Energy storage system management method and device, computer equipment and readable storage medium

Abstract

The invention relates to an energy storage system management method and device, computer equipment and a readable storage medium. The method comprises the following steps: acquiring an energy storage state data set of an energy storage system; performing prediction processing on the energy storage state data set based on a linear model to obtain a predicted value of each battery state parameter index, and determining a target jump state of the energy storage system in the next management period according to a state conversion rule and each predicted value; updating the energy storage state data set according to the predicted values to obtain an updated energy storage state data set, and predicting target hopping time of the energy storage system hopping to a target hopping state according to the nonlinear model and the updated energy storage state data set; the target jump state and the target jump time are used for managing the energy storage system. By adopting the method, the management efficiency of the energy storage system can be improved.

Description

Technical Field

[0001] The present application relates to the field of battery technology, and in particular to an energy storage system management method, apparatus, computer equipment, computer-readable storage medium, and computer program product. Background Art

[0002] With the large-scale grid connection of new energy and the growth of user-side energy storage demand, as well as the gradual completion and structural complexity of energy storage systems, accurate and real-time prediction of the status of energy storage systems has become a key issue in the battery field.

[0003] Traditionally, energy storage system management requires extracting indirect lifespan feature sets from a NASA dataset (a battery dataset). These indirect lifespan feature sets include constant voltage rise and fall charging time, constant voltage drop discharge time, isochronous charging voltage rise and fall, and constant voltage drop discharge voltage drop. Based on these indirect lifespan feature sets and the Extreme Learning Machine (ELM) model, the remaining useful life (RUL) of the energy storage system is predicted.

[0004] However, conventional technologies can only predict the RUL of an energy storage system, and cannot accurately manage the energy storage system based solely on the RUL. Therefore, current energy storage system management methods have low accuracy. Summary of the Invention

[0005] Based on this, it is necessary to provide an energy storage system management method, device, computer equipment, computer-readable storage medium and computer program product to address the above technical problems.

[0006] In a first aspect, the present application provides an energy storage system management method, comprising:

[0007] Obtaining an energy storage status data set of an energy storage system;

[0008] Performing prediction processing on the energy storage state data set based on a linear model to obtain predicted values ​​of each battery state parameter indicator, and determining a target transition state of the energy storage system in the next management cycle according to a state transition rule and each predicted value;

[0009] The energy storage status data set is updated according to each of the predicted values ​​to obtain an updated energy storage status data set, and a target transition time for the energy storage system to transition to the target transition state is predicted based on a nonlinear model and the updated energy storage status data set; the target transition state and the target transition time are used to manage the energy storage system.

[0010] In one embodiment, obtaining an energy storage status dataset of the energy storage system includes:

[0011] Obtain the initial scheduling plan dataset, initial battery system dataset, initial equipment status telesignaling dataset, and initial operation log dataset of the energy storage system to obtain the initial energy storage status dataset;

[0012] The initial energy storage state data set is integrated, and the integrated initial energy storage state data set is cleaned to obtain an energy storage state data set.

[0013] In one embodiment, before performing prediction processing on the energy storage status data set based on the linear model to obtain the predicted value of each battery status parameter indicator, the method further includes:

[0014] Acquire a sample data set; the sample data set includes each sample data subset;

[0015] Training an initial linear model and an initial nonlinear model based on each of the sample data subsets until the initial linear model and the initial nonlinear model meet a preset training stop condition;

[0016] The initial linear model is determined as a linear model, and the initial nonlinear model is determined as a nonlinear model.

[0017] In one embodiment, obtaining a sample data set includes:

[0018] Acquire an initial sample data set; the initial sample data set includes each initial sample data subset;

[0019] Obtaining an operating state definition set and a historical state data set of the energy storage system, and performing statistical analysis on the historical state data set and the operating state definition set to obtain a state transition rule;

[0020] Determining a sample state of each initial sample data subset based on the state transition rule, and updating the initial sample data subset according to the sample state to obtain a sample data subset;

[0021] The sample data subsets are combined to obtain a sample data set.

[0022] In one embodiment, the operating state definition set includes definition data of normal charging state, definition data of charging fault state, definition data of charging recovery state, definition data of normal discharging state, definition data of discharge fault state, definition data of discharge recovery state and definition data of standby state.

[0023] In one embodiment, the predictive processing of the energy storage state data set based on the linear model to obtain the predicted value of each battery state parameter indicator includes:

[0024] Determine a state parameter feature set, a hysteresis state parameter feature set and an accident feature of the energy storage system according to the energy storage state data set;

[0025] The state parameter feature set, the hysteresis state parameter feature set and the accident feature are input into a linear model, and the state parameter feature set, the hysteresis state parameter feature set and the accident feature are predicted by the linear model to obtain the predicted value of each battery state parameter indicator.

[0026] In one embodiment, determining the state parameter feature set, the hysteresis state parameter feature set, and the accident feature of the energy storage system according to the energy storage state data set includes:

[0027] Extracting basic characteristic values ​​corresponding to each battery status parameter indicator from the battery system data set in the energy storage status data set to obtain a status parameter feature set;

[0028] Extracting a hysteresis characteristic value corresponding to each battery state parameter indicator from the battery system data set to obtain a hysteresis state parameter feature set;

[0029] The operation log data set in the energy storage state data set is quantified according to an accident quantification algorithm to obtain accident characteristics.

[0030] In one embodiment, updating the energy storage state data set according to each of the predicted values ​​to obtain the updated energy storage state data set, and predicting a target transition time for the energy storage system to transition to the target transition state based on a nonlinear model and the updated energy storage state data set, includes:

[0031] Adding the accident characteristics and each of the predicted values ​​to the energy storage status dataset, and deleting the battery system dataset and the operation log dataset in the energy storage status dataset to obtain an updated energy storage status dataset;

[0032] Inputting the updated energy storage state data set into a nonlinear model, and performing prediction processing on the updated energy storage state data set through the nonlinear model to obtain a jump time period;

[0033] A target transition time for the energy storage system to transition to the target transition state is determined according to the transition time period and the current time.

[0034] In a second aspect, the present application further provides an energy storage system management device, comprising:

[0035] An acquisition module, used to obtain an energy storage status data set of an energy storage system;

[0036] a determination module for performing prediction processing on the energy storage state data set based on a linear model to obtain a predicted value of each battery state parameter indicator, and determining a target transition state of the energy storage system in the next management cycle according to a state transition rule and each predicted value;

[0037] A prediction module is configured to update the energy storage state data set according to each of the predicted values ​​to obtain an updated energy storage state data set, and predict a target transition time for the energy storage system to transition to the target transition state based on a nonlinear model and the updated energy storage state data set; the target transition state and the target transition time are used to manage the energy storage system.

[0038] In a third aspect, the present application further provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0039] Obtaining an energy storage status data set of an energy storage system;

[0040] Performing prediction processing on the energy storage state data set based on a linear model to obtain predicted values ​​of each battery state parameter indicator, and determining a target transition state of the energy storage system in the next management cycle according to a state transition rule and each predicted value;

[0041] The energy storage status data set is updated according to each of the predicted values ​​to obtain an updated energy storage status data set, and a target transition time for the energy storage system to transition to the target transition state is predicted based on a nonlinear model and the updated energy storage status data set; the target transition state and the target transition time are used to manage the energy storage system.

[0042] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the following steps are implemented:

[0043] Obtaining an energy storage status data set of an energy storage system;

[0044] Performing prediction processing on the energy storage state data set based on a linear model to obtain predicted values ​​of each battery state parameter indicator, and determining a target transition state of the energy storage system in the next management cycle according to a state transition rule and each predicted value;

[0045] The energy storage status data set is updated according to each of the predicted values ​​to obtain an updated energy storage status data set, and a target transition time for the energy storage system to transition to the target transition state is predicted based on a nonlinear model and the updated energy storage status data set; the target transition state and the target transition time are used to manage the energy storage system.

[0046] In a fifth aspect, the present application further provides a computer program product, comprising a computer program, which, when executed by a processor, implements the following steps:

[0047] Obtaining an energy storage status data set of an energy storage system;

[0048] Performing prediction processing on the energy storage state data set based on a linear model to obtain predicted values ​​of each battery state parameter indicator, and determining a target transition state of the energy storage system in the next management cycle according to a state transition rule and each predicted value;

[0049] The energy storage status data set is updated according to each of the predicted values ​​to obtain an updated energy storage status data set, and a target transition time for the energy storage system to transition to the target transition state is predicted based on a nonlinear model and the updated energy storage status data set; the target transition state and the target transition time are used to manage the energy storage system.

[0050] The above-mentioned energy storage system management method, apparatus, computer device, computer-readable storage medium, and computer program product obtain an energy storage system energy storage state data set; perform predictive processing on the energy storage state data set based on a linear model to obtain predicted values ​​for each battery state parameter indicator; and determine the target transition state of the energy storage system in the next management cycle based on state transition rules and each predicted value; update the energy storage state data set based on each predicted value to obtain an updated energy storage state data set; and predict the target transition time for the energy storage system to transition to the target transition state based on a nonlinear model and the updated energy storage state data set; the target transition state and target transition time are used to manage the energy storage system. Using this method, the energy storage state data set is predictively processed using a linear model, a nonlinear model, and state transition rules to obtain the target transition state and target transition time for the energy storage system in the next management cycle, thereby adjusting the energy storage system management plan based on the target transition state and target transition time, achieving precise management of the energy storage system and improving the management accuracy of the energy storage system. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following briefly introduces the drawings required for use in the embodiments of the present application or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying any creative work.

[0052] Figure 1 1 is a flow chart of an energy storage system management method according to an embodiment;

[0053] Figure 2 A schematic diagram of a process for obtaining an energy storage status data set in one embodiment;

[0054] Figure 3 Schematic diagram of a process for training an initial linear model and an initial nonlinear model in one embodiment;

[0055] Figure 4 A schematic diagram of a process for determining a sample data set in one embodiment;

[0056] Figure 5 A schematic diagram of a process for determining a predicted value in one embodiment;

[0057] Figure 6 A schematic diagram of a process for determining accident characteristics in one embodiment;

[0058] Figure 7 Schematic diagram of a process for predicting target transition time in one embodiment;

[0059] Figure 8 is a structural block diagram of an energy storage system management device in one embodiment;

[0060] Figure 9 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0061] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0062] With the large-scale grid connection of new energy and the growth of user-side energy storage demand, as well as the gradual completion and structural complexity of energy storage systems, accurate and real-time prediction of the status of energy storage systems has become a key issue in the battery field.

[0063] Traditionally, energy storage system management requires extracting indirect lifespan feature sets from a NASA dataset (a battery dataset). These indirect lifespan feature sets include constant voltage rise and fall charging time, constant voltage drop discharge time, isochronous charging voltage rise and fall, and constant voltage drop discharge voltage drop. Based on these indirect lifespan feature sets and the Extreme Learning Machine (ELM) model, the remaining useful life (RUL) of the energy storage system is predicted.

[0064] However, conventional technologies can only predict the RUL of an energy storage system, and cannot accurately manage the energy storage system based solely on the RUL. Therefore, current energy storage system management methods have low accuracy.

[0065] Furthermore, traditional technologies often focus on a single objective (for example, predicting only cycle life, remaining time, or state of charge), lacking the ability to integrate and collaboratively analyze multi-source data. This results in an inability to effectively resolve conflicts and complementarities between different data sources when dealing with complex systems, limiting the accuracy and generalization of the model. Furthermore, existing modeling techniques often use fixed thresholds or simple rules to divide states, lacking mechanisms for fitting historical data and dynamically adjusting empirical parameters. This makes it difficult for state transition rules to adapt to complex and changing operating environments, reducing the model's adaptability and robustness.

[0066] In addition, existing technologies often use a single model for prediction, which leads to the model's insufficient ability to handle linear trends and nonlinear mutations, making it difficult to achieve high-precision long-term and short-term predictions.

[0067] In summary, this application provides an energy storage system management method that predicts and processes energy storage state data sets through linear models, nonlinear models, and state transition rules. This method can obtain the target transition state and target transition time of the energy storage system in the next management cycle, thereby adjusting the energy storage system management plan based on the target transition state and target transition time, achieving precise management of the energy storage system and improving the management accuracy of the energy storage system. Furthermore, combining linear and nonlinear models to process energy storage state data sets can achieve precise capture of linear and nonlinear characteristics, improving the accuracy of target transition states and target transition times.

[0068] In addition, this application also accurately divides the state of the energy storage system, and during the training process of the initial linear model and the initial nonlinear model, compares the deviation between the predicted data and the actual data, continuously optimizes the model, improves the prediction accuracy and simulation efficiency, and enhances the accuracy of real-time state prediction.

[0069] In one embodiment, Figure 1 As shown, a method for managing an energy storage system is provided. This embodiment of the application takes the method applied to a computer device as an example for description. This embodiment of the application does not limit the execution device of the method for managing an energy storage system. The method includes the following steps 102 to 106:

[0070] Step 102: Acquire an energy storage status data set of the energy storage system.

[0071] During implementation, the computer device obtains the initial energy storage state data set of the energy storage system and preprocesses the initial energy storage state data set to obtain the energy storage state data set. The energy storage state data set includes the energy storage system scheduling plan data set, battery system data set, equipment status telemetering data set, and operation log data set.

[0072] Specifically, the computer device obtains the energy storage system's initial scheduling plan dataset, initial battery system dataset, initial device status telemetry dataset, and initial operation log dataset. The initial scheduling plan dataset represents the energy storage system's scheduling schedule. The battery system dataset represents the energy storage system's operating status. The initial device status telemetry dataset contains real-time status data for each device in the energy storage system. The initial operation log dataset contains the energy storage system's operating logs. The computer device integrates the initial energy storage status dataset and cleans the integrated initial energy storage status dataset to obtain the energy storage status dataset.

[0073] In an exemplary embodiment, in order to achieve precise management of the energy storage system, it is necessary to determine the state of the energy storage system in the current forecast period, the target transition state for the next forecast period, and the target transition time to the target transition state. The computer device obtains the energy storage system's initial scheduling plan dataset, initial battery system dataset, initial device status telesignaling dataset, and initial operation log dataset from the energy storage system, and constructs an initial energy storage status dataset based on the initial scheduling plan dataset, initial battery system dataset, initial device status telesignaling dataset, and initial operation log dataset. The computer device integrates the initial energy storage status dataset according to timestamps to obtain an integrated initial energy storage status dataset. The computer device then processes missing values ​​and removes outliers on the integrated initial energy storage status dataset to obtain an energy storage status dataset.

[0074] Optionally, the energy storage system may be, but is not limited to, a battery system. The embodiments of the present application do not limit the energy storage system.

[0075] Step 104 : Predictively process the energy storage state data set based on the linear model to obtain predicted values ​​of each battery state parameter indicator, and determine the target transition state of the energy storage system in the next management cycle based on the state transition rule and each predicted value.

[0076] During implementation, the computer device determines the energy storage system's state parameter feature set, hysteresis state parameter feature set, and accident characteristics based on the energy storage state data set. The computer device then predicts and processes the energy storage system's state parameter feature set and accident characteristics using a linear model to obtain predicted values ​​for each battery state parameter indicator. The computer device then determines the energy storage system's target transition state for the next prediction cycle based on the state transition rules and the predicted values. The target transition state refers to the energy storage system's state in the next prediction cycle, while a transition refers to the energy storage system's change from its current state in the current prediction cycle to its state in the next prediction cycle.

[0077] Specifically, the state transition rules are determined based on the operating state definition set and the historical state data set of the energy storage system. The operating state definition set contains the definition data of each state of the energy storage system. The states of the energy storage system include normal charging state, charging fault state, charging recovery state, normal discharging state, discharge fault state, discharge recovery state and standby state. Based on the energy storage state data set, the computer device determines the state parameter feature set, hysteresis state parameter feature set and accident characteristics of the energy storage system. Then, the computer device performs prediction processing on the state parameter feature set, hysteresis state parameter feature set and accident characteristics according to the linear model to obtain the predicted value of each battery state parameter indicator. Among them, the predicted value of each battery state parameter indicator is the value of each battery state parameter indicator of the energy storage system in the next prediction cycle predicted by the computer device. Based on the state transition rules and each predicted value, the computer device determines the state of the energy storage system in the next management cycle and determines the state as the target jump state.

[0078] In an exemplary embodiment, the charging state (including the normal charging state, the charging fault state, and the charging recovery state) can transition to the discharging state (including the normal discharging state, the discharging fault state, and the discharging recovery state), and the discharging state can also transition to the charging state. Both the charging state and the discharging state can transition to the standby state. The normal state (including the normal charging state and the normal discharging state) can transition to the fault state (including the charging fault state and the discharging fault state), and the fault state can transition to the recovery state (including the charging recovery state and the fault recovery state). The recovery state can transition to the fault state and the normal state. However, the normal state cannot transition to the recovery state. If the predicted values ​​of the battery status parameter indicators indicate that the energy storage system is not undergoing any charging or discharging operations, the computer device determines that the target transition state of the energy storage system is the standby state. If the predicted values ​​of the battery status parameter indicators indicate that the SOC (remaining battery capacity) of the energy storage system is increasing and the voltage is stable, the computer device determines that the target transition state of the energy storage system is the normal charging state. If the predicted values ​​of the battery status parameter indicators indicate that the energy storage system has triggered an alarm such as overvoltage (the current voltage is greater than the rated charging voltage) or overtemperature (the current temperature is greater than the rated temperature), the computer device determines that the target transition state of the energy storage system is the charging fault state. If the predicted values ​​of the battery status parameter indicators indicate that the energy storage system will resume normal charging after the fault is resolved, the computer device determines that the target transition state of the energy storage system is the charging recovery state.

[0079] If the predicted values ​​of the battery status parameter indicators indicate that the SOC of the energy storage system has decreased and meets the load demand, the computer device determines that the target transition state of the energy storage system is the normal discharge state. If the predicted values ​​of the battery status parameter indicators indicate that the energy storage system has triggered an alarm such as low voltage (the current voltage is less than the rated charging voltage) or overtemperature (the current temperature is less than the rated temperature), the computer device determines that the target transition state of the energy storage system is the discharge fault state. If the predicted values ​​of the battery status parameter indicators indicate that the energy storage system has resumed normal discharge after the fault is resolved, the computer device determines that the target transition state of the energy storage system is the discharge recovery state.

[0080] Step 106 , updating the energy storage state data set according to each predicted value to obtain an updated energy storage state data set, and predicting a target transition time for the energy storage system to transition to a target transition state based on the nonlinear model and the updated energy storage state data set.

[0081] Among them, the target jump state and target jump time are used to manage the energy storage system.

[0082] During implementation, the computer updates the energy storage system status dataset based on the predicted values ​​and accident characteristics, generating an updated energy storage system status dataset. The computer then inputs the updated energy storage system status dataset into a nonlinear model, using the unconstrained model to predict the target transition time for the energy storage system to reach the target state.

[0083] Specifically, the computer device updates the battery system dataset in the energy storage state dataset based on the predicted values ​​and updates the operation log dataset in the energy storage state dataset based on the accident characteristics, thereby obtaining an updated energy storage state dataset. The computer device then inputs the updated energy storage state dataset into a nonlinear model and uses the nonlinear model to perform predictive processing on the updated energy storage state dataset to obtain a transition time period. The transition time period is the interval required for the energy storage system to transition from its current state to a target transition state. Based on the interval and the current time, the computer device determines a target transition time for the energy storage system to transition from its current state to the target transition state.

[0084] In an optional embodiment, the computer device optimizes the initial scheduling plan data of the energy storage system based on the target jump time and the target jump state, so that the scheduling plan of the energy storage system becomes more reasonable, thereby reducing the probability of accidents in the energy storage system and extending the life of the energy storage system.

[0085] In an optional embodiment, the computer device performs predictive processing on the energy storage state dataset based on a linear model to obtain predicted values ​​for each battery state parameter indicator. The computer device then updates the energy storage state dataset based on the predicted values ​​and the accident characteristics to obtain an updated energy storage state dataset. Simultaneously, the computer device acquires real-time sensor data and inputs the updated energy storage state dataset and real-time sensor data into a nonlinear model. The nonlinear model then performs predictive processing on the updated energy storage state dataset and real-time sensor data to obtain a target transition time. The computer device determines the target transition state of the energy storage system for the next management cycle based on the updated energy storage state dataset and the state transition rules.

[0086] In the above-mentioned energy storage system management method, the energy storage state data set is predicted and processed through linear models, nonlinear models and state transition rules, which can obtain the target jump state and target jump time of the energy storage system in the next management cycle. The management plan of the energy storage system is adjusted according to the target jump state and target jump time, thereby realizing precise management of the energy storage system and improving the management accuracy of the energy storage system.

[0087] In an exemplary embodiment, Figure 2 As shown, the specific processing process of step 102 includes steps 202 to 204. Among them:

[0088] Step 202 : Acquire an initial scheduling plan dataset, an initial battery system dataset, an initial device status telesignaling dataset, and an initial operation log dataset of the energy storage system to obtain an initial energy storage status dataset.

[0089] During implementation, the computer device obtains the initial scheduling plan data set, initial battery system data set, initial equipment telemetering status data and initial operation log data of the energy storage system from the energy storage system, and combines the initial scheduling plan data, initial battery system data set, initial equipment status telemetering data set and initial operation log data set to obtain the initial energy storage status data set.

[0090] Specifically, the computer device obtains the energy storage system's initial scheduling plan dataset from the energy storage system. This initial scheduling plan dataset contains information such as the energy storage system's scheduling schedule and charging and discharging plans. Simultaneously, the computer device obtains the energy storage system's initial battery system dataset for the current management cycle. This initial battery system dataset contains the battery's remaining capacity (SOC) indicator, battery state of health (SOH) indicator, and real-time voltage, current, and temperature data. The computer device obtains real-time status data for each device in the energy storage system to obtain an initial device status telesignaling dataset. This device status telesignaling dataset contains real-time status data for devices in the energy storage system, such as the BMS (Battery Management System), PCS (Power Conversion System), and EMS (Energy Management System). Simultaneously, the computer device obtains the energy storage system's initial operation log dataset. This initial operation log dataset contains the energy storage system's warning log, error log, and timestamp records of key events. Key events are accidents.

[0091] Optionally, key events of the energy storage system include but are not limited to abnormal temperature, overcurrent, etc. This application does not limit key events.

[0092] Step 204 , performing data integration on the initial energy storage status data set, and performing data cleaning on the integrated initial energy storage status data set to obtain an energy storage status data set.

[0093] Among them, data cleaning includes missing value processing and outlier removal.

[0094] During implementation, the computer device integrates the data in the initial energy storage state data set according to the timestamp, processes missing values ​​and removes outliers on the integrated initial energy storage state data set, and obtains the energy storage state data set.

[0095] Specifically, the computer device aligns all data in the initial energy storage status dataset according to timestamps to obtain an integrated initial energy storage status dataset. The time resolution for aligning the initial energy storage status dataset is 1 minute. The computer device determines whether each data point in the initial energy storage status dataset is abnormal. If the data point is abnormal, the computer device removes the abnormal data or initializes the abnormal data. For example, if the data point is voltage data, the computer device determines whether the voltage data point is within a normal voltage range. If the voltage data point is not within the normal voltage range, the computer device removes the abnormal voltage data point. Then, for each data point in the initial energy storage status dataset, the computer device determines whether the data point is missing. If the data point is missing, the computer device determines whether the energy storage system is normal based on the log information corresponding to the data point. If the energy storage system is normal, the computer device loads the normal value. For example, if the energy storage system is normal and SOC(t) is missing, the computer device determines the supplementary value corresponding to the data point using linear interpolation and adds the supplementary value to the initial energy storage status dataset. The supplementary value is SOC(t) = SOC(t-1) + [SOC(t+1)- SOC(t-1)] / 2. Here, t is the time when the data was acquired. If the energy storage system is abnormal, the computer equipment will not process the data.

[0096] Optionally, the normal voltage range is less than or equal to 4.2V (V) and greater than or equal to 2.5V. The abnormal voltage range is greater than 4.2V or less than 2.5V. The normal voltage range is determined based on the properties of the energy storage system. This embodiment of the application does not limit the normal voltage range and the abnormal voltage range.

[0097] In this embodiment, the energy storage status data set is obtained by integrating and cleaning the initial energy storage status data set, thereby ensuring the data integrity of the energy storage status data set and facilitating subsequent data processing based on the energy storage status data set.

[0098] In an exemplary embodiment, before performing prediction processing on the energy storage state data set based on the linear model, it is necessary to first train the initial linear model and the initial nonlinear model to obtain the linear model and the nonlinear model. Figure 3 As shown, before step 104 is executed, the specific processing process of the energy storage system management method further includes steps 302 to 306. Among them:

[0099] Step 302: Obtain a sample data set.

[0100] The sample data set includes various sample data subsets.

[0101] During implementation, a computer device obtains an initial sample data set. Each initial sample data subset in the initial sample data set includes a historical energy storage state data set for two historical management cycles. Based on state transition rules, the computer device determines the state of each sample data subset during the two historical management cycles and updates the initial sample data subset based on the state, thereby obtaining a sample data subset. The computer device then combines the sample data subsets to obtain a sample data set.

[0102] Specifically, the computer device extracts a historical energy storage status dataset from the energy storage system, including each historical management cycle, and constructs each initial sample data subset based on each historical energy storage status dataset to obtain an initial sample data set. The computer device obtains an operating state definition set and a historical state dataset, and determines a state transition rule based on the operating state dataset and the historical state dataset. The computer device then determines the state of each initial sample data subset over two historical management cycles based on the state transition rule, and updates the initial sample data subset based on the state to obtain a sample data subset. The computer device then combines the sample data subsets to obtain a sample data set.

[0103] Step 304 : training the initial linear model and the initial nonlinear model based on each sample data subset until the initial linear model and the initial nonlinear model meet a preset training stop condition.

[0104] In an implementation, training stop conditions are pre-set in the computer device. The training stop conditions include a first training stop condition corresponding to the initial linear model and a second training stop condition corresponding to the linear model. The computer device trains the initial linear model and the initial nonlinear model based on each sample data subset until the initial linear model satisfies the first training stop condition and the initial nonlinear model satisfies the second training stop condition.

[0105] Specifically, the sample data subset includes historical energy storage status data subsets from two historical management periods. The two historical management periods are a first historical management period and a second historical management period. The first historical management period is shorter than the second historical management period. The computer device determines the historical energy storage status data subsets from each of the first historical management periods as sample energy storage status data subsets, and determines the historical energy storage status data subsets from the second historical management period as verification energy storage status data subsets. The computer device then determines a sample state parameter feature set, a sample hysteresis parameter feature set, and an accident feature based on the sample energy storage status data subsets. The computer device updates the sample energy storage status data subset based on the sample state parameter feature set, the sample hysteresis parameter feature set, and the accident feature to obtain a target sample state parameter feature set. Since the sample data subset contains the historical operating state of the energy storage system from the first historical management period and the verification state from the second historical management period, the computer device only needs to determine the verification transition time. The computer device then determines the verification transition time based on the verification energy storage status data subset and a preset operating state definition set. The computer device constructs the target sample data subset based on the verification transition time and the target sample state parameter feature set.

[0106] Then, for each target sample data subset, the computer device inputs the target sample state parameter feature set in the target sample data subset into the initial linear model, and performs prediction processing on the target sample state parameter feature set through the initial linear model to obtain a predicted jump state. Then, the computer device updates the parameters of the initial linear model based on the verification state and the predicted jump state until the initial linear model meets the first training stop condition. The computer device updates the target sample state parameter feature set based on the predicted jump state, and inputs the updated target sample state parameter feature set into the initial nonlinear model, and performs prediction processing on the updated target sample parameter feature set through the initial nonlinear model to obtain a predicted jump time. Then, the computer device updates the parameters of the initial nonlinear model based on the verification jump time and the predicted jump time until the nonlinear model meets the second training stop condition.

[0107] In an exemplary embodiment, for each target sample data subset, a computer device inputs a target sample state parameter feature set from the target sample data subset into an initial linear model, performs prediction processing on the target sample state parameter feature set using the initial linear model, and obtains a predicted transition state. The computer device then performs data operations on each predicted transition state and each verification state according to a preset loss algorithm to obtain a loss value. The computer device then updates the parameters of the initial linear model based on the loss value and a preset model optimization algorithm. The computer device then updates the target sample state parameter feature set based on the predicted conditional state to obtain an updated target sample state parameter feature set. The computer device then inputs the updated target sample state parameter feature set into an initial nonlinear model, performs prediction processing on the updated target sample parameter feature set using the initial nonlinear model, and obtains a predicted transition time. The computer device then performs data operations on each predicted transition time and each verification transition time according to a preset loss algorithm to obtain a loss value for the initial nonlinear model. The computer device then updates the parameters of the initial nonlinear model based on the preset model optimization algorithm and the loss value of the initial nonlinear model. The computer device determines whether the initial linear model satisfies a first training stop condition and determines whether the initial nonlinear model satisfies a second training stop condition. If the initial linear model satisfies the first training stop condition and the initial nonlinear model satisfies the second training stop condition, the computer device executes step 306 below. If the initial linear model does not satisfy the first training stop condition and / or the initial nonlinear model does not satisfy the second training stop condition, the computer device executes, for each target sample data subset, the step of inputting the target sample state parameter feature set in the target sample data subset into the initial linear model until the initial linear model satisfies the first training stop condition and the initial nonlinear model satisfies the second training stop condition.

[0108] Optionally, the first training stop condition may be, but is not limited to, that the loss value of the initial linear model does not exceed a preset first loss threshold, or that the accuracy of the initial linear model reaches a first accuracy threshold, or that the number of training rounds of the initial linear model reaches a preset first training round threshold. The second training stop condition may be, but is not limited to, that the loss value of the initial nonlinear model does not exceed a preset second loss threshold, or that the accuracy of the initial nonlinear model reaches a second accuracy threshold, or that the number of training rounds of the initial nonlinear model reaches a preset second training round threshold. The first loss threshold, the first accuracy threshold, the first training round threshold, the second loss threshold, the second accuracy threshold, and the second training round threshold are determined according to training requirements. The embodiment of the present application does not limit the first training stop condition and the second training stop condition.

[0109] Step 306: determine the initial linear model as a linear model, and determine the initial nonlinear model as a nonlinear model.

[0110] In an implementation, the computer device determines the initial linear model that meets the first training stop condition as the linear model, and determines the initial nonlinear model that meets the second training stop condition as the nonlinear model.

[0111] In this embodiment, the initial linear model and the initial nonlinear model are trained by the sample data set to obtain the linear model and the nonlinear model, thereby improving the accuracy of the linear model and the nonlinear model, facilitating the subsequent prediction of the target jump state and the target jump time based on the linear model and the nonlinear model, and improving the accuracy of the target jump state and the target jump time.

[0112] In an exemplary embodiment, Figure 4 As shown, the specific processing process of step 302 includes steps 402 to 408. Among them:

[0113] Step 402: Obtain an initial sample data set.

[0114] The initial sample data set includes various initial sample data subsets.

[0115] In implementation, the computer device obtains each initial sample data subset and constructs an initial sample data set according to each initial sample data subset.

[0116] Specifically, the computer device obtains a subset of the energy storage system's historical energy storage status data for each historical management cycle. Within each subset of the historical energy storage status data, the historical energy storage status data subsets from two consecutive historical management cycles are combined to construct an initial sample data subset, thereby obtaining each initial sample data subset. Thus, each initial sample data subset contains subsets of the historical energy storage status data from two historical management cycles. The computer device determines the preceding historical management cycle as the first management cycle and the succeeding historical management cycle as the second management cycle.

[0117] Step 404 : Acquire an operating state definition set and a historical state data set of the energy storage system, and perform statistical analysis on the historical state data set and the operating state definition set to obtain a state transition rule.

[0118] During implementation, the computer device obtains an operating state definition set and a historical state data set for the energy storage system. This historical state data set is a subset of the energy storage system's historical energy storage state data during each historical management cycle. Based on the operating state definition set, the computer device determines the energy storage system's historical operating state during each historical management cycle. The computer device then performs statistical analysis on each historical operating state and historical state data set to derive state transition rules.

[0119] In an optional embodiment, the computer device obtains an operating state definition set input by the target user. The operating state definition set contains various operating states of the energy storage system and the state definition data corresponding to each operating state. At the same time, the computer device obtains a subset of the historical energy storage state data of the energy storage system in each historical management period to obtain a historical state data set. For each historical management period, the computer device determines the historical operating state that matches the historical management period based on the historical energy storage state data subset and the operating state definition set of the historical management period. Then, the computer device performs statistical analysis on each historical energy storage state data subset and each historical operating state to obtain an initial state transition rule. The initial state transition rule characterizes that the transition conditions between the operating states depend on multiple quantitative indicators (envelope SOC dynamics, fault log quantification results, etc.). For example, assume that the state transition probability is , then the state transition probability is shown in the following formula (1):

[0120] (1)

[0121] In the above formula (1), is the status of the current management cycle (operating status), It is the transition state of the next management cycle. The change value of each battery status parameter indicator from the current management cycle to the next management cycle, Accident characteristics of the current management cycle, is a set of empirical parameters obtained by fitting the historical state data set.

[0122] Because energy storage systems may not fully transition through all operating states, some transition rules between operating states require manual supplementation. The computer device obtains supplementary transition rules recommended by experts and adds them to the initial state transition rules to generate state transition rules. These supplementary transition rules include scenarios such as how long battery overtemperature should be handled and how long battery overcurrent can be left untreated.

[0123] Step 406 : Determine the sample state of each initial sample data subset based on the state transition rule, and update the initial sample data subset according to the sample state to obtain the sample data subset.

[0124] In implementation, the computer device determines the sample state of each initial sample data subset according to the state transition rule, and then adds the sample state to the initial sample data subset to obtain the sample data subset.

[0125] Specifically, the initial sample data subset includes historical energy storage status data subsets for two historical management cycles. The sample status includes a historical operating status and a verification status. For each initial sample data subset, the computer device determines the historical operating status of the first historical management cycle based on the state transition rules and the historical energy storage status data subset of the first historical management cycle, and determines the verification status of the second historical management cycle based on the state transition rules and the historical energy storage status data subset of the second historical management cycle. The computer device then adds the historical operating status of the first historical management cycle and the verification status of the second historical management cycle to the initial sample data subset to obtain a sample data subset.

[0126] Step 408: Combine the sample data subsets to obtain a sample data set.

[0127] During implementation, the computer device combines the sample data subsets to obtain a sample data set.

[0128] In this embodiment, by adding sample states to the training sample set to obtain a sample data set, it is convenient to subsequently train the initial linear model and the initial nonlinear model based on the sample data set to obtain the linear model and the nonlinear model, thereby improving the accuracy of the linear model and the nonlinear model.

[0129] In an exemplary embodiment, the operating state definition set includes definition data of a normal charging state, definition data of a fault charging state, definition data of a recovery charging state, definition data of a normal discharging state, definition data of a fault discharging state, definition data of a recovery discharging state, and definition data of a standby state.

[0130] In implementation, the computer device pre-stores an operating state definition set. This operating state definition set stores various operating states of the energy storage system and definition data for each operating state. This definition data serves as the basis for determining the current operating state of the energy storage system. The operating states include a normal charging state, a charging fault state, a charging recovery state, a normal discharging state, a discharge fault state, a discharge recovery state, and a standby state. Therefore, the operating state definition set includes definition data for a normal charging state, a charging fault state, a charging recovery state, a normal discharging state, a discharge fault state, a discharge recovery state, and a standby state.

[0131] Specifically, the definition data of the normal charging state is that the SOC increases and the voltage is stable. The definition data of the charging fault state is the presence of alarm information such as triggering overvoltage (the current voltage is greater than the rated charging voltage) or overtemperature (the current temperature is greater than the rated temperature). The definition data of the charging recovery state is that normal charging is resumed after the fault is resolved. The definition data of the normal discharge state is that the SOC decreases and the load demand is met. The definition data of the discharge fault state is that alarm information such as triggering low voltage (the current voltage is less than the rated discharge voltage) or overcurrent (the current current exceeds the rated discharge current) is triggered. The definition data of the discharge recovery state is that normal discharge is resumed after the fault is resolved. The definition data of the standby state is no charging or discharging operation.

[0132] In this embodiment, by defining multiple operating states of the energy storage system, the granularity of the operating states is refined, which facilitates subsequent accurate management of the energy storage system based on each operating state.

[0133] In an exemplary embodiment, Figure 5 As shown, the specific process of performing prediction processing on the energy storage state data set based on the linear model in step 104 to obtain the predicted value of each battery state parameter index includes steps 502 to 504. Among them:

[0134] Step 502 : determining a state parameter feature set, a hysteresis state parameter feature set, and an accident feature of the energy storage system based on the energy storage state data set.

[0135] During implementation, the computer device is pre-configured with various battery status parameter indicators. The computer device extracts the basic characteristic values ​​of each battery status parameter indicator from the energy storage status dataset to obtain a status parameter feature set. The computer device also extracts the hysteresis basic characteristics of each battery status parameter indicator from the energy storage status dataset to obtain a hysteresis status parameter feature set. The computer device then quantifies the operation log dataset within the energy storage status dataset to obtain accident characteristics. The status parameter feature set represents the current state of the energy storage system. The hysteresis status parameter feature set represents the state of the energy storage system within a hysteresis collection period. The hysteresis collection period is the first n data collection periods before the current data collection period. Here, n is determined based on data processing requirements.

[0136] In step 504, the state parameter feature set, the hysteresis state parameter feature set and the accident feature are input into the linear model, and the linear model is used to perform prediction processing on the state parameter feature set, the hysteresis state parameter feature set and the accident feature to obtain the predicted value of each battery state parameter indicator.

[0137] Among them, the linear model is an autoregressive model.

[0138] During implementation, the computer device inputs the state parameter feature set, the lagged state parameter feature set and the accident feature into the autoregressive model, and performs prediction processing on the state parameter feature set, the lagged state parameter feature set and the accident feature through the autoregressive model to obtain the predicted value of each battery state parameter indicator.

[0139] In an exemplary embodiment, in an energy storage system, the change in battery status over time exhibits a clear linear trend, while also being subject to nonlinear influences from various emergencies (such as battery cluster disconnection and closing, single cell overheating, over-current limits, and EMS dispatch). Traditional single models have the following shortcomings: Autoregressive model (AR model): excels at capturing long-term and short-term linear dependencies in time series, but cannot effectively capture sudden, nonlinear events (such as equipment failures or drastic temperature fluctuations). Random Forest (RF) model: While it has advantages in handling nonlinear problems and can learn the interaction between sub-features through multiple decision trees, its ability to depict temporal trends and continuity features is relatively limited.

[0140] This application combines linear and nonlinear models to determine an autoregressive enhanced random forest prediction algorithm. This autoregressive enhanced random forest prediction algorithm integrates the advantages of linear and nonlinear models: the linear model is an autoregressive model and is also an autoregressive module (AR). The autoregressive module can quantify and capture the linear trend of state changes and provide short-term prediction values ​​for the model. The nonlinear model is a random forest model, which is also a random forest module. This random forest model can integrate real-time data from sensors and quantitative values ​​of various events, use ensemble learning to fully model nonlinear factors, and identify and predict sudden changes in battery status.

[0141] On the one hand, the above combination solves the shortcomings of a single model in describing the continuous changes in battery status. On the other hand, it can more accurately capture the state jumps caused by key events and achieve accurate prediction of the battery status jump points, thereby supporting the management system that runs the energy storage system management method to perform skip-step operation and reduce repeated calculations, thereby achieving the effect of saving system computing resources and real-time regulation.

[0142] The computer device determines the state parameter feature set, the hysteresis state parameter feature set, and the accident feature of the energy storage system based on the energy storage state data set. Then, the computer device inputs the state parameter feature set, the hysteresis state parameter feature set, and the accident feature into the autoregressive model, and performs prediction processing on the state parameter feature set, the hysteresis state parameter feature set, and the accident feature through the autoregressive model to obtain the predicted value of each battery state parameter indicator. The autoregressive model is shown in the following formula (2):

[0143] (2)

[0144] Among them, in the above formula (2) Indicates at time status indicators (including the predicted values ​​of various battery status parameter indicators). are model parameters, is the model order, is the random noise term. For the moment The values ​​of each battery status parameter indicator.

[0145] The computer device is based on the autoregressive model at the time Get a predicted value (trend value). The prediction process is shown in the following formula (3):

[0146] (3)

[0147] In the above formula (3), are model parameters, is the model order, Indicates at time The values ​​of each battery status parameter indicator. It is a set of predicted values ​​of each battery status parameter indicator. The linear continuation trend reflecting the state of the energy storage system in the short term in the future will serve as one of the enhanced features of the subsequent nonlinear model.

[0148] In this embodiment, a linear model is used to predict and process the state parameter feature set, the hysteresis state parameter feature set, and the accident characteristics. This can quantify and capture the linear trend of state changes, thereby determining the predicted value of each battery state parameter indicator. Each predicted value provides a basis for determining the target transition state.

[0149] In an exemplary embodiment, Figure 6 As shown, the specific processing process of step 502 includes steps 602 to 606. Among them:

[0150] Step 602 : extracting basic characteristic values ​​corresponding to each battery status parameter indicator from the battery system data set in the energy storage status data set to obtain a status parameter characteristic set.

[0151] Among them, the energy storage status dataset includes the battery system dataset.

[0152] In implementation, the computer device is pre-set with various battery status parameter indicators, extracts basic feature values ​​corresponding to the various battery status parameter indicators from the battery system data set, and constructs a status parameter feature set based on the basic feature values.

[0153] Specifically, the current management cycle includes various data collection cycles. The battery state parameter indicators are SOC, voltage, current, temperature, and SOH. The computer device determines the last data collection cycle as the target data collection cycle in each data collection cycle. Then, the computer device extracts the basic characteristic values ​​of SOC, voltage, current, temperature, and SOH of the target collection cycle from the battery system data set of the current management cycle, and constructs a state parameter feature set based on the basic characteristic values ​​of SOC, voltage, current, temperature, and SOH of the target collection cycle. Then, the computer device normalizes the state parameter feature set to obtain a normalized state parameter feature set.

[0154] Step 604 : extracting the hysteresis characteristic value corresponding to each battery state parameter indicator from the battery system data set to obtain a hysteresis state parameter characteristic set.

[0155] In implementation, the computer device extracts hysteresis eigenvalues ​​corresponding to each battery state parameter indicator from the battery data set. Hysteresis eigenvalues ​​are eigenvalues ​​whose collection period is earlier than the basic eigenvalue. The computer device then constructs a hysteresis state parameter feature set based on each hysteresis eigenvalue.

[0156] Specifically, the battery state parameter indicators are SOC, voltage, current, temperature and SOH. The computer device determines the hysteresis acquisition period based on the target acquisition period of the basic characteristic value. Then, the computer device extracts the hysteresis characteristic value of SOC, the hysteresis characteristic value of voltage, the hysteresis characteristic value of current, the hysteresis characteristic value of temperature and the hysteresis characteristic value of SOH in the hysteresis acquisition period from the battery system data set. The computer device constructs the hysteresis state parameter characteristic value based on each hysteresis characteristic value. Among them, if the target acquisition period of the basic characteristic value is n, the hysteresis acquisition period is n-1, n-2, n-3, n-4, n-5. The hysteresis order is 5. Then, the computer device normalizes the hysteresis state parameter feature set to obtain the normalized hysteresis state parameter feature set.

[0157] Optionally, the lag order may be, but is not limited to, 5, which is determined according to data processing requirements. The embodiment of the present application does not limit the lag order.

[0158] Step 606 : quantify the operation log data set in the energy storage status data set according to the accident quantification algorithm to obtain accident characteristics.

[0159] The energy storage status dataset includes the operation log dataset.

[0160] During implementation, the computer equipment quantifies each accident data in the operation log data set according to the accident quantification algorithm to obtain the accident characteristics.

[0161] Specifically, each type of accident has a corresponding accident weight. For each log data in the computer device operation log data set, the accident type represented by the log data is determined, and the number of occurrences of each accident type is counted. Then, the computer device performs weighted processing on the number of occurrences and accident weights of each accident type according to the accident quantification algorithm to obtain the accident characteristics. The weighted processing process is shown in the following formula (4):

[0162] (4)

[0163] In the above formula (4), Characteristics of the accident. The accident type per unit time The number of accidents that occurred. Accident type The weight of the accident. The accident type per unit time The number of accidents that occurred. Accident type The weight of the accident.

[0164] In this embodiment, the state parameter feature set, the hysteresis state parameter feature set and the accident feature are determined by centrally collecting the energy storage state data, which facilitates the subsequent prediction and processing of the state parameter feature set, the hysteresis state parameter feature set and the accident feature based on the linear model, thereby avoiding the linear model from directly processing the energy storage state data set, reducing the amount of data that the linear model needs to process, and improving the efficiency of the linear model.

[0165] In an exemplary embodiment, Figure 7 As shown, the specific processing process of step 504 includes steps 702 to 706. Among them:

[0166] Step 702 : Add the accident characteristics and each predicted value to the energy storage status data set, and delete the battery system data set and the operation log data set in the energy storage status data set to obtain an updated energy storage status data set.

[0167] During implementation, the computer device adds each predicted value to the energy storage status dataset and deletes the energy storage status dataset. At the same time, the computer device adds the accident characteristics to the energy storage status dataset and deletes the operation log dataset in the energy storage status dataset to obtain an updated energy storage status dataset.

[0168] Step 704 : input the updated energy storage state data set into the nonlinear model, and perform prediction processing on the updated energy storage state data set through the nonlinear model to obtain a jump time period.

[0169] The nonlinear model is a random forest model. The transition time period is the interval between the energy storage system transitioning from the current state to the target transition state.

[0170] During implementation, the computer device inputs the updated energy storage status data set into the random forest model, and performs prediction processing on the updated energy storage status data set through the random forest model to obtain the jump time period.

[0171] In an exemplary embodiment, a computer device acquires sensor data collected in real time The computer device constructs a feature vector based on the sensor data and the updated energy storage status data set. The feature vector is shown in the following formula (5):

[0172] (5)

[0173] In the above formula (5), is the sensor data, For each predicted value, is the accident characteristic of the current management cycle t. is the feature vector. is the set of basis and lagged eigenvalues. It is the equipment status telesignaling data of the current management cycle t. is the impact factor of the battery cluster opening / closing event in the energy storage system, which represents the impact of the battery cluster opening / closing operation in the current management cycle t on the energy storage system state. The impact factor of abnormal battery cell temperature events in the energy storage system indicates the degree of impact of the energy storage system state change caused by the battery temperature exceeding the set upper limit or falling below the lower limit in the current management cycle t. The current over-limit event impact factor indicates the degree of impact on the status caused by the battery operating current exceeding the maximum allowable current in the current management cycle t. It is obtained based on the processing of equipment status telesignaling data. and It is obtained based on the battery system data set processing.

[0174] The computer device inputs the feature vector into the random forest model, and the random forest model predicts the feature vector to obtain the jump time period. The random forest model contains N decision trees. Each decision tree Learn based on the input feature subset and output the predicted value (i.e., battery status prediction or transition signal). Combining the predictions of all trees, the final prediction value is obtained by majority voting or mean fusion, which is the output transition time period. The prediction process of the random forest model is shown in the following formula (6):

[0175] (6)

[0176] In the above formula (6), is the transition time period, is the input feature vector. Indicates the A decision tree, Indicates the number of decision trees.

[0177] Step 706: Determine a target transition time for the energy storage system to transition to a target transition state based on the transition time period and the current time.

[0178] During implementation, the computer device performs an addition operation on the transition time period and the current time to obtain a target transition time for the energy storage system to transition to a target transition state.

[0179] In an optional embodiment, the computer device can also determine the target jump state after determining the target jump time period. Specifically, in order to reflect the linear trend and nonlinear mutation effects at the same time, the overall model finally adopts the following comprehensive form to calculate the current state Transition to the next management cycle Probability As shown in the following formula (6):

[0180] (6)

[0181] In the above formula (6), is the current state, Status for the next management cycle. The sigmoid function can normalize the comprehensive input to the range of 0 to 1; is the bias term, and the rest The weight coefficient is obtained by fitting historical data. For each predicted value, is the accident characteristic of the current management cycle t. is the feature vector. is the set of basis and lagged eigenvalues. It is the equipment status telesignaling data of the current management cycle t. is the impact factor of the battery cluster opening / closing event in the energy storage system, which represents the impact of the battery cluster opening / closing operation in the current management cycle t on the energy storage system state. The impact factor of abnormal battery cell temperature events in the energy storage system indicates the degree of impact of the energy storage system state change caused by the battery temperature exceeding the set upper limit or falling below the lower limit in the current management cycle t. The current over-limit event impact factor indicates the degree of impact on the status caused by the battery operating current exceeding the maximum allowable current in the current management cycle t. It is obtained based on the processing of equipment status telesignaling data. and It is obtained based on the battery system data set processing.

[0182] This formula combines the linear trends predicted by the AR model with the nonlinear jump characteristics of the battery's operating state through the quantified values ​​of the impact of various events. By integrating and learning multi-dimensional features, it can ultimately more accurately predict the state transition points, thereby guiding the management system to perform step-by-step operations.

[0183] In this embodiment, the updated energy storage status data set is predicted and processed using a nonlinear model to obtain the target transition time of the energy storage system, thereby adjusting the management plan of the energy storage system according to the target transition time, achieving precise management of the energy storage system, and improving the management accuracy of the energy storage system.

[0184] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0185] Based on the same inventive concept, embodiments of the present application also provide an energy storage system management device for implementing the aforementioned energy storage system management method. The solution provided by this device is similar to the solution described in the aforementioned method. Therefore, the specific limitations of one or more energy storage system management device embodiments provided below can be found in the above-described limitations of the energy storage system management method and will not be further elaborated here.

[0186] In an exemplary embodiment, Figure 8 As shown, an energy storage system management device 800 is provided, including: an acquisition module 801, a determination module 802 and a prediction module 803, wherein:

[0187] The acquisition module 801 is used to acquire an energy storage status data set of the energy storage system.

[0188] The determination module 802 is used to perform prediction processing on the energy storage state data set based on the linear model to obtain the predicted value of each battery state parameter indicator, and determine the target jump state of the energy storage system in the next management cycle according to the state transition rule and each predicted value.

[0189] The prediction module 803 is used to update the energy storage state data set according to each predicted value to obtain an updated energy storage state data set, and predict the target transition time for the energy storage system to transition to the target transition state based on the nonlinear model and the updated energy storage state data set; the target transition state and target transition time are used to manage the energy storage system.

[0190] In an exemplary embodiment, the acquisition module 801 includes:

[0191] The first acquisition submodule is used to acquire the initial scheduling plan data set, the initial battery system data set, the initial equipment status telesignaling data set and the initial operation log data set of the energy storage system to obtain the initial energy storage status data set.

[0192] The first cleaning submodule is used to perform data integration on the initial energy storage state data set and perform data cleaning on the integrated initial energy storage state data set to obtain an energy storage state data set.

[0193] In an exemplary embodiment, the energy storage system management device 800 further includes:

[0194] The second acquisition module is used to acquire a sample data set; the sample data set includes various sample data subsets.

[0195] The training module is used to train the initial linear model and the initial nonlinear model based on each sample data subset until the initial linear model and the initial nonlinear model meet a preset training stop condition.

[0196] The second determining module is used to determine the initial linear model as a linear model and determine the initial nonlinear model as a nonlinear model.

[0197] In an exemplary embodiment, the second acquisition module includes:

[0198] The first acquisition submodule is used to acquire an initial sample data set; the initial sample data set includes various initial sample data subsets.

[0199] The second acquisition submodule is used to obtain the operating state definition set and the historical state data set of the energy storage system, and perform statistical analysis on the historical state data set and the operating state definition set to obtain the state transition rules.

[0200] The first determining submodule is configured to determine the sample state of each initial sample data subset based on a state transition rule, and update the initial sample data subset according to the sample state to obtain a sample data subset.

[0201] The first combining submodule is used to combine the sample data subsets to obtain a sample data set.

[0202] In an exemplary embodiment, the operating state definition set includes definition data of a normal charging state, definition data of a fault charging state, definition data of a recovery charging state, definition data of a normal discharging state, definition data of a fault discharging state, definition data of a recovery discharging state, and definition data of a standby state.

[0203] In an exemplary embodiment, the determination module includes a first prediction submodule and a second determination submodule. The first prediction submodule includes:

[0204] The third determination submodule is configured to determine a state parameter feature set, a hysteresis state parameter feature set, and an accident feature of the energy storage system according to the energy storage state data set.

[0205] The second prediction submodule is used to input the state parameter feature set, the hysteresis state parameter feature set and the accident feature into the linear model, and perform prediction processing on the state parameter feature set, the hysteresis state parameter feature set and the accident feature through the linear model to obtain the predicted value of each battery state parameter indicator.

[0206] In an exemplary embodiment, the third determining submodule includes:

[0207] The first extraction submodule is configured to extract basic characteristic values ​​corresponding to various battery status parameter indicators from the battery system data set in the energy storage status data set to obtain a status parameter characteristic set.

[0208] The second extraction submodule is used to extract the hysteresis characteristic value corresponding to each battery state parameter indicator from the battery system data set to obtain a hysteresis state parameter feature set.

[0209] The quantification submodule is used to quantify the operation log data set in the energy storage status data set according to the accident quantification algorithm to obtain accident characteristics.

[0210] In an exemplary embodiment, the prediction module 803 includes:

[0211] The first updating module is used to add the accident characteristics and each predicted value to the energy storage status data set, and delete the battery system data set and the operation log data set in the energy storage status data set to obtain an updated energy storage status data set.

[0212] The third prediction submodule is used to input the updated energy storage state data set into the nonlinear model, and perform prediction processing on the updated energy storage state data set through the nonlinear model to obtain a jump time period.

[0213] The fourth determining submodule is configured to determine a target transition time for the energy storage system to transition to a target transition state according to the transition time period and the current time.

[0214] Each module in the energy storage system management device described above can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in hardware form, or stored in a computer device memory in software form, allowing the processor to call and execute the corresponding operations of each module.

[0215] In an exemplary embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as shown in FIG. Figure 9 As shown. The computer device includes a processor, memory, an input / output interface, a communication interface, a display unit, and an input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are connected to the system bus via the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals via wired or wireless communication, and the wireless communication can be achieved via Wi-Fi, a mobile cellular network, near-field communication (NFC), or other technologies. When executed by the processor, the computer program implements a method for managing an energy storage system. The display unit of the computer device is used to form a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device casing, or an external keyboard, touchpad or mouse.

[0216] Those skilled in the art will understand that Figure 9 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0217] In one embodiment, a computer device is further provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above method embodiments when executing the computer program.

[0218] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0219] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0220] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. In particular, any reference to memory, database, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the various embodiments provided herein may be, but are not limited to, general-purpose processors, central processing units (CPUs), graphics processing units (GPUs), digital signal processors (DSPs), programmable logic devices (PLDs), quantum computing-based data processing logic devices, artificial intelligence (AI) processors, and the like.

[0221] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0222] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A method for managing an energy storage system, characterized in that: The method comprises: Obtaining an energy storage status data set of an energy storage system; Performing prediction processing on the energy storage state data set based on a linear model to obtain predicted values ​​of each battery state parameter indicator, and determining a target transition state of the energy storage system in the next management cycle according to a state transition rule and each predicted value; The energy storage status data set is updated according to each of the predicted values ​​to obtain an updated energy storage status data set, and a target transition time for the energy storage system to transition to the target transition state is predicted based on a nonlinear model and the updated energy storage status data set; the target transition state and the target transition time are used to manage the energy storage system.

2. The method according to claim 1, characterized in that The obtaining of the energy storage status data set of the energy storage system includes: Obtain the initial scheduling plan dataset, initial battery system dataset, initial equipment status telesignaling dataset, and initial operation log dataset of the energy storage system to obtain the initial energy storage status dataset; The initial energy storage state data set is integrated, and the integrated initial energy storage state data set is cleaned to obtain an energy storage state data set.

3. The method according to claim 1, characterized in that Before performing prediction processing on the energy storage state data set based on the linear model to obtain predicted values ​​of each battery state parameter indicator, the method further includes: Acquire a sample data set; the sample data set includes each sample data subset; Training an initial linear model and an initial nonlinear model based on each of the sample data subsets until the initial linear model and the initial nonlinear model meet a preset training stop condition; The initial linear model is determined as a linear model, and the initial nonlinear model is determined as a nonlinear model.

4. The method according to claim 3, characterized in that The obtaining of the sample data set includes: Acquire an initial sample data set; the initial sample data set includes each initial sample data subset; Obtaining an operating state definition set and a historical state data set of the energy storage system, and performing statistical analysis on the historical state data set and the operating state definition set to obtain a state transition rule; Determining a sample state of each initial sample data subset based on the state transition rule, and updating the initial sample data subset according to the sample state to obtain a sample data subset; The sample data subsets are combined to obtain a sample data set.

5. The method according to claim 4, characterized in that The operation state definition set includes definition data of normal charging state, definition data of fault charging state, definition data of charging recovery state, definition data of normal discharging state, definition data of fault discharging state, definition data of discharge recovery state and definition data of standby state.

6. The method according to claim 1, wherein The predictive processing of the energy storage state data set based on the linear model to obtain the predicted value of each battery state parameter index includes: Determine a state parameter feature set, a hysteresis state parameter feature set and an accident feature of the energy storage system according to the energy storage state data set; The state parameter feature set, the hysteresis state parameter feature set and the accident feature are input into a linear model, and the state parameter feature set, the hysteresis state parameter feature set and the accident feature are predicted by the linear model to obtain the predicted value of each battery state parameter indicator.

7. The method according to claim 6, characterized in that The determining of the state parameter feature set, the hysteresis state parameter feature set and the accident feature of the energy storage system according to the energy storage state data set includes: Extracting basic characteristic values ​​corresponding to each battery status parameter indicator from the battery system data set in the energy storage status data set to obtain a status parameter feature set; Extracting a hysteresis characteristic value corresponding to each battery state parameter indicator from the battery system data set to obtain a hysteresis state parameter feature set; The operation log data set in the energy storage state data set is quantified according to an accident quantification algorithm to obtain accident characteristics.

8. The method according to claim 6, characterized in that The step of updating the energy storage state data set according to each of the predicted values ​​to obtain an updated energy storage state data set, and predicting a target transition time for the energy storage system to transition to the target transition state based on a nonlinear model and the updated energy storage state data set, includes: Adding the accident characteristics and each of the predicted values ​​to the energy storage status dataset, and deleting the battery system dataset and the operation log dataset in the energy storage status dataset to obtain an updated energy storage status dataset; Inputting the updated energy storage state data set into a nonlinear model, and performing prediction processing on the updated energy storage state data set through the nonlinear model to obtain a jump time period; A target transition time for the energy storage system to transition to the target transition state is determined according to the transition time period and the current time.

9. An energy storage system management device, characterized in that: The device comprises: An acquisition module, used to obtain an energy storage status data set of an energy storage system; a determination module for performing prediction processing on the energy storage state data set based on a linear model to obtain a predicted value of each battery state parameter indicator, and determining a target transition state of the energy storage system in the next management cycle according to a state transition rule and each predicted value; A prediction module is configured to update the energy storage state data set according to each of the predicted values ​​to obtain an updated energy storage state data set, and predict a target transition time for the energy storage system to transition to the target transition state based on a nonlinear model and the updated energy storage state data set; the target transition state and the target transition time are used to manage the energy storage system.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.

Images