085] The tracking amount of the AGC command relative to the baseline is used as the frequency modulation mileage, as shown in the following formula:

[0086]

[0087] In the formula, For time period Frequency regulation mileage (unit: MW·min); For time period The power of AGC instructions Sample values ​​(unit: MW) For time period The baseline power of the first Sample values ​​(unit: MW) For the longest time period Each sample value (unit: min);

[0088] Time period The sum of all sampling points is the frequency modulation mileage.

[0089] Step 2.2: Establish an energy revenue model based on the charging and discharging of energy storage.

[0090] Energy gain As shown in the following formula:

[0091]

[0092] In the formula, For time period Electricity sales settlement price (unit: yuan / MWh) For time period Electricity purchase settlement price (unit: yuan / MWh).

[0093] Step 2.3: Based on the frequency regulation mileage and energy storage reserve capacity, establish an ancillary service revenue model;

[0094] The revenue from ancillary services is shown in the following formula:

[0095]

[0096] Among them, frequency modulation revenue As shown in the following formula:

[0097]

[0098] In the formula, For time period Frequency regulation mileage fee (unit: yuan / MW). For time period The performance coefficient has a range of (0,2).

[0099] Capacity billing As shown in the following formula:

[0100]

[0101] In the formula, For time period Reserve capacity (unit: MW); For time period Standby capacity fee (unit: yuan / MWh); For time period The availability coefficient, with a value range of 100%. .

[0102] Based on existing energy storage operation models, this invention proposes a revenue-cost integrated modeling method that collaboratively models the energy market, ancillary services market, and equipment physical characteristics.

[0103] Step 3: Determine the intraday decision parameters for each energy storage unit, and reuse the same intraday decision parameters within a natural month.

[0104] Based on the core physical behavior and objective trading rules of energy storage systems in different markets, the collaborative optimization strategy is parameterized to determine intraday decision parameters, including:

[0105] For time-shift arbitrage scenarios in the power market, the hourly start-stop decision parameters of the power market are parameterized as charging and discharging time windows and power limits, with the corresponding decision variables being the start time of the charging and discharging cycle, the end time, and the average power during the window period.

[0106] To address the demand for regulation capacity reservation for ancillary services such as frequency modulation, the day is divided into several key periods based on the differences in service demand during peak, valley, flat, and ramp-up periods. The time-segmented reservation decision parameters of the ancillary service market are parameterized as frequency modulation reserved capacity and frequency modulation mileage occupancy time sequence, with the corresponding decision variables being the unified reserved capacity level for each key period.

[0107] To avoid problems such as overcharging and over-discharging, cross-day and cross-month scheduling mathematical interfaces, and time-dimensional logical inconsistencies, the energy storage operation control strategy is parameterized as the power allocation and SOC trajectory for participating in the electric energy market, and decision variables such as daily cycle benchmark SOC level and optional daily net energy transfer amount are set.

[0108] Intraday decision parameters include: charging and discharging time window and power limit, frequency regulation reserved capacity and frequency regulation mileage occupancy sequence, power allocation and SOC trajectory for participating in the electricity market; by agreeing to reuse the same set of decision parameters every day within the month to reduce the optimization dimension, and the optimized intraday decision parameters are used as the final output of the independent energy storage multi-market collaborative optimization strategy under the electricity market;

[0109] This invention proposes a modeling method for decision parameterization on typical monthly days, which transforms the high-dimensional discrete hourly scheduling decision problem into a low-dimensional key strategy parameter optimization problem. This method improves solution efficiency while ensuring the practical guidance of the results, and constructs a mathematical intermediate layer that links the underlying physical operations with the top-level economic strategy.

[0110] Step 4: Establish constraints for intraday decision parameters.

[0111] Specifically, within the mixed-integer linear programming model, a series of deterministic mathematical constraints are established to transform the intraday decision parameters of the top-level output into the boundary conditions of the hourly running variables of the bottom level, including:

[0112] 1) Energy time shift constraint: Establish mathematical constraints so that within the "charging window" defined by the top-level parameters, the sum and average value of the charging power at each moment do not exceed the power upper limit of the window; similarly, within the "discharging window", the discharge power at each moment is also constrained by the corresponding parameters.

[0113] 2) Ancillary service capacity constraints: Establish mathematical constraints to ensure that, within each key time period defined by the top-level parameters, the power capacity reserved by the energy storage system for ancillary services such as frequency regulation is not lower than the reserved capacity parameter value for that time period.

[0114] 3) Resource mutual exclusion constraint: Establish logical constraints to ensure that at any given time, the power of energy storage is uniquely and non-conflictingly allocated among charging, discharging and providing ancillary services, and the total of these allocations does not exceed the rated power of the energy storage system.

[0115] This invention decomposes typical intraday core operational behaviors of energy storage into strategy modules with clear physical meaning; each module is configured with continuous / integer decision parameters to replace time-by-time binary and continuous variables; deterministic constraints are established between intraday decision parameters and underlying time-by-time operational variables within a mixed-integer linear programming model to achieve structured dimensionality reduction of the decision space.

[0116] Step 5: Based on the energy storage lifetime degradation model, establish a total degradation cost model.

[0117] Specifically, based on a lifetime degradation model characterizing the physical degradation of batteries, a nonlinear cost curve is generated to describe the relationship between the cumulative throughput of an energy storage system and its cumulative degradation cost. This nonlinear cost curve is fitted with multiple continuous straight lines to obtain the unit degradation cost coefficient within each segment interval. In the MILP model, a new set of decision variables and linear constraints are introduced to express the cumulative degradation cost as a linear weighted sum of the throughput of each segment and its corresponding cost coefficient. Through this technique, the complex nonlinear degradation cost is successfully transformed into a linear expression that can be simultaneously optimized within the MILP framework, thereby avoiding the bias caused by traditional a posteriori correction methods.

[0118] Energy storage parameter settings: Define the feasible domain and SOC evolution equation for energy storage operation, which are used for optimization solutions and plan verification to ensure physical feasibility; support upper limits of charge and discharge power, SOC boundary, efficiency and ramping constraints.

[0119] The power boundary is:

[0120] ,

[0121] In the formula, , Time periods Charging power and discharging power; , Time periods The upper limit of charging power and the upper limit of discharging power;

[0122] The dynamics of SOC (including efficiency, self-consumption, and self-discharge) and its boundaries are as follows:

[0123]

[0124]

[0125] In the formula, For time period The state of charge, , These are the charging and discharging efficiencies, respectively. Rated energy, Self-discharge rate , These are the lower and upper limits of the state of charge, respectively;

[0126] Power ramp-up and response are as follows:

[0127]

[0128] In the formula, For time period Net power at grid connection point Power ramp rate (unit: MW / h);

[0129] Based on annual equivalent full-charge cycles, the battery degradation cost is quickly estimated. The impact of temperature on energy storage life is comprehensively considered, and the revenue model is incorporated into the revenue optimization model.

[0130] The nonlinear cost curve model is as follows: In the formula, The cost coefficient for cyclic degradation (unit: yuan / cycle). The equivalent full cycle number, Total degradation cost;

[0131] ,

[0132] In the formula, To transmit power;

[0133] Meanwhile, this invention approximates the nonlinear characteristics of battery degradation cost using a linear segment, thereby incorporating energy storage degradation cost into the MILP objective and constraints. This enables degradation-runtime simultaneous optimization within a linear framework, and the total degradation cost is determined through lifetime curve fitting using the following relationship:

[0134] , ,

[0135] In the formula, For the first The fitted power flow rate within each segmented interval For the common The cumulative electricity consumption in each segment interval The number of segmented intervals, For the first Unit degradation cost coefficient for each segmented interval;

[0136] To achieve simultaneous optimization of revenue and degradation costs, this invention employs piecewise linearization technology, embedding the energy storage lifetime degradation model into the MILP revenue solution process in a linear approximation form, thus avoiding the bias caused by traditional posterior correction.

[0137] Step 6: Establish a two-layer optimization framework for energy storage operation strategy. At the bottom layer, based on the operational data of each energy storage unit on risk days, and with the goal of meeting the constraints of safe energy storage operation, adjust the constraints of intraday decision parameters. At the top layer, using the operational data of each energy storage unit on normal days, and with the goal of optimizing market economic objectives, determine the baseline values ​​of intraday decision parameters for each energy storage unit under the constraints of adjusted intraday decision parameters. At the bottom layer, based on the operational data of each energy storage unit on normal days and the baseline values ​​of intraday decision parameters, and with the goal of maximizing energy storage operating efficiency and minimizing energy loss, optimize the baseline values ​​of intraday decision parameters for each energy storage unit to obtain a multi-market collaborative control strategy for energy storage.

[0138] Specifically, step 6 includes:

[0139] Step 6.1: In the underlying layer of the mixed integer linear programming (MILP) model, based on the operating data of each energy storage on risk days, and with the goal of meeting the energy storage safety operation constraints, adjust the constraints of the intraday decision parameters.

[0140] Risk days are days characterized by extreme electricity price events. By using the extreme events of risk days as optimization conditions for intraday decision-making parameter constraints, and on the basis of ensuring the safe operation of energy storage, the energy storage collaborative control strategy is transformed from a single cost constraint into a market-oriented proactive adjustment strategy.

[0141] In this embodiment, high electricity price days present significant profit opportunities but also carry extremely high operational risks. Charging at maximum power before peak electricity prices may lead to excessively high State of Charge (SOC), while continuous discharge at maximum power during peak electricity prices may cause battery overheating, power exceeding limits, or even damage due to over-discharge. Although the profit from a single transaction is high, it causes irreversible damage to the battery, sacrificing countless future profit opportunities. Therefore, based on historical operational data from high electricity price days, the underlying MILP model learns which SOC and power combinations result in the fastest battery degradation or the temperature rise closest to the safety threshold. Furthermore, on high electricity price days, the underlying MILP model proactively tightens the constraints on intraday decision parameters, such as dynamically reducing the SOC upper limit from 95% to 85%. This may seem to limit the charging amount, but it ensures a healthier and more usable capacity for discharge during peak electricity prices, avoiding BMS protective shutdowns triggered by excessively high SOC; or it may shorten the maximum continuous discharge time from 1 hour to 45 minutes. This may seem to limit the discharge capacity, but it prevents the battery from overheating and derating due to prolonged high-power discharge, ensuring stable power output throughout the high-price window.

[0142] In this embodiment, based on historical data, the model can identify the pressure on safety from different operating modes on high frequency regulation price days or high energy price days. On high energy price days, the underlying MILP model moderately tightens the reserved capacity constraint for frequency regulation services, allowing more capacity to be used for energy arbitrage, because the marginal benefit of energy arbitrage is much higher than that of frequency regulation at this time. Conversely, on high frequency regulation price days, the underlying MILP model relaxes the penalty coefficient for frequent charging and discharging through the objective function, because the revenue from frequency regulation services is sufficient to cover the resulting battery loss costs. At the same time, it strictly enforces constraints to prevent SOC exceeding limits, ensuring that the battery is within an absolutely safe hard boundary under any market strategy.

[0143] In step 6.1, by using the operational data model of the risk day to train the model, the output of the model can dynamically adjust its tolerance for safe operation according to the market business risks represented by price signals, thereby achieving a comprehensive balance and maximization of returns for cross-market business risks and technical risks; at the same time, it can also serve as a feedforward optimization for subsequent steps in multiple iterations, realizing economic optimization under the premise of avoiding risks in advance.

[0144] Step 6.2, at the top level, using the operating data of each energy storage unit on normal days, with the goal of optimizing the market economy objective, and under the constraint of adjusting the intraday decision parameters, determine the benchmark value of the intraday decision parameters for each energy storage unit.

[0145] The optimal market economy objective is shown in the following formula:

[0146]

[0147] In the formula, For the optimal goal of a market economy, , , The first Energy revenue, ancillary service revenue, and total degradation cost of energy storage; The amount of energy stored within the market;

[0148] In this embodiment, the heuristic top layer uses normal day data obtained through clustering to generate baseline values ​​for intraday decision parameters based on risk avoidance, thus forming a baseline scheduling strategy. The top layer uses the "monthly unified schedule" parameter as the baseline value for intraday decision parameters and utilizes a heuristic search algorithm to perform global optimization under the constraints of intraday decision parameters. Its output includes, but is not limited to: the division of charging and discharging periods for typical normal days in each month, the allocation of power upper limits for each period, the frequency modulation capacity reservation ratio, and the SOC baseline setting.

[0149] Step 6.2 of this invention performs macro-strategy planning based on aggregated data of all energy storage, which reduces the complexity of the underlying optimization, while ensuring the coordination and consistency of the control strategies of all energy storage and meeting the global objectives. The obtained benchmark value is a robust benchmark control strategy learned from historical data of normal days, which takes into account both benefits and controllable risks. Furthermore, when determining the benchmark value of intraday decision parameters, the top layer also performs differentiated charging and discharging states and power allocation for each energy storage.

[0150] Step 6.3, at the bottom layer, based on the operating data of each energy storage on a normal day and the baseline values ​​of intraday decision parameters, with the goal of maximizing energy storage operating efficiency and minimizing energy loss, the baseline values ​​of intraday decision parameters of each energy storage are optimized to obtain the multi-market collaborative control strategy for energy storage.

[0151] Since the response performance of different energy storage systems varies, the final strategy output at the bottom layer is a control scheme that is the most economical and least energy-wasting for each execution path under a given control benchmark. This ensures that the final strategy is not only correct in terms of business model, but also optimal in terms of physical execution.

[0152] In this embodiment, the bottom layer uses the parameters output by the top layer as constraints to construct a refined energy storage physical operation model. This model takes the energy balance equation, the SOC dynamic equation and the power boundary constraint as its core, and solves the time-by-time charging and discharging power sequence and SOC trajectory to ensure that the energy closure, power constraints and safe operation conditions of the energy storage device are all met. The objective function adopts a dual-objective weighted form of maximizing comprehensive operating efficiency and minimizing energy loss, and obtains the optimal solution for energy allocation under the premise of ensuring that the constraints are met.

[0153] At the top level, using data from risk days as the test set, the baseline values ​​of the optimized intraday decision parameters for each energy storage system are evaluated based on feedback from the bottom level. The parameter search direction of the top level is dynamically updated based on the evaluation. This invention also uses the strategy variables (charge and discharge window, frequency regulation ratio, etc.) output by the top-level decision layer as input to the bottom-level model, and the running results returned by the bottom level (such as energy utilization rate, SOC curve smoothness, computational cost, etc.) as evaluation feedback to dynamically update the search direction. In particular, the data from risk days can be used as a test dataset for the reliability and robustness of the control strategy, thereby realizing strategy correction based on evaluation feedback. Through multiple iterations, the system achieves collaborative optimization from macro-level strategy to micro-level operating trajectory. The iteration termination condition is the convergence of the evaluation function or the number of iterations reaching a preset upper limit. The solution architecture proposed in this invention improves the computational feasibility of large-scale optimization problems and realizes dynamic feedback between strategy-level and control-level variables.

[0154] This invention employs a two-layer optimization framework to solve the operation strategy of energy storage systems; this framework separates long-term scheduling structure decisions from short-term operation trajectory optimization in order to balance computational efficiency and model accuracy.

[0155] Step 7: The optimized intraday decision parameter baseline values ​​for each energy storage system are used as the intraday execution plan; the same optimized intraday decision parameter baseline values ​​are reused within a calendar month to obtain a unified monthly schedule.

[0156] The final output is the annual energy storage charging and discharging schedule; based on historical data input, it solves the problem and outputs a monthly unified schedule and daily executable plans, providing monthly and annual revenue / cost breakdowns and profit indicators; the monthly unified schedule execution plan is used to support the operational planning of independent energy storage to achieve optimal market returns. The annual revenue / cost breakdown analyzes the proportion of different costs on a monthly and annual basis, facilitating decision-makers to further adjust the revenue model.

[0157] This invention proposes a collaborative optimization method and system for independent energy storage in a multi-market electricity market. The method constructs a parameterized dimensionality reduction model with a "monthly unified schedule" and a two-layer optimization framework. At the top layer, a heuristic algorithm is used for macro-level strategy optimization, while at the bottom layer, mixed-integer linear programming (MILP) is used for refined solution. This scheme achieves accurate modeling of complex constraints across multiple markets within a unified framework and considers total cost, aiming to ensure computational efficiency while outputting an annual operation plan that balances high returns and high feasibility. Addressing the technical challenges of unified modeling of multiple constraints, high computational load, and low solution efficiency in the collaborative operation of independent energy storage in multiple markets, this invention proposes a collaborative optimization method and system based on parameterized dimensionality reduction and two-layer optimization. This achieves accurate modeling and efficient solution of complex constraints across multiple markets, exhibiting good engineering feasibility and computational stability.

[0158] A two-layer optimization method for monthly unified schedule in independent energy storage under spot market multi-market collaborative optimization is proposed. It generates representative days for dimensionality reduction modeling on a monthly basis, determines the schedule structure through top-level heuristics, and performs fine-grained calculation of power and SOC trajectory at the bottom layer, while unifying the measurement of revenue and cost.

[0159] This invention also proposes an independent energy storage multi-market collaborative optimization system under the electricity market, comprising:

[0160] The data processing module is used to obtain the statistical distribution characteristics, peak-valley structure characteristics, and extreme event characteristics of the daily historical electricity price series, and to cluster them to obtain typical days; based on the extreme event characteristics, each typical day is divided into normal days and risk days;

[0161] The data preparation module is used to establish the energy revenue model and ancillary service revenue model for energy storage; determine the intraday decision parameters for each energy storage unit and establish the constraints for the intraday decision parameters; and establish the total degradation cost model based on the lifespan degradation model of energy storage.

[0162] The collaborative optimization module is used to establish a two-layer optimization framework for energy storage operation strategies. At the bottom layer, based on the operational data of each energy storage unit on risk days, the constraints of intraday decision parameters are adjusted to meet the constraints of safe energy storage operation. At the top layer, using the operational data of each energy storage unit on normal days, the baseline values ​​of intraday decision parameters for each energy storage unit are determined under the constraints of adjusted intraday decision parameters, with the goal of optimizing market economic objectives. At the bottom layer, based on the operational data of each energy storage unit on normal days and the baseline values ​​of intraday decision parameters, the baseline values ​​of intraday decision parameters for each energy storage unit are optimized to maximize energy storage operating efficiency and minimize energy loss, resulting in a multi-market collaborative control strategy for energy storage.

[0163] This disclosure can be a system, method, and / or computer program product. A computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for causing a processor to implement various aspects of this disclosure.

[0164] Computer-readable storage media can be tangible devices capable of holding and storing instructions for use by an instruction execution device. Computer-readable storage media can be, for example—but not limited to—electrical storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of computer-readable storage media include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital multifunction disc (DVD), memory sticks, floppy disks, mechanical encoding devices, such as punch cards or recessed protrusions storing instructions thereon, and any suitable combination of the foregoing. The computer-readable storage media used herein are not to be construed as transient signals themselves, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., light pulses through fiber optic cables), or electrical signals transmitted through wires.

[0165] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0166] Computer program instructions used to perform the operations of this disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, status setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may execute entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing the status information of the computer-readable program instructions to implement various aspects of this disclosure.

[0167] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A method for multi-market collaborative optimization of independent energy storage in an electricity market, characterized in that, include: Obtain the statistical distribution characteristics, peak-valley structure characteristics, and extreme event characteristics of daily historical electricity price sequences, and cluster them to obtain typical days; Based on the characteristics of extreme events, each typical day is divided into normal days and risk days; Establish energy revenue models and ancillary service revenue models for energy storage; Determine the intraday decision parameters for each energy storage unit and establish constraints for these parameters. A total degradation cost model is established based on the lifespan degradation model of energy storage. Establish a two-layer optimization framework for energy storage operation strategies; At the underlying level, based on the operational data of each energy storage facility on risk days, the constraints of intraday decision parameters are adjusted with the goal of meeting the constraints of safe operation of energy storage. At the top level, using the operational data of each energy storage unit on normal days, and with the goal of optimizing the market economy objective, the baseline values ​​of the intraday decision parameters for each energy storage unit are determined under the constraints of intraday decision parameter adjustments. At the bottom level, based on the operational data of each energy storage unit on normal days and the baseline values ​​of the intraday decision parameters, and with the goal of maximizing energy storage operating efficiency and minimizing energy loss, the baseline values ​​of the intraday decision parameters for each energy storage unit are optimized to obtain a multi-market collaborative control strategy for energy storage.

2. The method for multi-market collaborative optimization of independent energy storage under the electricity market according to claim 1, characterized in that, Statistical distribution characteristics, including: the mean and standard deviation of the daily historical electricity price series; Peak-valley structure characteristics include: peak-valley price difference, time of highest electricity price, and time of lowest electricity price; Extreme event characteristics include: the 95th and 5th percentiles of the electricity price series; Statistical distribution characteristics, peak-valley structure characteristics, and extreme event characteristics constitute a multidimensional feature vector.

3. The method for multi-market collaborative optimization of independent energy storage under the electricity market according to claim 2, characterized in that, The K-means clustering method based on feature weighted distance is used to cluster the multidimensional feature vectors of all days in a natural month, with the centroid as the typical day; The feature-weighted distance is shown in the following formula: In the formula, Weighted distance for features For the first Multidimensional feature vectors of historical days For the first Cluster The centroid vector is typical of daily data; The matrix is ​​a diagonal weight matrix. In the matrix, the weights corresponding to the peak-valley price difference in the peak-valley structure feature and the 95th quantile of the electricity price sequence in the extreme event feature are all greater than 1, while the weights corresponding to other features are 1.

4. The method for multi-market collaborative optimization of independent energy storage in the electricity market according to claim 1, characterized in that, Obtain energy storage operation data on normal days to determine the charging and discharging amount of energy storage to meet the net power of the grid connection point under constraints, and determine the frequency regulation mileage and energy storage reserve capacity based on the actual output of energy storage under automatic generation control. An energy revenue model is established based on the charging and discharging quantities of energy storage. A revenue model for ancillary services is established based on frequency regulation mileage and energy storage reserve capacity.

5. The method for multi-market collaborative optimization of independent energy storage in the electricity market according to claim 4, characterized in that, Energy gain As shown in the following formula: In the formula, For time period Electricity sales settlement price, For time period The electricity purchase settlement price; For time period The discharge amount is positive; For time period The charging amount is negative.

6. The method for multi-market collaborative optimization of independent energy storage in the electricity market according to claim 5, characterized in that, The tracking amount of the AGC command relative to the baseline is used as the frequency modulation mileage, as shown in the following formula: In the formula, For time period Frequency modulation mileage; For time period The power of AGC instructions Each sample value, For time period The baseline power of the first Each sample value, For the longest time period Each sample value; The revenue from ancillary services is shown in the following formula: Among them, frequency modulation revenue As shown in the following formula: In the formula, For time period FM mileage fee, For time period Performance coefficient; Capacity billing As shown in the following formula: In the formula, For time period The spare capacity; For time period Spare capacity fee; For time period Availability coefficient.

7. The method for multi-market collaborative optimization of independent energy storage in the electricity market according to claim 1, characterized in that, Intraday decision-making parameters include: charging and discharging time windows and power limits, frequency regulation reserved capacity and frequency regulation mileage occupancy sequence, power allocation for participation in the electricity market and SOC trajectory; The constraints on intraday decision parameters include: energy time-shift constraints, ancillary service capacity constraints, and resource mutual exclusion constraints.

8. The method for multi-market collaborative optimization of independent energy storage in the electricity market according to claim 1, characterized in that, Based on the lifetime degradation model characterizing the physical degradation properties of batteries, a nonlinear cost curve for energy storage is generated, as shown in the following equation: In the formula, This is the cyclic degradation cost coefficient. The equivalent full cycle number, Total degradation cost; The nonlinear cost curve is fitted with multiple continuous straight lines, and the total degradation cost is determined by the following relationship: ， ， In the formula, For the first The fitted power consumption within each segmented interval For the common The cumulative electricity consumption in each segment interval The number of segmented intervals, For the first The unit degradation cost coefficient for each segmented interval, To transmit power.

9. The method for multi-market collaborative optimization of independent energy storage in the electricity market according to claim 1, characterized in that, The optimal goal of a market economy is shown in the following formula: In the formula, For the optimal goal of a market economy, , , The first Energy revenue, ancillary service revenue, and total degradation cost of energy storage; The amount of energy stored within the market.

10. The method for multi-market collaborative optimization of independent energy storage in the electricity market according to claim 1, characterized in that, The method also includes: The optimized intraday decision-making parameter baseline values ​​for each energy storage system are used as the intraday execution plan; the same optimized intraday decision-making parameter baseline values ​​are reused within a calendar month to obtain a unified monthly schedule.

11. A multi-market collaborative optimization system for independent energy storage in an electricity market, used to implement the method described in any one of claims 1 to 10, characterized in that, include: The data processing module is used to obtain the statistical distribution characteristics, peak-valley structure characteristics, and extreme event characteristics of daily historical electricity price sequences, and to cluster them to obtain typical days; Based on the characteristics of extreme events, each typical day is divided into normal days and risk days; The data preparation module is used to establish the energy revenue model and ancillary service revenue model for energy storage; determine the intraday decision parameters for each energy storage unit and establish the constraints for the intraday decision parameters; and establish the total degradation cost model based on the lifespan degradation model of energy storage. The collaborative optimization module is used to establish a two-layer optimization framework for energy storage operation strategies. In the bottom layer, based on the operation data of each energy storage on risk days, the constraints of intraday decision parameters are adjusted with the goal of meeting the energy storage safety operation constraints. At the top level, using the operational data of each energy storage unit on normal days, and with the goal of optimizing the market economy objective, the baseline values ​​of the intraday decision parameters for each energy storage unit are determined under the constraints of intraday decision parameter adjustments. At the bottom level, based on the operational data of each energy storage unit on normal days and the baseline values ​​of the intraday decision parameters, and with the goal of maximizing energy storage operating efficiency and minimizing energy loss, the baseline values ​​of the intraday decision parameters for each energy storage unit are optimized to obtain a multi-market collaborative control strategy for energy storage.

12. A terminal, comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the method according to any one of claims 1-10.

13. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method according to any one of claims 1-10.

Images