A method and system for evaluating energy storage operation considering price life risk

Abstract

The present application relates to the technical field of energy storage system operation evaluation, and particularly relates to a kind of energy storage operation evaluation method and system considering tariff life risk, method includes: based on the tariff historical data to establish seasonal time series model, generate multiple sets of tariff scenario sequence;Establish the nonlinear mapping model of energy storage charge-discharge power and life degradation cost to calculate the life degradation cost caused by charge-discharge behavior;Solve independent optimization sub-problems for each tariff scenario sequence, output charge-discharge strategy, and obtain corresponding net income value;Aggregate the net income value of all tariff scenario sequences to generate income distribution, estimate the probability density function of fitting income distribution;Calculate conditional value at risk CVaR index;According to CVaR index and income distribution, generate operation risk evaluation report.Through the present application, the problem that the interaction influence of quantifying tariff fluctuation and battery life attenuation in extreme scenario is not coupled in the existing energy storage operation evaluation method, which leads to inaccurate evaluation is effectively solved.

Description

Technical Field

[0001] This invention relates to the field of energy storage system operation assessment technology, and in particular to an energy storage operation assessment method and system that takes into account electricity price lifetime risk. Background Technology

[0002] New energy storage technologies are a key strategic deployment for achieving energy transition, enhancing grid flexibility, and supporting power system security. Especially under the "dual carbon" target, their economic value in providing ancillary services such as peak shaving and frequency regulation through participation in electricity market energy arbitrage is increasingly prominent. However, the high proportion of renewable energy grid connection in the future will exacerbate random fluctuations in electricity prices, while energy storage systems face significant battery life degradation issues due to frequent charging and discharging. The interaction between these two factors directly affects the accuracy of the economic assessment of energy storage throughout its entire life cycle.

[0003] Current energy storage operation optimization methods have significant limitations in dealing with uncertainties. First, there are inherent flaws in modeling methods: mainstream uncertainty modeling techniques (such as probabilistic models, fuzzy models, and interval models) struggle to balance accuracy and practicality. Probabilistic models rely on complete probability distribution assumptions, which are difficult to accurately fit in real-world scenarios with complex and volatile electricity price fluctuations; fuzzy models require complex membership functions, resulting in excessively high data thresholds; while interval models simplify the process, they completely lose the characteristics of the probability distribution, leading to a distorted representation of the risk transmission mechanism. Second, there is a deficiency in decoupling core variables: existing methods either treat the coupling relationship between electricity price fluctuations and battery lifespan degradation in isolation, simplifying the impact of lifespan degradation only for short-term gains, or mechanically superimpose the two assessments after independent evaluation, failing to truly reflect the accelerating effect of charging and discharging behavior during peak electricity price periods on battery aging. Third, there is a lack of risk assessment dimensions: existing assessment systems often rely on single risk indicators such as expected returns, lacking the ability to quantitatively analyze multiple risks in extreme scenarios, such as a sudden drop in electricity prices coupled with a sudden drop in capacity. This leads to overly optimistic estimates in operational decisions, undermining the foundation for full life-cycle economic optimization. The aforementioned shortcomings collectively lead to a dilemma in energy storage economics assessment: oversimplification will amplify operational risks, while pursuing refined modeling will lead to excessively complex calculations.

[0004] The information disclosed in this background section is intended only to enhance the understanding of the general background of this disclosure and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0005] This invention provides a method and system for evaluating the operation of energy storage that takes into account the risk of electricity price lifetime, which can effectively solve the problems in the background art.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for assessing the operation of energy storage that takes into account electricity price lifetime risk, the method comprising: A seasonal time series model is established based on historical electricity price data to predict the estimated electricity price points at various times in the coming days, and multiple sets of electricity price scenario sequences are generated through Monte Carlo simulation. For each set of electricity price scenarios, a nonlinear mapping model is established between the energy storage charging and discharging power and the lifetime degradation cost. The nonlinear mapping model dynamically calculates the lifetime degradation cost caused by the charging and discharging behavior based on the relationship curve between the depth of discharge and cycle life of the energy storage battery. For each set of electricity price scenario sequences, solve an independent optimization subproblem with the objective of maximizing the peak-valley arbitrage profit minus the lifetime degradation cost. Under the constraints of charge-discharge state and battery capacity, output a matching charge-discharge strategy and obtain the corresponding net profit value. The net revenue values ​​of all the electricity price scenario sequences are aggregated to generate a revenue distribution, and the probability density function of the revenue distribution is fitted according to the kernel density estimate; The Conditional Value at Risk (CVaR) index is calculated based on the probability density function. The CVaR index represents the expected tail loss of the net income distribution under extremely unfavorable electricity price scenarios. An operational risk assessment report is generated based on the CVaR index and the statistical characteristics of the revenue distribution.

[0007] Furthermore, a seasonal time series model is established and an electricity price scenario series is generated, including: The historical electricity price data is fitted and predicted using a seasonal autoregressive integral moving average model to obtain the predicted electricity price sequence for future operating cycles. The electricity price prediction error is characterized based on the electricity price prediction sequence, and the electricity price prediction error is treated as a source of electricity price uncertainty. Based on the Monte Carlo simulation, random perturbation terms are generated by randomly sampling the electricity price prediction error, and the random perturbation terms are combined with the electricity price prediction sequence to generate multiple sets of electricity price scenario sequences; Multiple sets of the aforementioned electricity price scenario sequences are used as electricity price input data for the operation scheduling optimization model of the electrochemical energy storage system to characterize the electricity price uncertainty in the operating environment of the electrochemical energy storage system.

[0008] Furthermore, a seasonal autoregressive integral moving average model is used to fit and predict historical electricity price data, including: Seasonal pattern recognition is performed on the historical electricity price data, and trend changes are processed to make the historical electricity price data fit the fitting requirements of the seasonal autoregressive integral moving average model. The temporal correlation of the historical electricity price data is characterized based on non-seasonal and seasonal autoregressive terms to generate structural constraints on the electricity price prediction sequence. The correlation between the non-seasonal moving average term and the seasonal moving average term in the electricity price forecast error is characterized to improve the fitting ability of the electricity price forecast sequence to random fluctuations. The trend and seasonal trend components in the historical electricity price data are removed by differential processing, and the model parameters are fitted and determined to output the electricity price prediction sequence for generating multiple sets of electricity price scenario sequences in the Monte Carlo simulation.

[0009] Furthermore, multiple sets of the aforementioned electricity price scenario sequences are generated using the Monte Carlo simulation method, including: After obtaining the electricity price prediction sequence output by the seasonal autoregressive integral moving average model, the electricity price prediction error is sorted out to form an error characterization result that can be used for random sampling. Random sampling is performed on the error characterization results to obtain a random disturbance term that matches the electricity price prediction sequence, so that the random disturbance term reflects the random fluctuation characteristics of electricity prices; The random disturbance term is superimposed on the electricity price prediction sequence to form the electricity price scenario sequence, and the differences between different electricity price scenario sequences are characterized by the random sampling. The electricity price scenario sequence is collected and used as the electricity price input condition for the independent optimization sub-problem, so as to realize the modeling and expression of the electricity price uncertainty during the operation of the electrochemical energy storage system.

[0010] Furthermore, a nonlinear mapping model is established, including: An operation scheduling optimization model for an electrochemical energy storage system is established with the goal of maximizing net operating revenue within the operating cycle, and the transaction revenue generated from electricity market transactions is included in the revenue item of the operation scheduling optimization model. The equipment aging cost value characterized by the lifespan degradation cost is included in the cost item of the operation scheduling optimization model, and the lifespan degradation cost is made to correspond to the charging and discharging process to reflect the impact of lifespan degradation. In the operation scheduling optimization model, charging power constraints and discharging power constraints are set to limit the feasible range of charging and discharging power, and charging state evolution constraints are set to describe energy changes. The operation scheduling optimization model sets upper and lower limits of charge state and continuity constraints of charge state to ensure the feasibility of charge state and the consistency of scheduling cycle, and outputs a charging and discharging strategy that matches the electricity price scenario.

[0011] Furthermore, the cost of the said lifespan degradation includes: A battery degradation model based on the depth of discharge cycle is introduced into the operation scheduling optimization model, and a depth of discharge estimation algorithm is used to identify the charge and discharge cycle process of the electrochemical energy storage system. The total depth of discharge within the operating cycle is calculated based on the charge-discharge cycle process, and the total depth of discharge is allocated to the depth of discharge corresponding to multiple charge-discharge cycles to generate a set of depths of discharge for degradation assessment. The capacity loss corresponding to the set of discharge depths is estimated based on the relationship curve between the discharge depth and the number of cycles, and the total capacity decay for the day is obtained by summing up the capacity loss of each charge and discharge cycle. The total daily capacity decay is combined with the initial cost of the electrochemical energy storage system to convert it into an equivalent degradation cost, which is then written into the cost item of the operation scheduling optimization model as the lifetime degradation cost, so that the net operating revenue simultaneously reflects the transaction revenue and the equipment aging cost.

[0012] Furthermore, solving the independent optimization subproblems includes: Based on multiple sets of the electricity price scenario sequences, the operation scheduling optimization model is divided into multiple scheduling sub-problems according to the electricity price scenario, and each scheduling sub-problem is associated with the corresponding electricity price scenario. Maintain consistency in the charging power constraints, discharging power constraints, state of charge evolution constraints, upper and lower limits of state of charge constraints, state of charge continuity constraints, and battery degradation model across multiple scheduling sub-problems. The scheduling subproblems are solved independently using mathematical programming methods combined with mixed-integer linear programming to obtain the matching charging and discharging strategies, operational benefits, and battery degradation results for each of the electricity price scenarios. The solutions to the scheduling subproblems are aggregated to form a set of operating benefits and a set of battery degradation costs. The set of operating benefits and the set of battery degradation costs are then used for statistical analysis of the distribution of operating results and conditional risk value assessment.

[0013] Furthermore, the profit distribution and the fitted probability density function are generated, including: Statistical processing is performed on the set of operating revenues and the set of battery degradation costs to obtain the revenue distribution characteristics, and the set of operating revenues is sorted to form an ordered sequence of operating revenues for quantile representation. Based on the set of operating returns, a statistical measure is calculated to describe the central trend and volatility, and combined with the quantile characterization to reflect extreme characteristics, generating basic descriptive information of the distribution of operating results; The kernel density estimation method is used to perform nonparametric density estimation on the set of operating returns to form a continuous probability density distribution function, and the overall shape of the distribution of operating results and the tail risk characteristics are observed. Based on the distribution of the operating results, the conditional value of risk index is used to quantify the tail loss corresponding to the extreme unfavorable electricity price scenario and generate an evaluation result of the operating risk level. Combined with the changes in conditional value of risk under different confidence levels and the tail risk characteristics, a quantitative decision-making basis is provided for the selection of scheduling strategies for electrochemical energy storage systems.

[0014] An energy storage operation assessment system that takes into account electricity price lifetime risk, the system comprising: The scenario construction module establishes a seasonal time series model based on historical electricity price data, predicts the estimated electricity price points for various times in the next few days, and generates multiple sets of electricity price scenario sequences through Monte Carlo simulation. The lifetime determination module establishes a nonlinear mapping model between energy storage charging and discharging power and lifetime degradation cost for each set of electricity price scenario sequences. The nonlinear mapping model dynamically calculates the lifetime degradation cost caused by charging and discharging behavior based on the relationship curve between the depth of discharge and cycle life of the energy storage battery. The strategy matching module solves an independent optimization sub-problem for each set of electricity price scenario sequences. The objective is to maximize the peak-valley arbitrage revenue minus the lifetime degradation cost. Under the constraints of charge-discharge state and battery capacity, it outputs the matched charge-discharge strategy and obtains the corresponding net revenue value. The distribution fitting module aggregates the net revenue values ​​of all electricity price scenario sequences to generate a revenue distribution, and estimates the probability density function of the fitted revenue distribution based on kernel density. The indicator calculation module calculates the Conditional Value at Risk (CVaR) indicator based on the probability density function. The CVaR indicator represents the expected tail loss of the net income distribution under extremely unfavorable electricity price scenarios. The risk assessment module generates an operational risk assessment report based on the statistical characteristics of the CVaR index and the return distribution.

[0015] Furthermore, the scene construction module includes: The cycle prediction unit uses a seasonal autoregressive integral moving average model to fit historical electricity price data and perform prediction calculations to obtain the electricity price prediction sequence for future operating cycles. The error characterization unit characterizes the electricity price prediction error based on the electricity price prediction sequence and treats the electricity price prediction error as a source of electricity price uncertainty. The random perturbation unit generates random perturbation terms by randomly sampling the electricity price prediction error based on Monte Carlo simulation, and combines the random perturbation terms with the electricity price prediction sequence to generate multiple sets of electricity price scenario sequences; The scenario determination unit uses multiple sets of electricity price scenario sequences as electricity price input data for the operation scheduling optimization model of the electrochemical energy storage system to characterize the electricity price uncertainty in the operating environment of the electrochemical energy storage system.

[0016] The technical solution of this invention can achieve the following technical effects: Based on historical electricity price data, multiple Monte Carlo random scenario sequences are generated. Under each set of electricity price sequences, the charging and discharging strategy is optimized through a dynamic mapping model of discharge depth and lifespan loss. The scenario revenue distribution is aggregated and a probability density function is fitted. The tail risk is quantified using the CVaR index to achieve an integrated assessment of extreme scenarios and economic degradation. This effectively solves the problem of distorted operational decisions caused by the failure to simultaneously integrate the coupled effects of random electricity price fluctuations and battery lifespan loss in existing energy storage operation assessment methods, as well as the lack of quantification of multiple risk transmission mechanisms under extreme scenarios.

[0017] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, specific embodiments of this application are given below. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 A flowchart illustrating an energy storage operation assessment method that takes into account electricity price lifetime risk; Figure 2 A flowchart illustrating the process of generating an electricity price scenario sequence; Figure 3 A flowchart illustrating the process of establishing a nonlinear mapping model; Figure 4 A flowchart illustrating the process of solving independent optimization subproblems; Figure 5 A schematic diagram illustrating the process of generating the payoff distribution and fitting the probability density function. Detailed Implementation

[0020] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0022] Example 1; like Figure 1 As shown, this application provides a method for assessing the operation of energy storage that takes into account electricity price lifetime risk. The method includes: S10: Establish a seasonal time series model based on historical electricity price data to predict the estimated electricity price points at various times in the next few days, and generate multiple sets of electricity price scenario sequences through Monte Carlo simulation; S20: For each set of electricity price scenarios, establish a nonlinear mapping model between energy storage charging and discharging power and lifetime degradation cost. The nonlinear mapping model dynamically calculates the lifetime degradation cost caused by charging and discharging behavior based on the relationship curve between the depth of discharge and cycle life of the energy storage battery. S30: Solve an independent optimization subproblem for each set of electricity price scenarios, with the objective of maximizing the peak-valley arbitrage profit minus the lifetime degradation cost. Output a matching charge-discharge strategy under the constraints of charge-discharge state and battery capacity, and obtain the corresponding net profit value. S40: Aggregate the net revenue values ​​of all electricity price scenario sequences to generate a revenue distribution, and estimate and fit the probability density function of the revenue distribution based on kernel density. S50: Calculate the Conditional Value at Risk (CVaR) index based on the probability density function. The CVaR index represents the expected tail loss of the net income distribution under extremely unfavorable electricity price scenarios. S60: Generate an operational risk assessment report based on the statistical characteristics of the CVaR index and the revenue distribution.

[0023] Specifically, the first step is to obtain historical time-of-use electricity price data from the target electricity market as input. Ideally, continuous historical data for at least one seasonal period should be selected, such as hourly or 15-minute sampling covering at least several months. Missing values ​​are imputed, and abnormal peaks are corrected or removed based on statistical thresholds. The data is then subjected to seasonality testing and stabilization processing on a daily basis. Based on this, a seasonal time series forecasting model is established. Ideally, a seasonal autoregressive integral moving average is used to jointly fit the trend, seasonal, and random fluctuation terms of the electricity price. Parameter combinations are selected through rolling validation or information criteria, ensuring the model can depict both intraday peak-valley patterns and cross-day seasonal differences, yielding forecasts for multiple days and various time periods. A point-estimation sequence of electricity prices is generated. Subsequently, to characterize the uncertainty of electricity prices, the statistical characteristics of historical fitting residuals or prediction errors are preferably used as a source of random disturbance. Monte Carlo simulation is used to randomly sample the error term and superimpose it onto the point-estimation sequence, thereby generating multiple sets of future electricity price scenario sequences. The number of scenarios can be selected based on the risk assessment accuracy and computational resources, preferably from hundreds to thousands of sets. Stratified sampling or variance reduction techniques can be used to improve the coverage of tail-end scenarios. For each set of electricity price scenario sequences, a nonlinear mapping relationship between energy storage charging / discharging power and lifetime degradation cost is constructed to reflect the coupled impact of electricity prices, dispatch, and lifetime. Preferably, the relationship curve between battery discharge depth and cycle life is used as a basis, converting the discharge depth formed during dispatch into the current... The equivalent capacity degradation within a day or assessment period is then converted into an equivalent degradation cost based on the initial investment cost of the energy storage system or the cost of replaceable cells. This automatically suppresses strategies for deeper discharge and more frequent cycling at the same revenue level due to the increased degradation cost. In characterizing degradation, a maximum of two charge-discharge cycles per day is preferred to align with engineering operating habits. Multiple discharge segments within a day are combined to obtain the total discharge depth, which is then allocated to two equivalent cycles according to preset rules to read the corresponding capacity loss from the discharge depth lifetime curve, thus forming a degradation cost input for optimization. Subsequently, independent optimization sub-problems are solved for each electricity price scenario, with the optimization objective preferably set as the maximum peak-valley arbitrage revenue minus the lifetime degradation cost. The peak-valley arbitrage profit is determined by the scenario's electricity price and net power output. The constraints preferably include at least: upper limits for charging and discharging power, mutual exclusion constraints for charging and discharging (avoiding simultaneous charging and discharging in the same period), energy balance constraints for the evolution of state of charge with charging and discharging efficiency, upper and lower limits for state of charge, and continuity constraints for the state of charge at the beginning and end of the evaluation period to be consistent or to meet the preset terminal state of charge. In terms of solution method, it is preferable to express discrete logic such as mutual exclusion of charging and discharging as integer variables, and combine it with piecewise approximation of the degradation cost curve, so that each scenario sub-problem can be stably solved by a mixed integer linear programming solver, thereby outputting the optimal charging and discharging strategy for each time period in the scenario and obtaining the corresponding net profit value and degradation cost value.After obtaining net revenue samples from all scenarios, the data is aggregated to form a revenue distribution. The mean, standard deviation, and several quantiles are calculated to characterize the central trend and volatility of the revenue. Furthermore, kernel density estimation is used to smooth the discrete revenue samples, obtaining a continuous revenue probability density function to identify tail risks. In calculating risk indicators, a risk threshold for the revenue distribution is first determined at a given confidence level. Then, a conditional average is performed on the net revenue of the worst-case tail proportion scenarios to obtain the Conditional Value at Risk (CVaR), which characterizes the average loss level of net revenue under extremely unfavorable electricity price scenarios, thus simultaneously reflecting the comprehensive operational risk resulting from the superposition of market volatility and lifetime degradation costs. Finally, an operational risk assessment report is generated based on the CVaR indicator and the statistical characteristics of the revenue distribution. The report preferably includes at least: scenario generation parameters (prediction period, time granularity, number of scenarios, error sampling method), energy storage... Key parameters (rated capacity, power, efficiency, state of charge range, initial cost, and degradation conversion method), typical daily examples of the optimal scheduling strategy, interpretation of revenue distribution and tail risk, and suggestions on the impact of CVaR on scheduling conservatism and long-term economics at different confidence levels; for example, in an example application, for an electrochemical energy storage system with a rated power of 1 MW and a rated capacity of 2 MWh, the electricity price risk for the next 7 days can be assessed with a time granularity of 15 minutes, generating 1000 sets of electricity price scenarios and solving for 1000 net revenue samples respectively. Then, CVaR is calculated at a 95% confidence level as a conservative revenue floor, and the difference in revenue distribution under two standards—arbitrage without considering degradation and degradation cost—is compared to intuitively demonstrate the preferred implementation method of this invention, which achieves a balance between short-term revenue and long-term economics by incorporating lifetime degradation risk into the revenue objective and characterizing tail loss with CVaR.

[0024] The technical solution of this invention generates multiple Monte Carlo random scenario sequences based on historical electricity price data. Under each set of electricity price sequences, the charging and discharging strategy is optimized through a dynamic mapping model of discharge depth and lifespan loss. The scenario revenue distribution is aggregated and a probability density function is fitted. The tail risk is quantified using the CVaR index to achieve an integrated assessment of extreme scenarios and economic degradation. This effectively solves the problem of distorted operational decisions caused by the failure to simultaneously integrate the coupled effects of random fluctuations in electricity prices and battery lifespan loss in existing energy storage operation assessment methods, as well as the lack of quantification of multiple risk transmission mechanisms under extreme scenarios.

[0025] Furthermore, such as Figure 2 As shown, a seasonal time series model is established and an electricity price scenario series is generated, including: A seasonal autoregressive integral moving average model was used to fit historical electricity price data and perform prediction calculations to obtain the predicted electricity price sequence for future operating cycles. The electricity price forecasting error is characterized based on the electricity price forecasting sequence, and the electricity price forecasting error is treated as a source of electricity price uncertainty. Based on Monte Carlo simulation, random perturbation terms are generated by random sampling of electricity price prediction errors, and the random perturbation terms are combined with the electricity price prediction sequence to generate multiple sets of electricity price scenario sequences; Multiple sets of electricity price scenario sequences are used as electricity price input data for the operation scheduling optimization model of electrochemical energy storage system to characterize the electricity price uncertainty in the operating environment of electrochemical energy storage system.

[0026] As a preferred embodiment of the above, historical time-of-use electricity price data of the target electricity market is first obtained as a modeling sample. Continuous data covering the entire seasonal cycle is preferably selected, with hourly or 15-minute time granularities. Missing values, duplicate values, and abnormal spikes are then processed for consistency. For example, adjacent time intervals are used to imput individual missing points, and isolated spikes that clearly do not conform to market mechanisms are corrected or removed based on historical quantile thresholds. Simultaneously, the electricity price data is marked according to intraday and weekly cycles to facilitate subsequent extraction of seasonal patterns. After data cleaning, the historical electricity price sequence is fitted and predicted using a seasonal autoregressive integral moving average. Preferably, the trend changes, intraday repetition patterns, and random disturbances of the electricity price sequence are simultaneously incorporated into the same... The prediction framework weakens the influence of long-term trends by stabilizing historical sequences. During the fitting phase, a rolling training and validation method is preferred to select autoregressive, moving average, and seasonally relevant parameters for the model. This ensures the model captures both intraday peaks and troughs and reflects cross-day seasonal differences, thus outputting a time-by-time electricity price prediction sequence for the future operating cycle. After obtaining the electricity price prediction sequence, to explicitly transform the prediction error into a source of electricity price uncertainty, the residual sequence from the fitting phase or the rolling prediction error sequence is preferably used to characterize the error statistics. Descriptive statistics are performed on the mean bias, fluctuation scale, and possible heavy-tailed characteristics of the error. If necessary, error characteristics can be statistically analyzed separately for different time periods to avoid using a single error distribution to mask peaks and troughs. The prediction bias varies over time, thus making the error representation more closely reflect the time-varying heterogeneity of electricity price fluctuations. Based on this, Monte Carlo simulation is used to randomly sample the electricity price prediction error to generate random disturbance terms. Preferably, repeated sampling is performed using the empirical distribution of the error sequence as the sampling basis, allowing the disturbance terms to inherit the volatility and tail behavior of the real market error. The disturbance terms obtained from each sampling are then superimposed with the corresponding electricity price point prediction sequences according to their time indices, thereby forming multiple sets of future electricity price scenario sequences. The number of scenarios is preferably determined based on the risk assessment accuracy and computational resources, for example, set to hundreds to thousands of sets to obtain stable statistical results in subsequent revenue distribution and tail risk calculations. Simultaneously, stratified sampling can be preferably used to improve the probability of extreme disturbances being extracted. The system improves the coverage of tail scenarios by increasing the rate of electricity price scenarios. After generating the electricity price scenario sequence, a scenario consistency check is performed to ensure its engineering usability. For example, the system checks whether the range of electricity price values ​​in each time period is consistent with market rules, avoids significant unreasonable negative values ​​or large outliers caused by the superposition of disturbances, and performs boundary truncation on individual unreasonable points without changing the random nature of the system. Finally, the generated multiple sets of electricity price scenario sequences are used as the electricity price input data for the operation scheduling optimization model of the electrochemical energy storage system. This allows the charging and discharging strategies and net income results obtained from solving each scenario to reflect the random fluctuations introduced by error disturbances under the same prediction benchmark, thereby realizing the modeling and characterization of the uncertainty of electricity prices in the operating environment of the energy storage system.

[0027] Furthermore, a seasonal autoregressive integral moving average model is used to fit and predict historical electricity price data, including: Seasonal pattern recognition and trend processing are performed on historical electricity price data to make the historical electricity price data fit the requirements of the seasonal autoregressive integral moving average model. The temporal correlation of historical electricity price data is characterized by non-seasonal and seasonal autoregressive terms to generate structural constraints on the electricity price forecast sequence. The correlation between the non-seasonal moving average term and the seasonal moving average term in electricity price forecasting error is characterized to improve the ability of the electricity price forecasting sequence to fit random fluctuations. The model parameters are determined by removing trend and seasonal trend components from historical electricity price data through differential processing and fitting the model parameters to output an electricity price prediction sequence for generating multiple sets of electricity price scenario sequences in Monte Carlo simulation.

[0028] As a preferred embodiment of the above, the historical time-of-use electricity price data of the target electricity market is first preprocessed and seasonal pattern identified. Preferably, a continuous time series is constructed using a time granularity consistent with market clearing, such as hours or fifteen minutes. Missing points, duplicate points, and obvious abnormal peaks are then processed to ensure the series reflects real market fluctuations without being dominated by data flaws. Subsequently, the seasonal structure of the electricity price series is identified, preferably by simultaneously checking intraday and weekly repetition patterns. Different candidate seasonal cycles are compared, for example, using daily as the primary seasonal cycle and weekly as an auxiliary cycle for verification. The seasonal cycle used for modeling is determined by combining autocorrelation characteristics and interpretability. This provides a basis for setting the seasonal term in subsequent models. Regarding trend handling, it is preferable to first check the stationarity of the original electricity price series and then differencing the long-term and seasonal trends to make the processed series closer to stationarity in terms of mean and variance, thus meeting the fitting requirements of the seasonal autoregressive integral moving average model. The choice of the differencing order preferably follows the principle of sufficiently removing the trend while avoiding excessive differencing, i.e., retaining the necessary correlation structure for prediction after eliminating obvious trends and seasonal drift. After completing the above adaptation, the time correlation of the electricity price series is characterized using non-seasonal and seasonal autoregressive terms, preferably targeting the correlation of short-term lags and the correlation across seasonal cycles, respectively. Modeling is employed in two ways: the former constrains the continuity and regressiveness of electricity prices in adjacent time periods, while the latter constrains the recurring effects of the same seasonal location, such as the same time of day, across days or weeks. This forms a structural constraint on the future electricity price prediction sequence and improves the ability to maintain peak-valley patterns. Simultaneously, considering that electricity price prediction errors are often not independently and identically distributed over time, non-seasonal and seasonal moving average terms are introduced to characterize the correlation of prediction errors. Preferably, the autocorrelation of the fitting residuals is examined to determine whether an error-related term needs to be introduced, enabling the model to incorporate the portion of random fluctuations that can be explained by error correlation into the predictable structure, thereby improving the ability to predict short-term disturbances and peak spikes. The model exhibits strong fitting capabilities for random fluctuations such as peaks and troughs. Regarding parameter fitting, a phased parameter selection strategy is preferred: first, candidate non-seasonal and seasonal order combinations are selected based on correlation characteristics in the differencing sequence; then, the candidate combinations are compared using information criteria and rolling validation to select a parameter set that performs stably during both the training and validation periods, thus avoiding overfitting to historical data and resulting in extrapolation distortion. After fitting, the predicted electricity price sequence for each time period in the future operating cycle is output, and the fitting residuals or rolling prediction error statistics for the corresponding time period are preferably saved simultaneously, providing a consistent prediction benchmark and error source for generating multiple sets of electricity price scenario sequences through Monte Carlo simulation.

[0029] Furthermore, the Monte Carlo simulation method is used to generate multiple sets of electricity price scenario sequences, including: After obtaining the electricity price prediction sequence output by the seasonal autoregressive integral moving average model, the electricity price prediction error is sorted out to form an error characterization result that can be used for random sampling. Random sampling is performed on the error characterization results to obtain a random disturbance term that matches the electricity price prediction sequence, so that the random disturbance term reflects the random fluctuation characteristics of electricity prices; Random disturbance terms are superimposed on the electricity price prediction sequence to form an electricity price scenario sequence, and the differences between different electricity price scenario sequences are represented by random sampling. By aggregating a series of electricity price scenarios and using them as input conditions for independent optimization subproblems, we can model and express the uncertainty of electricity prices during the operation of electrochemical energy storage systems.

[0030] As a preferred embodiment of the above, after completing the fitting of the seasonal autoregressive integral moving average model to historical electricity prices and outputting the time-by-time electricity price prediction sequence for the future operating cycle, the electricity price prediction error is first sorted and standardized to ensure that the error can be directly used as the basis for random sampling. A preferred approach is to align the residual sequence obtained during the model fitting stage or the prediction error sequence obtained during the rolling prediction stage by time index, remove non-true error points introduced by factors such as missing data and abnormal peak correction, and statistically describe the mean bias and fluctuation amplitude of the error so that the error characterization results can truly reflect the random fluctuation characteristics of the market electricity price at different time points. Furthermore, considering that the electricity price error... Differences often exist across different time periods within a day. For example, errors are more easily amplified during peak periods and may be more concentrated during trough periods. It is preferable to group errors according to their intraday time position, that is, to group errors corresponding to the same time of day into the same group to form a time-segmented error sample pool. This ensures that the random disturbance terms generated by subsequent sampling not only inherit the overall fluctuation level but also retain the peak-trough error differences—a characteristic strongly correlated with energy storage arbitrage decisions—structurally. After obtaining sampleable error characterization results, Monte Carlo simulation is used for random sampling to generate random disturbance terms that match the electricity price prediction sequence point by point. Preferably, the empirical distribution of the error sample pool is used as the sampling source for repeated sampling, ensuring that each sampling is time-dimension related. Corresponding to the predicted sequence, in other words, for each time point in the predicted sequence, an error sample is extracted from the error sample pool at the corresponding time as the perturbation value at that time, thereby generating a complete perturbation sequence. Regarding the number of scenarios, the sampling number is preferably determined based on the stability requirements of the risk assessment and the computational resources for solving subsequent sub-problems, for example, taking hundreds to thousands of perturbation sequences to ensure sufficient sample support for the tails of the revenue distribution. Simultaneously, without changing the scope of the claims, a sampling strategy more biased towards the tails can be adopted to increase the chance of extreme perturbations being sampled, thus ensuring more comprehensive coverage of unfavorable electricity price scenarios in the scenario set. Subsequently, the random perturbation term is superimposed onto the electricity price prediction... The electricity price scenario sequence is formed from the prediction sequence. It is preferable to maintain the consistency principle of the same prediction benchmark and different disturbances. That is, all scenarios share the same point prediction sequence as the central path, and the differences only come from the disturbance terms obtained by random sampling. This allows the differences between different electricity price scenarios to be clearly represented by random sampling and ensures that the scenarios are comparable. In order to ensure that the generated scenarios are usable in engineering, it is preferable to make reasonable adjustments to each scenario after superposition. For example, check whether there are extreme values ​​that obviously do not conform to market rules, and perform boundary constraint processing without destroying the randomness characteristics. This ensures that the electricity price scenario can reflect random fluctuations without affecting the stability of subsequent scheduling solutions due to individual unusable points.Finally, the generated multiple sets of electricity price scenario sequences are aggregated and used as the electricity price input conditions for subsequent independent optimization sub-problems. This ensures that each sub-problem, under the same energy storage parameters and constraints, produces different optimal charging / discharging strategies and net profit results only due to different electricity price scenarios, thereby achieving a modeling expression of the electricity price uncertainty during the operation of electrochemical energy storage systems.

[0031] Furthermore, such as Figure 3 As shown, a nonlinear mapping model is established, including: Establish an operation scheduling optimization model for electrochemical energy storage systems with the goal of maximizing net operating revenue within the operating cycle, and include the transaction revenue generated from electricity market transactions in the revenue item of the operation scheduling optimization model; The cost item of the operation scheduling optimization model includes the equipment aging cost value represented by the life degradation cost, and the life degradation cost is made to correspond to the charging and discharging process to reflect the impact of life degradation. In the operation scheduling optimization model, charging power constraints and discharging power constraints are set to limit the feasible range of charging and discharging power, and charging state evolution constraints are set to describe energy changes. In the operation scheduling optimization model, upper and lower limits of charge state and continuity constraints of charge state are set to ensure the feasibility of charge state and the consistency of scheduling cycle, and a charging and discharging strategy that matches the electricity price scenario is output.

[0032] As a preferred embodiment of the above, an operation scheduling optimization process is first established with the overall objective of maximizing the net operating revenue of the electrochemical energy storage system within a given operating cycle. The operating cycle is preferably a single-day rolling cycle or a multi-day continuous cycle, and the time granularity is preferably consistent with the electricity market clearing granularity, such as an hour or fifteen minutes. Electricity price scenario sequences are used as external inputs to ensure that the scheduling results correspond to an uncertain electricity price environment. Regarding the construction of revenue items, the energy arbitrage revenue generated from selling electricity to and buying electricity from the electricity market during each time period is preferably included in the net revenue. Specifically, the transaction revenue generated from the clearing price and discharge power during the discharge period, and the electricity purchase expenditure generated from the clearing price and charging power during the charging period, are included in the same revenue accounting standard. The market transaction revenue is calculated by subtracting electricity purchase expenses from transaction revenue, enabling the model to automatically select charging and discharging times driven by peak-valley price differences. Regarding cost item construction, to reflect the coupling of lifespan degradation costs with charging and discharging behavior and their inclusion in scheduling decisions, it is preferable to simultaneously include the equipment aging cost value, represented by lifespan degradation costs, in the net revenue. Furthermore, this lifespan degradation cost is not a fixed constant but dynamically changes with the charging and discharging strategy. In specific implementation, it is preferable to extract discharge depth information based on the charging and discharging trajectory within the operating cycle, and map deep discharge and frequent cycling behaviors to a larger capacity decay amount based on the relationship curve between discharge depth and cycle life. Then, the capacity decay amount is converted into an equivalent degradation cost based on the initial battery cost or the cost of replaceable cells, ensuring that the same... Under the arbitrage benefit of electricity prices, if the profit is obtained through deeper discharge depth, the increased degradation cost will suppress it in the objective function, thus achieving an endogenous trade-off between short-term gains and long-term lifetime. To ensure that the degradation cost corresponds to the charge-discharge process and is operable, a degradation characterization method based on discharge depth estimation is preferred, and a maximum of two equivalent charge-discharge cycles per day is constrained on a daily scale: after merging multiple discharge behaviors within the day to obtain the total discharge depth, it is then allocated to the discharge depth of two equivalent cycles according to a preset rule, and the corresponding capacity loss is read and accumulated to obtain the daily capacity decay, which is then converted into the daily degradation cost and included in the cost item, so that the degradation cost changes with the strategy, thus constituting the charge-discharge power, discharge depth, The nonlinear mapping relationship between capacity decay and degradation cost; in terms of constraint setting, to ensure the engineering feasibility of the output strategy, it is preferable to set charging power constraints and discharging power constraints simultaneously during the scheduling process to limit the feasible range of charging and discharging power. In order to avoid non-physical behavior caused by simultaneous charging and discharging at the same time, it is preferable to restrict the charging and discharging states to only one at the same time through mutual exclusion logic. At the same time, a state of charge evolution constraint is set to describe the process of energy change with charging and discharging. It is preferable to consider charging efficiency and discharging efficiency so that the energy increment of the state of charge increases when charging according to efficiency and the energy decrease when discharging according to efficiency. This evolution is continuously passed through time periods throughout the entire operation cycle to ensure energy conservation and strategy execution.Furthermore, to ensure the feasibility of the state of charge (SOC) and the consistency of the operating cycle, upper and lower limits for SOC are preferably set to keep the SOC within the safe range allowed by the equipment. A continuity constraint for SOC is also set to ensure that the SOC at the end of the scheduling cycle is consistent with the SOC at the beginning of the scheduling cycle or meets the preset terminal SOC requirements. This avoids the problem of unsustainable cross-cycle discharge caused by overdraft discharge and makes the strategies under different electricity price scenarios comparable. In terms of output, for each electricity price scenario sequence, a time-by-time charging and discharging strategy matching that electricity price scenario is obtained under the combined effect of the aforementioned revenue and degradation cost items, as well as constraints such as power and SOC. Simultaneously, the corresponding market transaction revenue value, lifetime degradation cost value, and net revenue value are output, thus providing basic data for subsequent scenario-based aggregation to form revenue distribution and tail risk assessment.

[0033] Furthermore, the cost of lifespan degradation should be taken into account, including: A battery degradation model based on the depth of cycle discharge is introduced into the operation scheduling optimization model, and a depth of discharge estimation algorithm is used to identify the charge and discharge cycle process of the electrochemical energy storage system. The total depth of discharge within the operating cycle is calculated based on the charge-discharge cycle process, and the total depth of discharge is allocated to the depth of discharge corresponding to multiple charge-discharge cycles to generate a set of depth of discharge for degradation assessment. The capacity loss corresponding to the set of discharge depths is estimated based on the relationship curve between discharge depth and cycle number, and the total capacity decay of each charge and discharge cycle is obtained by summing up the capacity loss of each cycle. The total daily capacity decay is combined with the initial cost of the electrochemical energy storage system to convert it into an equivalent degradation cost, which is then written into the cost item of the operation scheduling optimization model as a lifetime degradation cost, so that the net operating revenue reflects both the transaction revenue and the equipment aging cost.

[0034] As a preferred embodiment of the above, in the scheduling of the electrochemical energy storage system aimed at maximizing net operating benefits, a battery degradation model based on the depth of charge / discharge is simultaneously introduced. This prevents degradation costs from being treated as a fixed constant, instead allowing them to be dynamically determined by the charge / discharge strategy. This achieves the coupled inclusion of charge / discharge behavior, depth of discharge, capacity decay, and equivalent cost. Specifically, firstly, within a given operating cycle, preferably a single day, an hour with the same granularity as market clearing, or a 15-minute interval, the charge / discharge cycle process is identified based on the charge / discharge power sequence and state of charge sequence obtained from the scheduling. Preferably, a depth of discharge estimation algorithm is used to find the state of charge transition from charging to discharging on the state of charge curve. The system identifies the local high point of discharge and the local low point at the end of discharge, treating the continuous discharge interval as a single discharge process. It also merges and organizes multiple discharge segments within a day, such as brief recharging or zero-power intervals, to obtain the discharge depth information that can be attributed to cyclic aging within the operating cycle. Based on this, the total discharge depth within the operating cycle is calculated. The total discharge depth is preferably characterized by the ratio of the daily equivalent discharge energy to the rated capacity, ensuring it covers the cumulative impact of multiple discharges within the day and facilitates alignment with subsequent discharge depth versus cycle number curves. Furthermore, to ensure consistency between the degradation assessment and the disclosed information and to facilitate stable operation in scheduling optimization... The embedded method preferably sets a maximum of two equivalent charge-discharge cycles per day. When there are more than two actual discharge processes within a day, instead of calculating the lifetime for each individual cycle, the total discharge depth is first obtained, and then allocated into multiple sets according to a preset allocation rule, preferably the discharge depths corresponding to two charge-discharge cycles, forming a set of discharge depths for degradation assessment. This allows for characterizing the daily cycle aging intensity with a finite number of equivalent cycles, avoiding instability in degradation calculations and optimization solutions caused by fragmented cycles. Subsequently, the capacity loss corresponding to the set of discharge depths is estimated based on the relationship curve between discharge depth and cycle count. The preferred approach is to estimate the discharge depth for each equivalent cycle from... The relationship curve is read or converted to obtain the tolerable cycle level at the discharge depth, and the equivalent cycle's lifespan consumption is converted into capacity decay. The capacity loss of each equivalent cycle is then summarized to obtain the total capacity decay for the day, so that deeper discharge naturally corresponds to higher capacity loss in the model. Finally, the total capacity decay for the day is combined with the initial cost of the electrochemical energy storage system to convert it into equivalent degradation cost, and written as a lifespan degradation cost into the cost item of the operation scheduling optimization model. This allows the net operating revenue to simultaneously reflect the electricity market transaction revenue and equipment aging cost during accounting, thereby automatically suppressing the strategy of excessively deep discharge or excessive cycling under the same electricity price arbitrage space.

[0035] Furthermore, such as Figure 4 As shown, solving the independent optimization subproblems includes: Based on multiple sets of electricity price scenario sequences, the operation scheduling optimization model is broken down into multiple scheduling sub-problems according to the electricity price scenario, and each scheduling sub-problem is associated with the corresponding electricity price scenario. Maintain consistency in charging power constraints, discharging power constraints, state of charge evolution constraints, upper and lower limits of state of charge constraints, state of charge continuity constraints, and battery degradation model across multiple scheduling subproblems. The scheduling subproblem is solved independently using mathematical programming methods combined with mixed-integer linear programming to obtain the matching charging and discharging strategies, operational benefits and battery degradation results under various electricity price scenarios. The solutions to the scheduling subproblems are aggregated to form a set of operating revenues and a set of battery degradation costs. These sets are then used for statistical analysis of the distribution of operating results and conditional risk value assessment.

[0036] As a preferred embodiment of the above, after obtaining multiple sets of future electricity price scenario sequences, it is preferable to use them as inputs for multiple possible operating trajectories of the same energy storage system under an uncertain electricity price environment. Following the principle of one-to-one correspondence between scenarios and consistent model structure, the established operation scheduling optimization model is decomposed according to the electricity price scenarios, forming multiple independent scheduling sub-problems. Each scheduling sub-problem only uses its corresponding electricity price scenario sequence for the electricity price input, while the equipment parameters, scheduling cycle, and time granularity of the energy storage system remain completely consistent, thereby ensuring the comparability of charging and discharging strategies obtained under different scenarios. During the decomposition process, it is preferable to explicitly establish a mapping relationship between scenario number, electricity price sequence, and sub-problems. For example, the first set of electricity price scenarios can be used as the input for the first sub-problem, and the second set of electricity price scenarios can be used as the input for the second sub-problem, until all scenarios are fully covered. This ensures that the solution results for each subsequent set can be traced back to the corresponding electricity price situation and used for statistical analysis. Furthermore, to ensure that the differences between scenarios only come from electricity price disturbances rather than changes in model caliber, it is preferable to use the same constraint system and the same battery degradation characterization method in all scheduling sub-problems. Specifically, this includes: setting upper limit constraints on charging power and discharging power to limit the feasible power range, while maintaining consistent mutual exclusion logic for charging and discharging states to avoid non-physical simultaneous charging and discharging at the same time; and for the state of charge... The evolution of the state of charge (SBC) adopts the same energy balance standard, continuously considering charging and discharging efficiency and time step size, so that the SBC is transferred time-by-time with the charging and discharging power and satisfies energy conservation. Consistent upper and lower limit ranges are set for the SBC to meet safe operation requirements, and consistent SBC continuity constraints are set to ensure that the SBC at the end of the scheduling cycle is consistent with the initial SBC or meets the preset terminal requirements, thereby avoiding artificially increasing short-term gains in a particular scenario by overdrawing the terminal SBC. Simultaneously, all sub-problems share the same battery degradation model and its discharge depth estimation and equivalent degradation cost conversion caliber, ensuring that the inhibitory effect of degradation costs on the strategy remains consistent under different electricity price scenarios, thus enabling realistic... This study reflects the difference in the impact of the coupling of electricity price uncertainty and lifespan degradation costs on revenue and strategy. In terms of solution method, it is preferred to use mathematical programming to solve each sub-problem independently, and combine it with mixed integer linear programming to handle discrete decisions such as charge-discharge mutual exclusion. This allows each sub-problem to output the optimal charge-discharge strategy for each time period under a given electricity price scenario, and at the same time obtain the corresponding operating revenue and battery degradation results. The operating revenue is preferably calculated by deducting the equivalent degradation cost from the market transaction revenue. The battery degradation result is preferably output as the daily capacity decay amount and its converted degradation cost value, so as to observe the synergistic changes of revenue fluctuation and aging cost in subsequent risk assessment.After solving all sub-problems, the preferred approach is to collect and organize the results to form a set of operational revenues and battery degradation costs covering all electricity price scenarios, maintaining an index relationship consistent with the scenario numbers. This allows for simultaneous tracking of whether low-revenue degradation levels are abnormally high or whether high-revenue costs come at a significant lifespan penalty when statistically analyzing revenue distribution, analyzing tail-end scenarios, and calculating conditional value of risk.

[0037] Furthermore, such as Figure 5 As shown, the generation of the profit distribution and the fitting probability density function include: Statistical processing is performed on the set of operating revenues and the set of battery degradation costs to obtain the revenue distribution characteristics, and the set of operating revenues is sorted to form an ordered sequence of operating revenues for quantile representation. Based on the calculation of statistics to describe the central trend and volatility of the operating results set, and combined with quantile characterization to reflect extreme characteristics, basic descriptive information of the distribution of operating results is generated. The kernel density estimation method is used to perform nonparametric density estimation on the set of operating returns to form a continuous probability density distribution function, and the overall shape of the distribution of operating results and the tail risk characteristics are observed. Based on the distribution of operational results, the Conditional Value at Risk (VaR) index is used to quantify the tail loss corresponding to extreme unfavorable electricity price scenarios and generate operational risk level evaluation results. Combined with the changes in VaR under different confidence levels and tail risk characteristics, quantitative decision-making basis is provided for the selection of scheduling strategies for electrochemical energy storage systems.

[0038] As a preferred embodiment, the set of operating revenues and the set of battery degradation costs are first aligned and consistent. Preferably, each operating revenue is aligned with its corresponding degradation cost using the scenario number as an index. Invalid samples resulting from solution failures, abnormal inputs, or failure to meet constraints are removed. The revenue caliber is then verified to ensure that it simultaneously reflects both transaction revenues and equipment aging costs, thereby guaranteeing that the risk sources reflected in subsequent distribution statistics truly include the combined impact of electricity price uncertainty and lifespan degradation costs. Subsequently, the set of operating revenues is sorted to form an ordered sequence of operating revenues, and quantiles are used to characterize these revenues. Preferably, at least several position points representing normal and extreme levels are extracted, for example, to characterize poor performance. The distribution of running results, including the lower quantile, the median quantile (characterizing the neutral case), and the upper quantile (characterizing the better case), not only describes the central location but also directly reveals the lower bound characteristics of the tail interval's returns. Furthermore, without exceeding the scope of the claims, the set of degradation costs can be synchronously sorted or grouped according to the same index as the returns to observe whether low-return tails are accompanied by abnormally high degradation costs, thus providing support for explaining the tail mechanism. Regarding the calculation of basic statistics, it is preferable to calculate statistics describing the central trend and volatility based on the running return set, such as using the sample mean to represent the average return level and the sample dispersion to represent the return volatility, combined with the aforementioned quantiles to represent the inverse trend. To reflect extreme characteristics, the same set of results can simultaneously present different information, thus forming a basic description of the distribution of operating results. Regarding continuous density fitting, to avoid the problem that discrete histogram statistics alone are insufficient to stably identify tail shapes and multimodal characteristics, a kernel density estimation method is preferred to perform nonparametric density estimation on the operating profit set. This involves superimposing a smoothing kernel function around each profit sample and appropriately selecting the smoothing degree, so that the estimated continuous probability density distribution function can retain the overall shape of the distribution without excessively amplifying noise. This facilitates observation of risk characteristics such as whether the profit distribution exhibits skewness, the presence of multimodal structures, and whether the tails have long extensions. Regarding risk quantification, it is preferable to base the operation results on the aforementioned... The conditional value of risk index is introduced to quantify the tail loss corresponding to extreme unfavorable electricity price scenarios. In specific implementation, under a given confidence level, the tail threshold position is first determined by using an ordered payoff sequence. Then, the worst tail samples falling below the threshold are conditionally averaged to obtain the average payoff level or average loss level when entering the worst case under the given confidence level. This is used as the result of the operational risk level evaluation. This calculation can be repeated at multiple confidence levels to obtain a set of conditional values ​​of risk that vary with the confidence level. This is used to characterize the degree of tail bottom line decline or loss aggravation when the more conservative the confidence level, i.e., the higher the confidence level. Thus, the tail risk is transformed from a qualitative observation into a comparable quantitative basis that can be used to constrain the selection of scheduling strategies.Finally, the continuous distribution shape formed by the aforementioned basic statistical information and kernel density estimation is jointly interpreted with the conditional value of risk (VHR) results at different confidence levels. The preferred output is conclusive information that can directly guide scheduling strategy selection. For example, when the VHR shows significantly low tail returns, it suggests a more conservative charging / discharging strategy in subsequent operations or parameter settings to reduce tail loss exposure. Conversely, when tail risk is controllable and average returns and volatility indicators are favorable, the strategy can be allowed to moderately increase its intensity while meeting degradation cost constraints.

[0039] Example 2; Based on the same inventive concept as the energy storage operation assessment method considering electricity price lifetime risk in the foregoing embodiments, the present invention also provides an energy storage operation assessment system considering electricity price lifetime risk, the system comprising: The scenario construction module establishes a seasonal time series model based on historical electricity price data, predicts the estimated electricity price points for various times in the next few days, and generates multiple sets of electricity price scenario sequences through Monte Carlo simulation. The lifetime determination module establishes a nonlinear mapping model between energy storage charging and discharging power and lifetime degradation cost for each set of electricity price scenario sequences. The nonlinear mapping model dynamically calculates the lifetime degradation cost caused by charging and discharging behavior based on the relationship curve between the depth of discharge and cycle life of the energy storage battery. The strategy matching module solves an independent optimization subproblem for each set of electricity price scenario sequences. The objective is to maximize the peak-valley arbitrage revenue minus the lifetime degradation cost. Under the constraints of charge-discharge state and battery capacity, it outputs the matched charge-discharge strategy and obtains the corresponding net revenue value. The distribution fitting module aggregates the net revenue values ​​of all electricity price scenario sequences to generate a revenue distribution, and estimates the probability density function of the fitted revenue distribution based on kernel density. The indicator calculation module calculates the Conditional Value at Risk (CVaR) indicator based on the probability density function. The CVaR indicator represents the expected tail loss of the net income distribution under extremely unfavorable electricity price scenarios. The risk assessment module generates an operational risk assessment report based on the statistical characteristics of the CVaR index and the return distribution.

[0040] The adjustment system described above in this invention can effectively realize an energy storage operation assessment method that takes into account electricity price lifetime risk. The technical effects it can achieve are as described in the above embodiments, and will not be repeated here.

[0041] Furthermore, the scene construction module includes: The cycle prediction unit uses a seasonal autoregressive integral moving average model to fit historical electricity price data and perform prediction calculations to obtain the electricity price prediction sequence for future operating cycles. The error characterization unit characterizes the electricity price prediction error based on the electricity price prediction sequence and treats the electricity price prediction error as a source of electricity price uncertainty. The random perturbation unit generates random perturbation terms by randomly sampling the electricity price prediction error based on Monte Carlo simulation, and combines the random perturbation terms with the electricity price prediction sequence to generate multiple sets of electricity price scenario sequences; The scenario determination unit uses multiple sets of electricity price scenario sequences as electricity price input data for the operation scheduling optimization model of the electrochemical energy storage system to characterize the electricity price uncertainty in the operating environment of the electrochemical energy storage system.

[0042] Similarly, the above-mentioned optimization schemes for the system can also achieve the optimization effects corresponding to the methods in Embodiment 1, which will not be repeated here.

[0043] Although this application has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made thereto without departing from the spirit and scope of this application. Accordingly, this specification and drawings are merely exemplary illustrations of the application as defined herein, and are to be considered as covering any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Thus, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A method for assessing the operation of energy storage that takes into account electricity price lifetime risk, characterized in that, The method includes: A seasonal time series model is established based on historical electricity price data to predict the estimated electricity price points at various times in the coming days, and multiple sets of electricity price scenario sequences are generated through Monte Carlo simulation. For each set of electricity price scenarios, a nonlinear mapping model is established between the energy storage charging and discharging power and the lifetime degradation cost. The nonlinear mapping model dynamically calculates the lifetime degradation cost caused by the charging and discharging behavior based on the relationship curve between the depth of discharge and cycle life of the energy storage battery. For each set of electricity price scenario sequences, solve an independent optimization subproblem with the objective of maximizing the peak-valley arbitrage profit minus the lifetime degradation cost. Under the constraints of charge-discharge state and battery capacity, output a matching charge-discharge strategy and obtain the corresponding net profit value. The net revenue values ​​of all the electricity price scenario sequences are aggregated to generate a revenue distribution, and the probability density function of the revenue distribution is fitted according to the kernel density estimate; The Conditional Value at Risk (CVaR) index is calculated based on the probability density function. The CVaR index represents the expected tail loss of the net income distribution under extremely unfavorable electricity price scenarios. An operational risk assessment report is generated based on the CVaR index and the statistical characteristics of the revenue distribution.

2. The energy storage operation assessment method considering electricity price lifetime risk according to claim 1, characterized in that, Establishing a seasonal time series model and generating electricity price scenario sequences, including: The historical electricity price data is fitted and predicted using a seasonal autoregressive integral moving average model to obtain the predicted electricity price sequence for future operating cycles. The electricity price prediction error is characterized based on the electricity price prediction sequence, and the electricity price prediction error is treated as a source of electricity price uncertainty. Based on the Monte Carlo simulation, random perturbation terms are generated by randomly sampling the electricity price prediction error, and the random perturbation terms are combined with the electricity price prediction sequence to generate multiple sets of electricity price scenario sequences; Multiple sets of the aforementioned electricity price scenario sequences are used as electricity price input data for the operation scheduling optimization model of the electrochemical energy storage system to characterize the electricity price uncertainty in the operating environment of the electrochemical energy storage system.

3. The energy storage operation assessment method considering electricity price lifetime risk according to claim 2, characterized in that, A seasonal autoregressive integral moving average model was used to fit and predict historical electricity price data, including: Seasonal pattern recognition is performed on the historical electricity price data, and trend changes are processed to make the historical electricity price data fit the fitting requirements of the seasonal autoregressive integral moving average model. The temporal correlation of the historical electricity price data is characterized based on non-seasonal and seasonal autoregressive terms to generate structural constraints on the electricity price prediction sequence. The correlation between the non-seasonal moving average term and the seasonal moving average term in the electricity price forecast error is characterized to improve the fitting ability of the electricity price forecast sequence to random fluctuations. The trend and seasonal trend components in the historical electricity price data are removed by differential processing, and the model parameters are fitted and determined to output the electricity price prediction sequence for generating multiple sets of electricity price scenario sequences in the Monte Carlo simulation.

4. The energy storage operation assessment method considering electricity price lifetime risk according to claim 2, characterized in that, Multiple sets of the aforementioned electricity price scenario sequences were generated using the Monte Carlo simulation method, including: After obtaining the electricity price prediction sequence output by the seasonal autoregressive integral moving average model, the electricity price prediction error is sorted out to form an error characterization result that can be used for random sampling. Random sampling is performed on the error characterization results to obtain a random disturbance term that matches the electricity price prediction sequence, so that the random disturbance term reflects the random fluctuation characteristics of electricity prices; The random disturbance term is superimposed on the electricity price prediction sequence to form the electricity price scenario sequence, and the differences between different electricity price scenario sequences are characterized by the random sampling. The electricity price scenario sequence is collected and used as the electricity price input condition for the independent optimization sub-problem, so as to realize the modeling and expression of the electricity price uncertainty during the operation of the electrochemical energy storage system.

5. The energy storage operation assessment method considering electricity price lifetime risk according to claim 1, characterized in that, Establish a nonlinear mapping model, including: An operation scheduling optimization model for an electrochemical energy storage system is established with the goal of maximizing net operating revenue within the operating cycle, and the transaction revenue generated from electricity market transactions is included in the revenue item of the operation scheduling optimization model. The equipment aging cost value characterized by the lifespan degradation cost is included in the cost item of the operation scheduling optimization model, and the lifespan degradation cost is made to correspond to the charging and discharging process to reflect the impact of lifespan degradation. In the operation scheduling optimization model, charging power constraints and discharging power constraints are set to limit the feasible range of charging and discharging power, and charging state evolution constraints are set to describe energy changes. The operation scheduling optimization model sets upper and lower limits of charge state and continuity constraints of charge state to ensure the feasibility of charge state and the consistency of scheduling cycle, and outputs a charging and discharging strategy that matches the electricity price scenario.

6. The energy storage operation assessment method considering electricity price lifetime risk according to claim 5, characterized in that, The cost of the aforementioned lifespan degradation includes: A battery degradation model based on the depth of discharge cycle is introduced into the operation scheduling optimization model, and a depth of discharge estimation algorithm is used to identify the charge and discharge cycle process of the electrochemical energy storage system. The total depth of discharge within the operating cycle is calculated based on the charge-discharge cycle process, and the total depth of discharge is allocated to the depth of discharge corresponding to multiple charge-discharge cycles to generate a set of depths of discharge for degradation assessment. The capacity loss corresponding to the set of discharge depths is estimated based on the relationship curve between the discharge depth and the number of cycles, and the total capacity decay for the day is obtained by summing up the capacity loss of each charge and discharge cycle. The total daily capacity decay is combined with the initial cost of the electrochemical energy storage system to convert it into an equivalent degradation cost, which is then written into the cost item of the operation scheduling optimization model as the lifetime degradation cost, so that the net operating revenue simultaneously reflects the transaction revenue and the equipment aging cost.

7. The energy storage operation assessment method considering electricity price lifetime risk according to claim 1, characterized in that, Solving independent optimization subproblems, including: Based on multiple sets of the electricity price scenario sequences, the operation scheduling optimization model is divided into multiple scheduling sub-problems according to the electricity price scenario, and each scheduling sub-problem is associated with the corresponding electricity price scenario. Maintain consistency in the charging power constraints, discharging power constraints, state of charge evolution constraints, upper and lower limits of state of charge constraints, state of charge continuity constraints, and battery degradation model across multiple scheduling sub-problems. The scheduling subproblems are solved independently using mathematical programming methods combined with mixed-integer linear programming to obtain the matching charging and discharging strategies, operational benefits, and battery degradation results for each of the electricity price scenarios. The solutions to the scheduling subproblems are aggregated to form a set of operating benefits and a set of battery degradation costs. The set of operating benefits and the set of battery degradation costs are then used for statistical analysis of the distribution of operating results and conditional risk value assessment.

8. The energy storage operation assessment method considering electricity price lifetime risk according to claim 1, characterized in that, Generate the payoff distribution and fit the probability density function, including: Statistical processing is performed on the set of operating revenues and the set of battery degradation costs to obtain the revenue distribution characteristics, and the set of operating revenues is sorted to form an ordered sequence of operating revenues for quantile representation. Based on the set of operating returns, a statistical measure is calculated to describe the central trend and volatility, and combined with the quantile characterization to reflect extreme characteristics, generating basic descriptive information of the distribution of operating results; The kernel density estimation method is used to perform nonparametric density estimation on the set of operating returns to form a continuous probability density distribution function, and the overall shape of the distribution of operating results and the tail risk characteristics are observed. Based on the distribution of the operating results, the conditional value of risk index is used to quantify the tail loss corresponding to the extreme unfavorable electricity price scenario and generate an evaluation result of the operating risk level. Combined with the changes in conditional value of risk under different confidence levels and the tail risk characteristics, a quantitative decision-making basis is provided for the selection of scheduling strategies for electrochemical energy storage systems.

9. An energy storage operation assessment system that takes into account electricity price lifetime risk, characterized in that, The system includes: The scenario construction module establishes a seasonal time series model based on historical electricity price data, predicts the estimated electricity price points for various times in the next few days, and generates multiple sets of electricity price scenario sequences through Monte Carlo simulation. The lifetime determination module establishes a nonlinear mapping model between energy storage charging and discharging power and lifetime degradation cost for each set of electricity price scenario sequences. The nonlinear mapping model dynamically calculates the lifetime degradation cost caused by charging and discharging behavior based on the relationship curve between the depth of discharge and cycle life of the energy storage battery. The strategy matching module solves an independent optimization sub-problem for each set of electricity price scenario sequences. The objective is to maximize the peak-valley arbitrage revenue minus the lifetime degradation cost. Under the constraints of charge-discharge state and battery capacity, it outputs the matched charge-discharge strategy and obtains the corresponding net revenue value. The distribution fitting module aggregates the net revenue values ​​of all electricity price scenario sequences to generate a revenue distribution, and estimates the probability density function of the fitted revenue distribution based on kernel density. The indicator calculation module calculates the Conditional Value at Risk (CVaR) indicator based on the probability density function. The CVaR indicator represents the expected tail loss of the net income distribution under extremely unfavorable electricity price scenarios. The risk assessment module generates an operational risk assessment report based on the statistical characteristics of the CVaR index and the return distribution.

10. The energy storage operation assessment system considering electricity price lifetime risk according to claim 9, characterized in that, The scene construction module includes: The cycle prediction unit uses a seasonal autoregressive integral moving average model to fit historical electricity price data and perform prediction calculations to obtain the electricity price prediction sequence for future operating cycles. The error characterization unit characterizes the electricity price prediction error based on the electricity price prediction sequence and treats the electricity price prediction error as a source of electricity price uncertainty. The random perturbation unit generates random perturbation terms by randomly sampling the electricity price prediction error based on Monte Carlo simulation, and combines the random perturbation terms with the electricity price prediction sequence to generate multiple sets of electricity price scenario sequences; The scenario determination unit uses multiple sets of electricity price scenario sequences as electricity price input data for the operation scheduling optimization model of the electrochemical energy storage system to characterize the electricity price uncertainty in the operating environment of the electrochemical energy storage system.

Images