An energy storage system operation and maintenance strategy optimization method based on digital twinning

Abstract

The application discloses a kind of based on digital twinning energy storage system operation strategy optimization method, belong to the storage and intelligent power grid technical field of electric energy, it includes: obtaining the whole process information of energy storage power station;Establish the digital twin synchronization model of energy storage system and calibration;Based on digital twin model, generate the prediction data of current time and carry out residual error analysis with real-time data, generate quantitative diagnostic indicators;According to quantitative diagnostic indicators and fault mode, adjust the parameters of digital twin synchronization model, ensure that model and energy storage system actual state keep synchronization;Real-time data is input into adjusted digital twin model, calculate the future operation index of energy storage system, generate simulation running data;According to simulation running data, fault mode and operation rule library, generate non-periodic operation strategy.The application adopts adaptive calibration digital twin model, can fuse real-time diagnosis and forward-looking optimization, significantly improve energy storage system operation efficiency and reliability.

Description

Technical Field

[0001] This invention relates to the fields of electrical energy storage and smart grids, and in particular to a method for optimizing the operation and maintenance strategy of energy storage systems based on digital twins. Background Technology

[0002] Energy storage systems, as a crucial component of modern power systems, play a key role in enhancing grid flexibility, promoting renewable energy consumption, and ensuring power quality. They achieve the spatial and temporal transfer of energy by charging when there is surplus power and discharging when there is a shortage, making them one of the core technological equipment for building new power systems. The safe and efficient operation of energy storage systems directly affects the stability and economy of the entire power system; therefore, their scientific and effective operation and maintenance are essential. Currently, the operation and maintenance of energy storage systems mainly relies on periodic inspections and status monitoring based on data acquisition and monitoring systems. Maintenance personnel inspect and maintain the equipment according to fixed time cycles, while simultaneously acquiring basic operating parameters such as voltage, current, and temperature through monitoring systems. Some advanced systems use simplified physical or empirical models to estimate the battery's health and state of charge, triggering alarms based on preset thresholds. In terms of operational strategies, a single-objective optimization method based on economic efficiency or meeting specific grid service needs is typically used to formulate fixed charging and discharging schedules.

[0003] However, existing technical solutions have significant drawbacks. Regular maintenance models cannot promptly detect sudden failures and gradual performance degradation, easily leading to over-maintenance or under-maintenance, resulting in low operational efficiency. Monitoring and estimation methods based on simple models, failing to adequately consider multi-physics coupling effects and individual equipment differences, cause their model accuracy to decline significantly as the system ages, leading to inaccurate condition assessments. Furthermore, single-objective optimization strategies often neglect long-term equipment lifespan degradation and operational safety risks, easily sacrificing equipment health for short-term economic gains, making it difficult to maximize the value of energy storage assets throughout their entire lifecycle. Summary of the Invention

[0004] To address the aforementioned issues, this invention provides a digital twin-based method for optimizing the operation and maintenance strategy of energy storage systems. This method employs a technical solution that involves constructing a digital twin synchronization model, performing residual analysis and dynamic calibration, and combining it with a multi-objective decision engine for strategy optimization. This approach enables predictive, adaptive, and globally optimal operation and maintenance of energy storage systems, comprehensively improving the system's safety, reliability, and economic benefits.

[0005] The above objectives can be achieved through the following approach:

[0006] A method for optimizing the operation and maintenance strategy of an energy storage system based on digital twins includes: acquiring full-process information of the energy storage power station, including real-time and historical data of battery cluster temperature difference, power converter switching transient characteristics, grid connection point harmonic spectrum, and battery status, as well as fault cases and operation and maintenance records of the energy storage power station; establishing a digital twin synchronization model of the energy storage system, and using the historical data to train and calibrate the battery physical parameter model and dynamic operation model of the digital twin synchronization model; converting the fault cases and operation and maintenance records into a fault mode and operation and maintenance rule base; and generating predictive data for the current moment based on the digital twin synchronization model and historical data, and combining it with... Real-time data is subjected to residual analysis to calculate the mean and standard deviation of the residuals between the two data sets, generating quantitative diagnostic indicators. Based on these quantitative diagnostic indicators and in conjunction with fault modes, the parameters of the digital twin synchronization model are adjusted to ensure that the digital twin synchronization model remains synchronized with the actual operating state of the energy storage system. Real-time data is then input into the adjusted digital twin synchronization model to estimate the future power output, thermal management response, and grid compatibility indicators of the energy storage system, generating simulation operation data. Based on the simulation operation data, fault modes, and the operation and maintenance rule base, a multi-objective decision engine is used to generate non-periodic operation and maintenance strategies for adjusting the charging and discharging parameters of the energy storage system and optimizing the power converter control strategy.

[0007] Optionally, the step of using the historical data to train and calibrate the battery physical parameter model and dynamic operation model of the digital twin synchronization model includes: cleaning and standardizing the historical data to obtain a training dataset; using the training dataset, performing deep learning training on the battery's charging and discharging behavior, capacity decay, and internal resistance changes through a long short-term memory network model to generate the battery physical parameter model and dynamic operation model.

[0008] Optionally, the deep learning training of the battery's charging and discharging behavior, capacity decay, and internal resistance change using a long short-term memory network model includes: iteratively training the long short-term memory network model using the training dataset; in the iterative training, using an adaptive learning rate optimization algorithm to dynamically adjust the weight updates of the long short-term memory network model, accelerating convergence and avoiding local optima; and using the iteratively trained model as the battery physical parameter model and dynamic running model.

[0009] Optionally, the quantitative diagnostic indicators include: the residual mean represents the systematic deviation between the digital twin synchronization model and the energy storage system; and the residual standard deviation represents the volatility and instability of the digital twin synchronization model.

[0010] Optionally, adjusting the parameters of the digital twin synchronization model based on the quantitative diagnostic indicators and in conjunction with the fault modes includes: determining that the digital twin synchronization model has a deviation when the deviation between the residual mean and the residual standard deviation exceeds a preset threshold; determining the parameters in the digital twin synchronization model that need to be adjusted based on the deviation direction and degree indicated by the quantitative diagnostic indicators and in conjunction with the fault modes; and using an online parameter optimization algorithm to adaptively fine-tune the parameters that need to be adjusted until the deviation between the prediction result and the real-time data is restored to within the threshold.

[0011] Optionally, the calculation of the future power output, thermal management response, and grid compatibility indicators of the energy storage system includes: inputting the real-time data into an adjusted digital twin synchronization model to simulate various future operating strategies of the energy storage system; calculating the impact of each strategy on the power output, thermal management response, battery life, and economic benefits of the energy storage system; and integrating the simulation results and calculation results to generate simulation operation data, which includes different operation and maintenance strategies and corresponding performance evaluation indicators.

[0012] Optionally, generating the non-periodic operation and maintenance strategy for adjusting the charging and discharging parameters of the energy storage system and optimizing the power converter control strategy includes: configuring a multi-objective decision engine based on the real-time data, weighting factors and priority rules contained in the operation and maintenance rule base, wherein the multi-objective decision engine is used to handle multiple conflicting or interrelated objectives and select a balance point among them; inputting various operation and maintenance strategies and their corresponding performance evaluation indicators contained in the simulation running data into the multi-objective decision engine; using the multi-objective decision engine to weigh the performance evaluation indicators and select a superior operation and maintenance strategy; and converting the operation and maintenance strategy into instructions for adjusting the charging and discharging parameters of the energy storage system and the power converter control strategy to generate the non-periodic operation and maintenance strategy.

[0013] Optionally, the multi-objective decision engine includes: an improved NSGA-III algorithm and fuzzy decision, wherein: the improved NSGA-III algorithm is used to solve the Pareto front of various operation and maintenance strategies in the simulation running data; and the fuzzy decision is used to combine the fault modes with the weight factors and priority rules in the operation and maintenance rule base to select high-quality operation and maintenance strategies from the Pareto front.

[0014] Optionally, configuring the multi-objective decision engine based on the fault modes and the weighting factors and priority rules contained in the operation and maintenance rule base includes: extracting dynamic risks based on the dynamic changes in battery cluster temperature difference, power converter switching transient characteristics, and grid connection point harmonic spectrum in the real-time data; mapping the dynamic risks to dynamic weighting factors and priority rules of the multi-objective decision engine according to the fault modes and the operation and maintenance rule base; and configuring the multi-objective decision engine using the dynamic weighting factors and priority rules to ensure that the generation of operation and maintenance strategies prioritizes responses to dynamic risk events.

[0015] Based on the same inventive concept, this invention also provides a digital twin-based energy storage system operation and maintenance strategy optimization system. The system includes: an information acquisition module for acquiring full-process information of the energy storage power station, including real-time and historical data on battery cluster temperature difference, power converter switching transient characteristics, grid connection point harmonic spectrum, and battery status, as well as fault cases and operation and maintenance records of the energy storage power station; a model building module for establishing a digital twin synchronization model of the energy storage system, using the historical data to train and calibrate the battery physical parameter model and dynamic operation model of the digital twin synchronization model, and converting the fault cases and operation and maintenance records into a fault mode and operation and maintenance rule base; and a diagnostic analysis module for generating current... The system uses a multi-objective decision engine to generate quantitative diagnostic indicators by combining real-time predicted data with residual analysis to calculate the mean and standard deviation of the residuals. A parameter adjustment module adjusts the parameters of the digital twin synchronization model based on the quantitative diagnostic indicators and fault modes, ensuring synchronization between the digital twin synchronization model and the actual operating state of the energy storage system. A simulation calculation module inputs real-time data into the adjusted digital twin synchronization model to calculate the future power output, thermal management response, and grid compatibility indicators of the energy storage system, generating simulation operation data. A strategy generation module uses a multi-objective decision engine based on the simulation operation data, fault modes, and the operation and maintenance rule base to generate non-periodic operation and maintenance strategies for adjusting the charging and discharging parameters of the energy storage system and optimizing the power converter control strategy.

[0016] Compared with the prior art, the present invention has the following advantages:

[0017] 1. This invention improves the predictability and accuracy of energy storage system operation and maintenance. By constructing a digital twin model that is synchronized with the physical entity in real time and introducing residual analysis for dynamic calibration, it can accurately capture subtle state deviations caused by factors such as aging and environmental changes. This enables the system to provide early warnings of potential failure risks from a data perspective, realizing a shift from passive response maintenance to proactive predictive maintenance, and significantly improving the scientific basis and accuracy of operation and maintenance decisions;

[0018] 2. This invention achieves global and adaptive optimization of energy storage system operation and maintenance strategies. Through a multi-objective decision engine, it incorporates multiple conflicting operation and maintenance objectives, such as safety, economy, and reliability, into a unified optimization framework. Furthermore, by dynamically mapping real-time risk events into decision weights, the operation and maintenance strategy can adaptively adjust according to the most pressing needs, finding the globally optimal balance point under complex operating conditions, thus avoiding the limitations of traditional single-objective optimization.

[0019] 3. This invention enhances the safety and economic efficiency of energy storage system operation. By using simulation calculations to predict the future impact of different strategies, high-risk operations that could accelerate equipment aging or cause safety accidents can be effectively avoided. Simultaneously, by generating and executing non-periodic, on-demand operation and maintenance strategies, unnecessary maintenance costs are reduced. Furthermore, through refined charge and discharge control and power converter management, the energy efficiency and overall economic benefits of the energy storage system are maximized while ensuring safety, thus extending the effective service life of the assets.

[0020] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description

[0021] 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 some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a flowchart illustrating an energy storage system operation and maintenance strategy optimization method based on digital twins according to an embodiment of the present invention.

[0023] Figure 2 This is a dynamic calibration graph of the residual mean and residual standard deviation in an embodiment of the present invention.

[0024] Figure 3 This is a strategy trade-off diagram of the multi-objective decision engine in an embodiment of the present invention.

[0025] Figure 4 This is a schematic diagram of the structure of an energy storage system operation and maintenance strategy optimization system based on digital twin according to an embodiment of the present invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] Reference Figure 1 One embodiment of the present invention proposes an optimization method for the operation and maintenance strategy of energy storage system based on digital twin. It adopts an adaptively calibrated digital twin model, which can integrate real-time diagnosis and forward-looking optimization, and significantly improve the operation and maintenance efficiency and reliability of energy storage system.

[0028] The method described in this embodiment specifically includes:

[0029] Acquire full-process information of the energy storage power station, including real-time and historical data of battery cluster temperature difference, power converter switching transient characteristics, grid connection point harmonic spectrum and battery status, as well as fault cases and operation and maintenance records of the energy storage power station.

[0030] Specifically, a multi-dimensional sensor network deployed within the energy storage power station acquires multi-parameter operational data streams in real time. This data stream includes temperature distribution within the battery clusters, transient voltage and current characteristics of the power converter during switching operations, and harmonic spectrum data at the grid connection point. Simultaneously, historical data for these parameters is extracted from the energy storage power station, along with past fault cases and maintenance records.

[0031] Establish a digital twin synchronization model of the energy storage system, and use the historical data to train and calibrate the battery physical parameter model and dynamic operation model of the digital twin synchronization model; convert the fault cases and operation and maintenance records into a fault mode and operation and maintenance rule base.

[0032] Specifically, historical data is used to train and calibrate the battery physical parameter model and dynamic operation model of the digital twin synchronization model. Furthermore, fault cases and maintenance records are transformed into a structured fault mode and maintenance rule base. Historical fault cases and maintenance records of the energy storage power station are cleaned to remove irrelevant information. Key features, such as fault type, occurrence time, relevant sensor data, and manual intervention measures, are extracted from the text descriptions using natural language processing techniques or keyword matching. Using the extracted features, a knowledge graph or rule-based expert system is constructed. "Battery cluster temperature difference continuously exceeding the threshold" can be defined as a "fault mode" and associated with "maintenance rules" such as "reducing the charge / discharge rate" and "enhancing the power of the heat dissipation system." This association can be stored in the form of IF-THEN rules, for example:

[0033] IF diagnostic indicators: 'Increased temperature rise prediction deviation' AND deviation degree > 0.5°C, THEN adjustment parameters: 'Battery equivalent internal resistance' AND adjustment direction == 'Upward adjustment';

[0034] IF diagnostic indicators: 'grid-connected harmonics exceed the standard' AND deviation degree > threshold, THEN adjustment parameters: 'PCS switching timing' AND adjustment direction 'optimization'.

[0035] Based on the digital twin synchronization model and historical data, predictive data for the current moment is generated, and residual analysis is performed in conjunction with real-time data to calculate the mean and standard deviation of the residuals between the two, thereby generating quantitative diagnostic indicators.

[0036] Specifically, in the digital twin synchronization model, historical data is used to generate predicted data that the energy storage system should achieve at the current moment. Then, this predicted data is compared with real-time data, and the mean and standard deviation of the residuals are calculated. These two values ​​together constitute a quantitative diagnostic indicator characterizing the operating status. The mapping between the quantitative diagnostic indicator and model parameter adjustments can be established using regression models or lookup tables from machine learning. For example, a regression model can be trained using historical data, with the inputs being the "mean residual" and "standard deviation of the residual," and the output being the "percentage of parameter value to be adjusted." Alternatively, a lookup table can be created to map different ranges of residual indicators to predefined parameter adjustment amounts. For example, when the mean residual is in the range [0.2, 0.4] and the standard deviation is in the range [0.1, 0.3], the internal resistance parameter is adjusted by +1.5%. When the mean residual is in the range [0.4, 0.6] and the standard deviation is in the range [0.3, 0.5], the internal resistance parameter is adjusted by +3.0%.

[0037] Based on the quantitative diagnostic indicators and in conjunction with the fault modes, adjust the parameters of the digital twin synchronization model to ensure that the digital twin synchronization model remains synchronized with the actual operating state of the energy storage system.

[0038] Specifically, based on the degree of deviation of quantitative diagnostic indicators, it is determined that the digital twin synchronization model has biases. By referring to knowledge in the fault mode and operation and maintenance rule base, the model parameters that need to be adjusted are identified. Subsequently, these parameters are adjusted to ensure that the model's prediction results remain synchronized with the actual operating state of the energy storage system.

[0039] By inputting real-time data into the adjusted digital twin synchronization model, the future power output, thermal management response, and grid compatibility indicators of the energy storage system are calculated, generating simulation operation data.

[0040] Specifically, the latest real-time data is used as the starting condition and input into the adjusted digital twin synchronization model. The model calculates the possible operating states of the energy storage system over a period of time, including power output, thermal management effectiveness, and grid compatibility indicators. These calculation results are then integrated to generate simulation operation data for strategy evaluation.

[0041] Based on simulation data, fault modes, and operation and maintenance rules, a multi-objective decision engine is used to generate non-periodic operation and maintenance strategies for adjusting the charging and discharging parameters of the energy storage system and optimizing the power converter control strategy.

[0042] Specifically, simulation data is input into a multi-objective decision engine. Simultaneously, the decision engine is configured by combining fault modes with weighting factors and priority rules contained in the operation and maintenance rule base. The decision engine weighs multiple performance evaluation metrics and selects the optimal operation and maintenance strategy. This strategy is then translated into specific charging and discharging parameters and power converter control commands, generating non-periodic operation and maintenance strategies and issuing them for execution.

[0043] An operation and maintenance strategy optimization method integrating real-time data perception, digital twin model self-calibration, and a multi-objective decision engine is adopted. This method enables a shift from passive response to proactive prediction in operation and maintenance through accurate prediction of system operating status and real-time residual analysis. It incorporates multiple interrelated operation and maintenance objectives, such as safety, economy, and efficiency, into a unified decision-making framework, significantly improving the operation and maintenance efficiency, operational reliability, and overall economic benefits of energy storage systems.

[0044] Optionally, training and calibrating the battery physical parameter model and dynamic operation model of the digital twin synchronization model using the historical data includes:

[0045] The historical data is cleaned and standardized to obtain a training dataset;

[0046] Specifically, the collected historical data may contain missing values, outliers, and parameters with different units of measurement. Preprocessing this data involves first filling or removing missing data points through interpolation or deletion. Then, statistical methods are used to identify and smooth outliers. Finally, the processed data is standardized to map all parameter values ​​to a uniform range, eliminating the impact of unit of measurement differences on model training. This standardization process can be performed using the following formula:

[0047] ,

[0048] in, For standardized data, This is the original data. The average value of the data. denoted as the standard deviation of the data.

[0049] Using the training dataset, a long short-term memory network model is used to train the battery's charging and discharging behavior, capacity decay, and internal resistance changes through deep learning, generating a battery physical parameter model and a dynamic operation model.

[0050] Specifically, the preprocessed training dataset is input into a Long Short-Term Memory (LSTM) network model for deep learning training. The LSTM model utilizes gating mechanisms to selectively remember and forget information, making it well-suited for processing time-series data from energy storage system operation. The goal of the training process is to enable the LSTM model to learn and simulate the battery's charging and discharging behavior, capacity decay over time, and internal resistance changes with state, ultimately generating a battery physical parameter model and dynamic operation model that accurately reflects the battery's characteristics. During forward and backward propagation in training, the LSTM model involves a series of gating unit operations. The calculation formulas for the forget gate, input gate, and output gate are as follows:

[0051] ,

[0052] ,

[0053] ,

[0054] in, , Forget gate, input gate, and output gate respectively Output at any moment It is the sigmoid activation function. and For the weight matrix and bias vector, This is the hidden state from the previous moment. This represents the input vector at the current moment. Through these operations, the LSTM model is able to capture and learn the complex nonlinear characteristics of the energy storage system in the time dimension.

[0055] Optionally, the deep learning training of the battery's charging and discharging behavior, capacity decay, and internal resistance changes using a long short-term memory network model includes:

[0056] The Long Short-Term Memory network model is iteratively trained using the training dataset.

[0057] Specifically, the cleaned and standardized training dataset is divided into a training set, a validation set, and a test set. The LSTM model is iteratively trained using the training set through forward and backward propagation to minimize the loss function. During training, the model's performance on the validation set is used to determine if it is overfitting, and the training hyperparameters are adjusted accordingly to ensure good generalization ability.

[0058] In the iterative training, an adaptive learning rate optimization algorithm is used to dynamically adjust the weight updates of the long short-term memory network model, accelerate convergence, and avoid local optima.

[0059] Specifically, the Adam optimization algorithm is used to update the weights of the LSTM model, accelerating model convergence and avoiding getting trapped in local optima. The Adam algorithm combines the advantages of RMSprop and Momentum optimization algorithms, calculating an adaptive learning rate for each weight parameter. This allows for more efficient training of the model in sparse gradient problems. In each training iteration, the weight update formula for the Adam algorithm is:

[0060] ,

[0061] in, for Weight parameters at time points, For learning rate, for The gradient first moment estimate after time deviation correction. for The gradient second moment estimate after time deviation correction. To prevent extremely small constants with a denominator of zero.

[0062] The model trained iteratively is used as the battery physical parameter model and dynamic operation model.

[0063] Specifically, training stops when the LSTM model achieves a preset accuracy on the test set and the loss function converges. At this point, the trained model is considered the final battery physical parameter model and dynamic operation model. These two models are structurally unified, but they respectively focus on simulating the physical characteristics of the battery and the overall dynamic operation of the system, together forming the core of the digital twin synchronization model.

[0064] Optionally, the quantitative diagnostic indicators include:

[0065] The residual mean represents the systematic deviation between the digital twin synchronization model and the energy storage system;

[0066] Specifically, the mean residual can be obtained by calculating the residual between the predicted data of the digital twin synchronization model and the actual operating data of the energy storage system at the current moment. This mean residual represents the deviation between the model's prediction and the actual situation. If the mean residual continuously deviates from zero, it indicates that the parameters of the digital twin model may no longer accurately reflect the current state of the energy storage system, and there is a systematic bias. For example, as the battery ages, its internal resistance increases. If the model parameters are not updated synchronously, the mean residual between the predicted voltage and the actual voltage will remain positive.

[0067] The residual standard deviation characterizes the volatility and instability of the digital twin synchronization model.

[0068] Specifically, the residual standard deviation can be obtained by calculating the degree of residual fluctuation between the predicted data of the digital twin synchronization model and the actual operating data of the energy storage system. This residual standard deviation represents the random fluctuation between the model's prediction and the actual situation. If the residual standard deviation fluctuates within a normal range, it indicates that the model's prediction is stable. However, if the residual standard deviation suddenly increases, it may indicate that a sudden or intermittent anomaly has occurred within the energy storage system, such as a power converter experiencing a transient overshoot under specific operating conditions, leading to increased data volatility.

[0069] Optionally, adjusting the parameters of the digital twin synchronization model based on the quantitative diagnostic indicators and in conjunction with the fault mode includes:

[0070] When the deviation between the residual mean and the residual standard deviation exceeds a preset threshold, it is determined that there is a deviation in the digital twin synchronization model;

[0071] Specifically, by statistically analyzing historical data of the energy storage system under normal operating conditions, the normal fluctuation range of the residual mean and residual standard deviation can be determined. This fluctuation range is used as a threshold. If the currently calculated residual mean or residual standard deviation exceeds this threshold range, it is determined that there is a deviation between the digital twin synchronization model and the actual operating state of the energy storage system. For example, if the residual mean between the model's predicted battery voltage and the actual voltage remains positive for a long period, it indicates that the model's internal resistance parameter may be lower than the actual value. Figure 2 As shown, the curves illustrating the changes in the mean and standard deviation of residuals over time when performing residual analysis on the operating status of an energy storage system, intuitively reflect the adaptive adjustment process of the model parameters. When the residuals exceed the threshold, the model parameters are adjusted, and the residuals subsequently return to the normal range.

[0072] Based on the direction and degree of deviation indicated by the quantitative diagnostic indicators, and in conjunction with the fault mode, determine the parameters that need to be adjusted in the digital twin synchronization model;

[0073] Specifically, deviations in quantitative diagnostic indicators are used as input and matched against the fault mode and operation and maintenance rule base. This rule base contains rich historical data and expert knowledge, enabling it to correlate specific deviation patterns with specific model parameter deviations. For example, if a systematic deviation is determined in the model, the rule base may indicate the need to correct the capacity degradation parameter in the battery physical parameter model.

[0074] Using an online parameter optimization algorithm, the parameters that need to be adjusted are adaptively fine-tuned until the deviation between the prediction result and the real-time data is restored to within the threshold.

[0075] Specifically, online parameter optimization algorithms based on Kalman filtering or particle swarm optimization are used to adjust the determined model parameters in real time with small amplitude. The purpose of the adjustment is to minimize the deviation between the predicted data and the real-time data, so that the model can quickly converge to parameter values ​​that match the actual operating state of the energy storage system. This adaptive fine-tuning process continues until the residual mean and residual standard deviation recover to within the threshold, ensuring that the accuracy of the digital twin synchronization model remains online. The specific implementation of the Kalman filtering (EFK) algorithm as the online parameter optimization algorithm consists of three parts: the construction of the state equation and observation equation, the setting and updating of the covariance matrix, and the achievement of the objective.

[0076] Construction of state and observation equations: The model parameters that need to be adjusted, such as the battery equivalent internal resistance and capacity decay coefficient, are used as the state vectors of the EKF. The state equations of the EKF are based on the dynamic operating equations of the digital twin synchronization model, describing the evolution of these parameters over time. The observation equations use real-time data from the energy storage system, such as voltage, current, and temperature, as observation vectors. This equation correlates the model parameters with measurable real-time data. State equation: ,in, It is the control input from the previous moment. This is process noise. Observation equation: ,in, It is observation noise. (Function) This corresponds to the predictive function of a digital twin model, which is to predict the output based on the model parameters.

[0077] Setting and updating the covariance matrix: In order to capture the uncertainty of parameter changes, it is necessary to set the process noise covariance matrix. and observation noise covariance matrix .matrix The value can be set based on the typical drift rate of the parameter in historical data, matrix The determination is based on the sensor's measurement accuracy. The EKF algorithm dynamically updates the state covariance matrix in each iteration. This is to reflect the uncertainty in parameter estimation.

[0078] Objective: The algorithm iterates by continuously comparing real-time data with model predictions and using the residuals to correct the state vector. This process continues until the deviation between the prediction and real-time data falls within a preset threshold. When the mean and standard deviation of the residuals are within the threshold, the Kalman filter gain decreases accordingly, and the parameter adjustment range becomes smaller, thus achieving precise adaptive fine-tuning.

[0079] Optionally, the projected future power output, thermal management response, and grid compatibility indicators of the energy storage system include:

[0080] The real-time data is input into the adjusted digital twin synchronization model to simulate various future operating strategies of the energy storage system;

[0081] Specifically, a parameter-adjusted digital twin synchronization model is used as the basis for calculations, and the latest real-time data is used as the starting state to simulate the operation of the energy storage system over a period of time. This simulation process considers various operation and maintenance strategies, such as charging and discharging power, duration, and time periods under different grid load and electricity price fluctuation scenarios, to evaluate the potential effectiveness of various strategies.

[0082] Based on the aforementioned operating strategies, the impact of each strategy on the energy output, thermal management response, battery life, and economic benefits of the energy storage system is calculated.

[0083] Specifically, when simulating multiple operating strategies, key performance indicators (KPIs) for each strategy are calculated simultaneously. Energy output indicators include total charge / discharge capacity and power curves. Thermal management response indicators include the battery cluster's temperature rise rate and temperature difference variation. Battery life indicators are evaluated by simulating battery aging under different charge / discharge depths and rates. Economic efficiency indicators are measured by calculating energy trading revenue and system loss costs.

[0084] The simulation results and the extrapolation results are integrated to generate simulation operation data, which includes different operation and maintenance strategies and corresponding performance evaluation indicators.

[0085] Specifically, all simulated and extrapolated performance metrics are correlated with corresponding operational strategies. This data is structured into simulation runtime data, forming a multi-dimensional decision matrix. This simulation runtime data demonstrates the advantages and disadvantages of each strategy.

[0086] Optionally, the non-periodic operation and maintenance strategy for adjusting the charging and discharging parameters of the energy storage system and optimizing the power converter control strategy includes:

[0087] Based on the real-time data, the weight factors and priority rules contained in the operation and maintenance rule base, the multi-objective decision engine is configured, wherein the multi-objective decision engine is used to process multiple conflicting or related objectives and select a balance point from them.

[0088] Specifically, the multi-objective decision engine is an algorithmic decision-making system whose decision logic is configured through weighting factors and priority rules. Weighting factors are used to allocate importance among multiple objectives such as economic benefits, battery life, and system safety. Priority rules are adjusted based on dynamic risk events in real-time data. For example, when the temperature difference between battery clusters suddenly increases in real-time data, the priority rules dynamically increase the weight of safety, ensuring that the decision engine prioritizes safety-enhancing operational strategies.

[0089] The various operation and maintenance strategies and their corresponding performance evaluation indicators contained in the simulation running data are input into the multi-objective decision engine;

[0090] Specifically, simulation data is used as input to the multi-objective decision engine. This data includes multiple alternative operation and maintenance strategies, each accompanied by performance evaluation metrics such as power output, thermal management response, battery life, and economic benefits over a future period. These metrics form the basis for the decision engine's trade-offs and selection.

[0091] The multi-objective decision engine is used to weigh performance evaluation metrics and select the best operation and maintenance strategy.

[0092] Specifically, the multi-objective decision engine performs a weighted sum of performance metrics for multiple objectives based on pre-configured weighting factors to achieve a trade-off between objectives. Assume the objective function of the multi-objective decision engine is... This function is designed to achieve economic objectives. Technical objectives and environmental goals The balance. The objective function can be expressed as:

[0093] ,

[0094] in, As decision variables, , , These are dynamic weighting factors, adjusted based on real-time operational status and the rule base. Ultimately, the decision engine will select the factors that optimize the objective function. Optimal operation and maintenance strategy. For example... Figure 3As shown, the performance evaluation results of the multi-objective decision engine when weighing different operation and maintenance strategies are presented. The color intensity in the figure represents the performance of different strategies in multiple performance indicators such as economic benefits, battery life, and system stability; the darker the color, the better the performance.

[0095] The operation and maintenance strategy is converted into instructions for adjusting the charging and discharging parameters of the energy storage system and the control strategy of the power converter, thereby generating an unscheduled operation and maintenance strategy.

[0096] Specifically, the optimal operation and maintenance strategies selected by the multi-objective decision engine are transformed into executable control commands. This includes decomposing the charging and discharging strategy into specific charging and discharging power, duration, and time periods, and issuing these to the energy storage system's controller. Simultaneously, the power converter control strategy is transformed into parameters for adjusting its equivalent output impedance to optimize grid compatibility.

[0097] Optionally, the multi-objective decision engine includes: an improved NSGA-III algorithm and fuzzy decision, wherein:

[0098] The improved NSGA-III algorithm is used to solve the Pareto front of various operation and maintenance strategies in the simulation running data;

[0099] Specifically, the traditional NSGA-III algorithm uses static reference points to guide population convergence. However, the operational objectives of energy storage systems change dynamically with real-time risk events. A dynamic reference point generation mechanism is introduced, adjusting the position and distribution of reference points in real time based on dynamic weight factors and priority rules mapped from failure modes to the operation and maintenance rule base. For example, when the system faces the risk of thermal runaway, the weight of objectives related to thermal management response in the multi-objective space increases, guiding the algorithm to find Pareto optimal solutions in directions with better thermal management performance, ensuring a trade-off while prioritizing safety. In the selection operation of the standard NSGA-III algorithm, selection is mainly based on non-dominated ordering and reference point correlation. A risk preference factor is integrated into the selection operator. Specifically, when facing high-risk events, the selection operation prioritizes and enhances solutions that can effectively reduce risk, thus considering risk factors as an implicit constraint or preference during algorithm iteration. To accelerate convergence and improve algorithm efficiency, an elite library based on historical best solutions is introduced in each iteration. This elite pool is periodically mixed with the current population to ensure that previously discovered high-quality solutions are not lost during algorithm evolution. This mechanism significantly reduces the time required to obtain high-quality solutions, especially when dealing with dynamic risks requiring rapid responses, such as grid harmonic exceedance events.

[0100] The improved NSGA-III algorithm is the core component of the multi-objective decision engine, used to optimize various operational strategies in simulation data. In each iteration, the improved NSGA-III algorithm updates the population in the following way:

[0101] ,

[0102] in, For the current population, For the newly generated solution set, This is the selection operator. This operator selects the solution with the highest non-dominant level and best diversity from the merged population to generate the next generation. .

[0103] Fuzzy decision-making is used to combine the failure modes with weighting factors and priority rules in the operation and maintenance rule base to select high-quality operation and maintenance strategies from the Pareto front.

[0104] Specifically, the fuzzy decision-making module receives the Pareto front solution set obtained by the improved NSGA-III algorithm and evaluates each solution in the Pareto front by combining the weight factors and priority rules in the failure mode and operation and maintenance rule base. Fuzzy decision-making constructs fuzzy membership functions to transform the performance of each solution on different performance indicators into fuzzy values, and then uses fuzzy inference rules for comprehensive evaluation. For example, for a decision problem involving three objectives—economic efficiency, safety, and lifespan—the comprehensive evaluation value of its fuzzy decision can be calculated as follows:

[0105] ,

[0106] in, To solve The overall evaluation value, , , solutions respectively The fuzzy membership degree is considered in terms of economy, security, and lifespan. Ultimately, the solution with the highest comprehensive evaluation value is selected as the superior operation and maintenance strategy.

[0107] Optionally, configuring the multi-objective decision engine based on the fault modes and the weighting factors and priority rules contained in the operation and maintenance rule base includes:

[0108] Dynamic risks are extracted based on the dynamic changes in battery cluster temperature difference, power converter switching transient characteristics, and grid connection point harmonic spectrum in the real-time data.

[0109] Specifically, by detecting abnormal fluctuations and analyzing trends in real-time data streams, dynamic events can be identified. For example, when the temperature difference of the battery cluster exceeds the normal rate of change within a short period of time, or when abnormal overshoot occurs in the switching transient characteristics of the power converter, the system will mark it as a dynamic event. These events are converted into a dynamic risk value to quantify the risk level currently faced by the system.

[0110] The dynamic risks are mapped to dynamic weighting factors and priority rules of a multi-objective decision engine based on failure modes and operation and maintenance rule base.

[0111] Specifically, the extracted dynamic risk values ​​are used as input and matched against the fault modes and the operation and maintenance rule base. The rule base stores the mapping relationship between different risk levels and the weight factors and priority rules in the multi-objective decision engine. When the dynamic risk value is low, the rule base maps weight factors that are biased towards economic benefits; when the dynamic risk value is high, the rule base maps weight factors and priority rules that are biased towards system security.

[0112] By utilizing the aforementioned dynamic weighting factors and priority rules, a multi-objective decision engine is configured to ensure that the generation of operation and maintenance strategies prioritizes responses to dynamic risk events.

[0113] Specifically, the dynamic weighting factors and priority rules obtained from the mapping are loaded into the multi-objective decision engine in real time, serving as constraints and preferences in its decision-making process. This ensures that when generating operation and maintenance strategies, the decision engine can dynamically adjust its trade-off criteria based on the dynamic risks currently faced by the system, prioritizing strategies that can effectively respond to and avoid risks, thereby achieving the adaptability of the operation and maintenance strategy.

[0114] Based on the same inventive concept, such as Figure 4 As shown, the present invention also provides a digital twin-based energy storage system operation and maintenance strategy optimization system, the system comprising:

[0115] Information acquisition module: used to acquire full-process information of energy storage power station. The full-process information includes real-time and historical data of battery cluster temperature difference, power converter switching transient characteristics, grid connection point harmonic spectrum and battery status, as well as fault cases and operation and maintenance records of energy storage power station.

[0116] Model building module: used to establish a digital twin synchronization model of the energy storage system, use the historical data to train and calibrate the battery physical parameter model and dynamic operation model of the digital twin synchronization model, and convert the fault cases and operation and maintenance records into a fault mode and operation and maintenance rule base;

[0117] Diagnostic analysis module: Based on the digital twin synchronization model and historical data, it generates predicted data for the current moment, and performs residual analysis in combination with real-time data to calculate the mean and standard deviation of the residuals between the two, and generates quantitative diagnostic indicators.

[0118] Parameter adjustment module: used to adjust the parameters of the digital twin synchronization model according to the quantitative diagnostic indicators and the fault mode, so as to ensure that the digital twin synchronization model keeps synchronized with the actual operating state of the energy storage system;

[0119] Simulation and extrapolation module: Used to input real-time data into the adjusted digital twin synchronization model to extrapolate the future power output, thermal management response and grid compatibility indicators of the energy storage system, and generate simulation operation data;

[0120] Strategy generation module: Based on simulation data, fault modes, and operation and maintenance rule base, this module uses a multi-objective decision engine to generate non-periodic operation and maintenance strategies for adjusting the charging and discharging parameters of the energy storage system and optimizing the power converter control strategy.

[0121] Example 1

[0122] To verify the feasibility of this invention in practice, it was applied to an energy storage power station. This energy storage power station is responsible for peak shaving and valley filling and frequency regulation of the power grid, and its operational stability and economic benefits are crucial to power grid security. Traditional operation and maintenance models mainly rely on fixed maintenance cycles and responses after failures, which are difficult to cope with inconsistent battery aging, thermal runaway risks, and increasingly stringent power grid harmonic regulations, resulting in high operation and maintenance costs and reduced system availability.

[0123] In this embodiment, the energy storage power station deployed the energy storage system operation and maintenance strategy optimization system based on digital twin proposed in this invention. Through the information acquisition module, real-time information on the entire process, including battery cluster temperature difference, power converter switching transient characteristics, and grid connection point harmonic spectrum, was collected. The model building module utilized the power station's historical operating data from the past 12 months, and trained and calibrated a dedicated digital twin synchronization model for the power station using a time series learning model and an adaptive learning rate optimization algorithm, establishing a fault mode and operation and maintenance rule base. During a 6-month trial operation, the health status of the energy storage system was diagnosed in real-time, simulated, and the operation and maintenance strategy was dynamically optimized.

[0124] During the trial operation, the diagnostic analysis module continuously performed residual analysis on the predicted data and real-time data of the digital twin synchronization model. For example, due to accelerated aging of some battery modules, the average residual between the actual temperature rise and the model prediction was continuously detected to exceed the preset threshold of 0.5℃. The parameter adjustment module was triggered, and based on the fault mode library, it diagnosed that the battery equivalent internal resistance model parameters were inaccurate. It immediately started the online parameter optimization algorithm to fine-tune the parameters, so that the residual was restored to the threshold range within 20 minutes, completing the dynamic synchronization of the model.

[0125] The multi-objective decision engine of this invention demonstrated its superiority in responding to sudden risks. During a high-rate charging process, the temperature difference of battery cluster No. 3 exceeded the safety threshold of 5°C. The dynamic weight update mechanism was immediately triggered, increasing the weight factor of the "thermal management response" objective by 50% based on the fault mode and the operation and maintenance rule base. After analyzing various alternative strategies on the Pareto front, the multi-objective decision engine ultimately selected an operation and maintenance strategy that prioritized reducing the charging current and enhancing the power of the heat dissipation system at the expense of some charging efficiency. This strategy was converted into specific charging and discharging parameter adjustment instructions and sent to the BMS, successfully reducing the temperature difference to a safe range within 30 minutes and effectively avoiding the risk of thermal runaway.

[0126] In addition, due to the startup of high-power equipment in a nearby industrial area, the total harmonic distortion rate at the grid connection point temporarily exceeded the standard. Upon detecting this event, the weighting factor of the "grid compatibility" target was immediately increased by 30%. The non-periodic maintenance strategy output by the strategy generation module was correspondingly converted into power converter switching timing optimization instructions. By adjusting the PWM carrier frequency, a specific harmonic cancellation mode was activated, enabling the grid-connected harmonics to return to within the grid specification requirements within 10 minutes.

[0127] Through the application of this invention, the operation and maintenance mode of this energy storage power station has been transformed from passive response to proactive prediction. Trial operation data shows that compared with the same period last year using the traditional operation and maintenance mode, the average temperature difference of the battery cluster has decreased by 15%, the grid connection harmonic non-compliance events have decreased by 70%, the predicted annualized battery capacity degradation rate has decreased by about 5%, and the overall operating efficiency has improved by 3%.

[0128] Table 1. Energy Storage System Condition Diagnosis and Model Synchronization Data Table

[0129]

[0130] Table 2 Data Table of Multi-Objective Decision Making and Dynamic Weight Adjustment for Energy Storage Systems

[0131]

[0132] Table 3 Performance Comparison Before and After Energy Storage System Operation and Maintenance Strategy Optimization

[0133]

[0134] As can be seen from the data in Tables 1 to 3 above, this invention has achieved significant results in the practical application of energy storage power stations. The data in Table 1 confirms the effectiveness of the diagnostic analysis and parameter adjustment modules, which can detect deviations between the model and the physical entity in real time and quickly complete adaptive calibration, ensuring the high fidelity of the digital twin model. Table 2 clearly shows the dynamic decision-making process of the multi-objective decision engine when facing specific risk events. By adjusting weight factors in real time, it achieves risk-driven and intelligent strategy selection, ensuring the pertinence and timeliness of operation and maintenance decisions. The performance comparison results before and after optimization in Table 3 intuitively demonstrate the comprehensive advantages of this invention in improving system safety, reliability, and economy.

[0135] It should be noted that the electrical connections between the various units described above do not necessarily represent direct or indirect connections. Any indirect connection method can be applied to the embodiments of the present invention as long as it achieves the purpose of the present invention. The above descriptions are merely exemplary embodiments of the present invention and should not be construed as limiting the scope of the present invention.

[0136] All equivalent changes and modifications made in accordance with the teachings of this invention are still within the scope of this invention. Those skilled in the art will readily conceive of other embodiments of this invention upon considering the specification and the disclosure of practical truth. This application is intended to cover any variations, uses, or adaptations of this invention that follow the general principles of this invention and include common knowledge or conventional techniques in the art not described herein.

Claims

1. A method for optimizing the operation and maintenance strategy of an energy storage system based on digital twins, characterized in that, The method includes: Acquire full-process information of the energy storage power station, including real-time and historical data of battery cluster temperature difference, power converter switching transient characteristics, grid connection point harmonic spectrum and battery status, as well as fault cases and operation and maintenance records of the energy storage power station. Establish a digital twin synchronization model of the energy storage system, and use the historical data to train and calibrate the battery physical parameter model and dynamic operation model of the digital twin synchronization model; convert the fault cases and operation and maintenance records into a fault mode and operation and maintenance rule base. Based on the digital twin synchronization model and historical data, predictive data for the current moment is generated, and residual analysis is performed in conjunction with real-time data to calculate the mean and standard deviation of the residuals between the two, thereby generating quantitative diagnostic indicators. Based on the quantitative diagnostic indicators and in conjunction with the fault modes, adjust the parameters of the digital twin synchronization model to ensure that the digital twin synchronization model remains synchronized with the actual operating state of the energy storage system. By inputting real-time data into the adjusted digital twin synchronization model, the future power output, thermal management response, and grid compatibility indicators of the energy storage system are calculated, generating simulation operation data. Based on simulation data, fault modes, and operation and maintenance rules, a multi-objective decision engine is used to generate non-periodic operation and maintenance strategies for adjusting the charging and discharging parameters of the energy storage system and optimizing the power converter control strategy.

2. The method for optimizing the operation and maintenance strategy of an energy storage system based on digital twins according to claim 1, characterized in that, The battery physical parameter model and dynamic operation model of the digital twin synchronization model trained and calibrated using the historical data include: The historical data is cleaned and standardized to obtain a training dataset; Using the training dataset, a long short-term memory network model is used to train the battery's charging and discharging behavior, capacity decay, and internal resistance changes through deep learning, generating a battery physical parameter model and a dynamic operation model.

3. The method for optimizing the operation and maintenance strategy of an energy storage system based on digital twins according to claim 2, characterized in that, The deep learning training of the battery's charging and discharging behavior, capacity decay, and internal resistance changes using a long short-term memory network model includes: The Long Short-Term Memory network model is iteratively trained using the training dataset. In the iterative training, an adaptive learning rate optimization algorithm is used to dynamically adjust the weight updates of the long short-term memory network model, accelerate convergence, and avoid local optima. The model trained iteratively is used as the battery physical parameter model and dynamic operation model.

4. The method for optimizing the operation and maintenance strategy of an energy storage system based on digital twins according to claim 1, characterized in that, The quantitative diagnostic indicators include: The residual mean represents the systematic deviation between the digital twin synchronization model and the energy storage system; The residual standard deviation characterizes the volatility and instability of the digital twin synchronization model.

5. The method for optimizing the operation and maintenance strategy of an energy storage system based on digital twins according to claim 1, characterized in that, The step of adjusting the parameters of the digital twin synchronization model based on the quantitative diagnostic indicators and in conjunction with the fault mode includes: When the deviation between the residual mean and the residual standard deviation exceeds a preset threshold, it is determined that there is a deviation in the digital twin synchronization model; Based on the direction and degree of deviation indicated by the quantitative diagnostic indicators, and in conjunction with the fault mode, determine the parameters that need to be adjusted in the digital twin synchronization model; Using an online parameter optimization algorithm, the parameters that need to be adjusted are adaptively fine-tuned until the deviation between the prediction result and the real-time data is restored to within the threshold.

6. The method for optimizing the operation and maintenance strategy of an energy storage system based on digital twins according to claim 1, characterized in that, The projected future power output, thermal management response, and grid compatibility indicators of the energy storage system include: The real-time data is input into the adjusted digital twin synchronization model to simulate various future operating strategies of the energy storage system; Based on the aforementioned operating strategies, the impact of each strategy on the energy output, thermal management response, battery life, and economic benefits of the energy storage system is calculated. The simulation results and the calculated results are integrated to generate simulation operation data, which includes different operation and maintenance strategies and corresponding performance evaluation indicators.

7. The method for optimizing the operation and maintenance strategy of an energy storage system based on digital twins according to claim 1, characterized in that, The non-periodic operation and maintenance strategy for adjusting the charging and discharging parameters of the energy storage system and optimizing the power converter control strategy includes: Based on the real-time data, the weight factors and priority rules contained in the operation and maintenance rule base, the multi-objective decision engine is configured, wherein the multi-objective decision engine is used to process multiple conflicting or related objectives and select a balance point from them. The various operation and maintenance strategies and their corresponding performance evaluation indicators contained in the simulation running data are input into the multi-objective decision engine; The multi-objective decision engine is used to weigh performance evaluation metrics and select the best operation and maintenance strategy. The operation and maintenance strategy is converted into instructions for adjusting the charging and discharging parameters of the energy storage system and the control strategy of the power converter, thereby generating an unscheduled operation and maintenance strategy.

8. The method for optimizing the operation and maintenance strategy of an energy storage system based on digital twins according to claim 1, characterized in that, The multi-objective decision engine includes: an improved NSGA-III algorithm and fuzzy decision-making, wherein: The improved NSGA-III algorithm is used to solve the Pareto front of various operation and maintenance strategies in the simulation running data; Fuzzy decision-making is used to combine the failure modes with weighting factors and priority rules in the operation and maintenance rule base to select high-quality operation and maintenance strategies from the Pareto front.

9. The method for optimizing the operation and maintenance strategy of an energy storage system based on digital twins according to claim 7, characterized in that, The configuration of the multi-objective decision engine based on the fault modes and the weight factors and priority rules contained in the operation and maintenance rule base includes: Dynamic risks are extracted based on the dynamic changes in battery cluster temperature difference, power converter switching transient characteristics, and grid connection point harmonic spectrum in the real-time data. The dynamic risks are mapped to dynamic weighting factors and priority rules of a multi-objective decision engine based on failure modes and operation and maintenance rule base. By utilizing the aforementioned dynamic weighting factors and priority rules, a multi-objective decision engine is configured to ensure that the generation of operation and maintenance strategies prioritizes responses to dynamic risk events.

10. A digital twin-based energy storage system operation and maintenance strategy optimization system, characterized in that, The system is used in the method for optimizing the operation and maintenance strategy of an energy storage system based on digital twins as described in any one of claims 1-9, the system comprising: Information acquisition module: used to acquire full-process information of energy storage power station. The full-process information includes real-time and historical data of battery cluster temperature difference, power converter switching transient characteristics, grid connection point harmonic spectrum and battery status, as well as fault cases and operation and maintenance records of energy storage power station. Model building module: used to establish a digital twin synchronization model of the energy storage system, use the historical data to train and calibrate the battery physical parameter model and dynamic operation model of the digital twin synchronization model, and convert the fault cases and operation and maintenance records into a fault mode and operation and maintenance rule base; Diagnostic analysis module: Based on the digital twin synchronization model and historical data, it generates predicted data for the current moment, and performs residual analysis in combination with real-time data to calculate the mean and standard deviation of the residuals between the two, and generates quantitative diagnostic indicators. Parameter adjustment module: used to adjust the parameters of the digital twin synchronization model according to the quantitative diagnostic indicators and in combination with the fault mode, so as to ensure that the digital twin synchronization model keeps synchronized with the actual operating state of the energy storage system; Simulation and extrapolation module: Used to input real-time data into the adjusted digital twin synchronization model to extrapolate the future power output, thermal management response and grid compatibility indicators of the energy storage system, and generate simulation operation data; Strategy generation module: Based on simulation data, fault modes, and operation and maintenance rule base, this module uses a multi-objective decision engine to generate non-periodic operation and maintenance strategies for adjusting the charging and discharging parameters of the energy storage system and optimizing the power converter control strategy.

Images