Battery energy storage device fire prevention and control method and system

Abstract

The application discloses a battery energy storage device fire prevention and control method and system, comprising: collecting real-time fire characteristic parameters of the battery energy storage device; making a fire-fighting decision based on the real-time fire characteristic parameters and a preset fire-fighting strategy and generating a fire-fighting instruction; obtaining effect feedback after the fire-fighting instruction is executed; optimizing and updating the fire-fighting strategy based on the real-time fire characteristic parameters and the effect feedback; and applying the optimized and updated fire-fighting strategy to subsequent fire-fighting decisions of the battery energy storage device. The method significantly improves the accuracy and self-adaptive ability of battery energy storage device fire prevention and control through real-time monitoring, digital twin simulation and closed-loop optimization mechanism, realizes dynamic adjustment and low-risk verification of the fire-fighting strategy, and effectively enhances safety and reliability.

Description

Technical Field

[0001] This invention relates to the field of fire safety technology for battery energy storage systems, and in particular to a fire prevention and control method and system for battery energy storage equipment. Background Technology

[0002] With the rapid development of the new energy industry, battery energy storage devices are widely used in energy storage power stations and other fields, highlighting the increasing importance of their fire safety management. Currently, fire safety in energy storage power stations mainly relies on various sensors and suppression units for fire detection and suppression. However, the lack of a centralized intelligent management platform leads to data silos between systems, making it difficult to comprehensively grasp the safe operating status of energy storage power stations. Existing fire prevention strategies are often fixed and preset, unable to be dynamically adjusted according to actual fire conditions, and lack quantitative assessment and closed-loop optimization mechanisms for fire suppression effectiveness, affecting the accuracy and adaptability of fire prevention and control. Summary of the Invention

[0003] To address the technical problems existing in the background art, this invention proposes a fire prevention and control method for battery energy storage devices, comprising the following steps:

[0004] S1. Collect real-time fire characteristic parameters of battery energy storage equipment;

[0005] S2. Make fire-fighting decisions and generate fire-fighting instructions based on real-time fire characteristic parameters and preset fire-fighting strategies;

[0006] S3. Obtain feedback on the effect of fire command execution;

[0007] S4. Optimize and update fire protection strategies based on real-time fire characteristic parameters and effect feedback;

[0008] S5. Apply the optimized and updated fire protection strategy to the subsequent fire protection decisions of the battery energy storage device;

[0009] Step S4 includes:

[0010] S41. Based on the feedback of the effect after the execution of fire orders and the changes in fire characteristic parameters, generate a deviation evaluation index between the actual fire extinguishing efficiency and the expected target.

[0011] S42. Input the deviation evaluation index and real-time fire characteristic parameters into the adaptive analysis model to identify the dimensions to be optimized in the fire protection strategy. The dimensions include response triggering conditions, priority rules and threshold parameters.

[0012] S43. By using a rules engine in conjunction with historical fire protection data, dynamically adjust the identified dimensions to be optimized and generate optimized and updated fire protection strategies.

[0013] Furthermore, S41 specifically includes:

[0014] S411. Using the fire characteristic parameters at the moment the fire command is executed as the initial state, the fire command is simulated through the digital twin of the battery energy storage device to generate the expected fire extinguishing efficiency curve.

[0015] S412. Based on the initial state, and combining the effect feedback after the execution of the fire command and the time-series change data of fire characteristic parameters within the monitoring period, generate the actual fire extinguishing efficiency curve.

[0016] S413. Calculate the difference between the actual fire extinguishing efficiency curve and the expected fire extinguishing efficiency curve at each time point, and integrate to generate a deviation evaluation index.

[0017] Furthermore, the digital twin is specifically a digital twin of the battery energy storage device constructed based on the physical structure, electrical characteristics, and historical operating data of the battery energy storage device, which can simulate the battery thermal runaway process and the response behavior of the fire extinguishing system; the expected fire extinguishing efficiency curve is specifically the trajectory of the predicted fire characteristic parameters changing over time.

[0018] This invention utilizes a digital twin to simulate the expected fire extinguishing efficiency curve and combines it with actual monitoring data to generate the actual fire extinguishing efficiency curve. By calculating the differences in characteristic parameters between the two at different time points and integrating them, a deviation evaluation index is obtained, enabling a refined and quantitative evaluation of the fire extinguishing process. This provides a reliable data foundation for subsequent strategy optimization and enhances the scientific nature and reliability of fire prevention and control.

[0019] Furthermore, the identification of dimensions to be optimized in the fire protection strategy specifically includes: the adaptive analysis model extracts the indicator features of the deviation evaluation index and the time series data of the fire characteristic parameters, combines the indicator features and the time series data to obtain the optimization weights of multiple dimensions, and defines the dimensions whose optimization weights are greater than the preset weight thresholds as the dimensions to be optimized.

[0020] This invention extracts time-series data of indicator features and fire characteristic parameters through an adaptive analysis model, calculates the optimization weights for each dimension, and automatically identifies dimensions requiring optimization whose weights exceed thresholds. This enables targeted strategy adjustments, avoids redundant global strategy modifications, improves optimization efficiency, and enhances the system's adaptability and operational stability. This mechanism can accurately pinpoint weaknesses in fire protection strategies, avoid blind adjustments, improve optimization efficiency, and ensure that each strategy update targets the most critical areas for improvement, thereby continuously enhancing the overall system's fire prevention and control capabilities.

[0021] Furthermore, S43 specifically includes:

[0022] S431. Based on historical fire protection data of battery energy storage equipment, the rule engine generates multiple candidate adjustment schemes for the target to be optimized. The target to be optimized is the dimension with the largest optimization weight.

[0023] S432. Input the candidate adjustment schemes into the digital twin of the battery energy storage device in sequence for simulation verification, and obtain the simulated fire extinguishing efficiency curves corresponding to each candidate adjustment scheme.

[0024] S433. Calculate the simulation deviation evaluation index between each simulated fire extinguishing efficiency curve and the expected fire extinguishing efficiency curve, and select the candidate adjustment scheme with the smallest simulation deviation evaluation index as the optimized and updated fire protection strategy.

[0025] In this invention, the rule engine generates multiple candidate adjustment schemes for fire protection strategies based on historical fire protection data. The effects of each scheme are verified through digital twin simulation, and the scheme with the smallest simulation deviation is selected as the final optimization strategy. This ensures that the adjustment of fire protection strategies is based on historical experience and has been verified virtually, thereby improving the accuracy and reliability of fire protection decisions and reducing actual operational risks.

[0026] The present invention also provides a fire prevention and control system for battery energy storage equipment, comprising: a data acquisition module, an execution module, a station-level host, and a cloud management platform;

[0027] The data acquisition module communicates with the station-level host and is used to collect real-time fire characteristic parameters of the battery energy storage equipment and upload them to the station-level host.

[0028] The station-level host is connected to the execution module and is used to make fire-fighting decisions and generate fire-fighting instructions based on real-time fire characteristic parameters and preset fire-fighting strategies. The fire-fighting instructions are then sent to the execution module. The host is also used to obtain feedback on the effects of the fire-fighting instructions after they are executed and to upload the feedback and real-time fire characteristic parameters to the cloud management platform.

[0029] The execution module is used to execute the fire command and return the effect feedback data generated after the command execution to the station-level host.

[0030] The cloud management platform communicates with the station-level host to receive the real-time fire characteristic parameters and the effect feedback data, and executes the following optimization process:

[0031] Deviation assessment unit: Based on the feedback data of the effect after the execution of fire orders and the changes in fire characteristic parameters, it generates deviation assessment indicators between the actual fire extinguishing effectiveness and the expected target.

[0032] Dimension identification unit: Inputs the deviation evaluation index and real-time fire characteristic parameters into the adaptive analysis model to identify the dimensions to be optimized in the fire protection strategy. The dimensions include response triggering conditions, priority rules and threshold parameters.

[0033] Rule Adjustment Unit: Through the rule engine and in conjunction with historical fire protection data, it dynamically adjusts the identified dimensions to be optimized, and generates optimized and updated fire protection strategies;

[0034] The cloud management platform will also distribute the optimized and updated fire protection strategy to the station-level host;

[0035] The station-level host is also used to receive the optimized and updated fire protection strategy issued by the cloud management platform, and apply the optimized and updated fire protection strategy to the subsequent fire protection decisions of the battery energy storage equipment.

[0036] This invention proposes a fire prevention and control method for battery energy storage devices. This method collects fire characteristic parameters of the battery energy storage device in real time, generates fire commands based on a preset fire prevention strategy, obtains feedback on the execution results, and optimizes and updates the fire prevention strategy based on real-time parameters and feedback, applying this optimization to subsequent decision-making. This method effectively improves the accuracy and adaptability of fire prevention and control, and the effectiveness of the fire prevention strategy is ensured through simulation verification, thereby enhancing the safety and operational reliability of battery energy storage devices and reducing fire risk. Attached Figure Description

[0037] Figure 1 This is an overall flowchart of the fire prevention and control method for battery energy storage devices proposed in this invention;

[0038] Figure 2 This is a partial flowchart of the fire prevention and control method for battery energy storage devices proposed in this invention;

[0039] Figure 3 This is a partial flowchart of the fire prevention and control method for battery energy storage devices proposed in this invention;

[0040] Figure 4 This is a partial flowchart of the fire prevention and control method for battery energy storage devices proposed in this invention;

[0041] Figure 5 This is a schematic diagram of the fire prevention and control system for battery energy storage devices proposed in this invention. Detailed Implementation

[0042] Reference Figures 1 to 4 The present invention proposes a fire prevention method for battery energy storage devices, comprising the following steps:

[0043] S1. Collect real-time fire characteristic parameters of the battery energy storage device. In this embodiment, the fire characteristic parameters are temperature data, smoke concentration data, CO concentration data, and VOC concentration data, which are collected by various sensors deployed in the battery compartment and then uploaded to the station-level host.

[0044] S2. Make fire-fighting decisions and generate fire-fighting instructions based on real-time fire characteristic parameters and preset fire-fighting strategies.

[0045] In this embodiment, the station-level host performs protocol parsing and format conversion on the received fire characteristic parameters, then calls the alarm service to perform a fire risk assessment. The real-time fire characteristic parameters are compared with multi-level alarm thresholds in the preset fire protection strategy: if the parameters reach the first-level alarm threshold, an early warning instruction is generated to alert operators to the anomaly and prompt them to conduct an inspection; if the parameters reach the second-level alarm threshold, a warning signal is generated and sent to the client to prompt manual intervention, while simultaneously sending a signal to the battery management system to restrict operation; if the parameters reach the third-level alarm threshold, a fire extinguishing instruction is generated and issued to the corresponding fire extinguishing actuator to initiate fire extinguishing actions, while simultaneously sending a signal to the battery management system to stop charging and discharging. All instruction generation is based on real-time data and strategy matching results, and the instruction type and parameter status at the trigger time are fully recorded, providing crucial input for subsequent fire extinguishing effectiveness evaluation and strategy updates.

[0046] S3. Obtain feedback on the effect of fire command execution.

[0047] In this embodiment, the effect feedback information includes confirmation of the fire extinguishing actuator's operational status, the actual status of the extinguishing agent release, the command response status of the battery management system, and confirmation signals for manual intervention. When the station-level host issues a fire command, the fire extinguishing actuator in the execution module returns its startup status signal, such as a status code indicating successful startup, startup failure, or equipment malfunction. Simultaneously, the extinguishing agent release device uses built-in sensors to provide feedback on whether the extinguishing agent is released normally and whether the release amount meets expectations. Furthermore, upon receiving a command to restrict operation or stop charging / discharging, the battery management system sends a confirmation response to the station-level host, indicating that the command has been executed and the battery operating status has been adjusted. For scenarios requiring manual intervention, operators can submit an operation confirmation report through the station-level host's human-machine interface, recording the results of manual inspections or actions. All this effect feedback data is collected in real-time through the communication link between the station-level host and the execution module. Hardware status feedback is uploaded via the CAN bus protocol, while manual confirmation information is input through a touchscreen interface and received by the station-level host monitoring service.

[0048] S4. Optimize and update fire-fighting strategies based on real-time fire characteristic parameters and effect feedback. Specifically, this includes:

[0049] S41. Based on the feedback of the effects after the execution of fire orders and the changes in fire characteristic parameters, generate a deviation evaluation index between the actual fire extinguishing effectiveness and the expected target, specifically including:

[0050] S411. Using the fire characteristic parameters at the moment the fire command is executed as the initial state, the fire command is simulated through the digital twin of the battery energy storage device to generate the expected fire extinguishing efficiency curve.

[0051] The digital twin is specifically a digital twin of the battery energy storage device constructed based on the physical structure, electrical characteristics, and historical operating data of the battery energy storage device, which can simulate the battery thermal runaway process and the response behavior of the fire extinguishing system; the expected fire extinguishing efficiency curve is specifically the trajectory of the predicted fire characteristic parameters changing over time.

[0052] S412. Based on the initial state, and combining the effect feedback after the execution of the fire command and the time-series change data of fire characteristic parameters within the monitoring period, generate the actual fire extinguishing efficiency curve.

[0053] S413. Calculate the difference between the actual fire extinguishing efficiency curve and the expected fire extinguishing efficiency curve at each time point, and integrate to generate a deviation evaluation index.

[0054] In this embodiment, real-time data from various sensors in the battery compartment at the moment of fire command execution are acquired as the initial state of fire characteristic parameters. Based on this initial state and the fire command, a pre-built digital twin of the battery energy storage device is invoked. Through its embedded thermal runaway and fire extinguishing response simulation model, the expected fire extinguishing efficiency curve is derived. This curve, with time as the horizontal axis and each fire characteristic parameter as the vertical axis, predicts the trajectory of each parameter over time under ideal conditions. Based on the same initial state, combined with the effect feedback after the execution of the fire command and the temporal change data of the fire characteristic parameters within the monitoring period, an actual fire extinguishing efficiency curve is generated, reflecting the evolution process of each fire characteristic parameter in the real environment.

[0055] For each fire characteristic parameter, the deviation between the actual fire extinguishing efficiency curve and the expected fire extinguishing efficiency curve at each time point is calculated to form a deviation sequence corresponding to that fire characteristic parameter. The squared error integral of the deviation sequence for each fire characteristic parameter is calculated and then normalized to obtain the sub-item deviation index corresponding to that fire characteristic parameter, which is used to quantify the degree of deviation of a single parameter relative to the expectation. The sub-item deviation indices of all fire characteristic parameters are weighted and fused to obtain a comprehensive deviation index, which is used to quantitatively assess the overall degree of deviation of fire extinguishing efficiency. The sub-item deviation index and the comprehensive deviation index together constitute the deviation assessment index.

[0056] In the weighted fusion process, the weights of each sub-item deviation index are dynamically determined as follows: Contribution data of each fire characteristic parameter in previous fires is extracted from the historical fire database as initial prior knowledge for weight allocation; based on real-time collected time-series data of fire characteristic parameters, the rate of change and standard deviation of each parameter within the monitoring period are calculated as the basis for real-time sensitivity assessment; historical contribution data and real-time sensitivity data are input into a two-layer neural network model based on an attention mechanism. The first layer of this network is a fully connected layer, using the ReLU activation function to map the two types of features to the same high-dimensional space; the second layer is an attention scoring layer, using the Softmax function to output the weight coefficients of each fire characteristic parameter. This weight determination mechanism fully considers the differences in the importance of fire characteristic parameters in historical firefighting scenarios and the response characteristics under the current fire situation, ensuring that the weight allocation conforms to long-term statistical patterns while adapting to real-time operational changes.

[0057] This dynamic weighting method adaptively highlights the dominant role of key fire characteristic parameters in performance evaluation, significantly improving the accuracy of the comprehensive deviation index in representing actual fire extinguishing effectiveness. Compared with fixed-weight schemes, this dynamic weighting method based on multi-source feature fusion overcomes the limitations of fixed parameter importance settings in traditional fire assessments, effectively mitigating evaluation bias caused by dynamic changes in fire scenarios, and thus providing a more accurate decision-making basis for subsequent strategy optimization.

[0058] S42. Input the deviation evaluation index and real-time fire characteristic parameters into the adaptive analysis model to identify the dimensions to be optimized in the fire protection strategy. The dimensions include response triggering conditions, priority rules and threshold parameters.

[0059] The identification of dimensions to be optimized in fire protection strategies specifically includes: extracting the indicator features of deviation evaluation indicators and the time series data of fire characteristic parameters from the adaptive analysis model; combining the indicator features and the time series data to obtain the optimization weights of multiple dimensions; and defining the dimensions whose optimization weights are greater than the preset weight thresholds as the dimensions to be optimized.

[0060] The adaptive analysis model employs a temporal convolutional network (TCNN) in its first stage to extract dynamic pattern features of fire characteristic parameters within the monitoring period. The TCNN captures the changing trends and fluctuations at different time scales. In the second stage, the dynamic pattern features extracted by the TCNN are fused with various deviation evaluation indicators at the feature level to form a joint feature vector containing real-time fire status and performance deviation. In the third stage, the joint feature vector is input into a multi-branch attention module. This module sets an independent branch for each adjustable dimension of the fire strategy. Each branch contains a fully connected layer used to calculate the correlation strength between the joint feature vector and the decision criteria for that dimension. The outputs of all branches are then Softmax normalized to obtain the optimization weights for each dimension. Finally, dimensions with optimization weights greater than a preset weight threshold are identified as dimensions to be optimized.

[0061] In this process, the features extracted by the temporal convolutional network guide the analysis direction of the attention module, while the optimization weights output by the attention module further determine the optimization focus of the rule engine. This analysis model based on a multi-dimensional attention mechanism overcomes the limitations of traditional methods that rely on human experience to set the optimization direction, accurately identifying the dimension in the current fire-fighting strategy that is least suited to the actual fire situation. This provides a clear target for the subsequent optimization of the rule engine, significantly improving the efficiency and accuracy of strategy optimization.

[0062] S43. By leveraging a rules engine and historical fire safety data, dynamically adjust identified dimensions requiring optimization to generate optimized and updated fire safety strategies. Specifically, this includes:

[0063] S431. Based on historical fire protection data of battery energy storage equipment, the rule engine generates multiple candidate adjustment schemes for the target to be optimized. The target to be optimized is the dimension with the highest optimization weight.

[0064] In this embodiment, the rule engine works by combining a generative rule system and case reasoning. First, the rule engine constructs a multi-dimensional retrieval condition, which includes the type of the current fire, the initial feature parameter range, and the type of the dimension to be optimized. Based on the retrieval condition, the engine retrieves the k most similar historical cases to the current scenario from the historical fire database, forming a similar case set.

[0065] By analyzing the similar case set and comparing the fire-fighting strategies and corresponding comprehensive deviation indices adopted in different cases, implicit strategy-effectiveness correlation rules are uncovered. For example, for some fire modes that initially show rapid temperature rise but relatively low smoke concentration, historical cases have shown that strategies that improve temperature monitoring sensitivity generally have significantly better comprehensive deviation indices than strategies that use default parameters. Based on these statistically significant strategy-effectiveness correlation rules, multiple candidate adjustment schemes are generated for the optimization target.

[0066] S432. Input the candidate adjustment schemes into the digital twin of the battery energy storage device in sequence for simulation verification, and obtain the simulated fire extinguishing efficiency curves corresponding to each candidate adjustment scheme.

[0067] S433. Calculate the simulation deviation evaluation index between each simulated fire extinguishing efficiency curve and the expected fire extinguishing efficiency curve, and select the candidate adjustment scheme with the smallest simulation deviation evaluation index as the optimized and updated fire protection strategy.

[0068] This invention generates candidate strategies based on historical data, uses digital twins to achieve low-risk trial and error, and selects the optimal strategy based on quantitative indicators. This breaks through the limitations of traditional methods that rely on fixed rules and manual parameter tuning, and realizes the closed-loop optimization and dynamic adaptive capability of fire protection strategies.

[0069] S5. Apply the optimized and updated fire protection strategy to the subsequent fire protection decisions of the battery energy storage equipment.

[0070] Reference Figure 5 A fire prevention and control system for battery energy storage equipment includes: a data acquisition module, an execution module, a station-level host, and a cloud management platform;

[0071] The data acquisition module communicates with the station-level host and is used to collect real-time fire characteristic parameters of the battery energy storage equipment and upload them to the station-level host.

[0072] The station-level host is connected to the execution module and is used to make fire-fighting decisions and generate fire-fighting instructions based on real-time fire characteristic parameters and preset fire-fighting strategies. The fire-fighting instructions are then sent to the execution module. The host is also used to obtain feedback on the effects of the fire-fighting instructions after they are executed and to upload the feedback and real-time fire characteristic parameters to the cloud management platform.

[0073] The execution module is used to execute the fire command and return the effect feedback data generated after the command execution to the station-level host.

[0074] The cloud management platform communicates with multiple station-level hosts through a message middleware. It optimizes and updates fire protection strategies based on real-time fire characteristic parameters and feedback on the effects of executed fire commands, and then distributes the optimized and updated strategies to the corresponding station-level hosts. Specifically, it includes the following units:

[0075] The deviation assessment unit is used to generate deviation assessment indicators between the actual fire extinguishing effectiveness and the expected target based on the effect feedback after the execution of fire orders and changes in fire characteristic parameters.

[0076] The dimension identification unit is used to identify the dimensions to be optimized in fire protection strategies based on deviation evaluation indicators and fire characteristic parameters using an adaptive analysis model.

[0077] The rule adjustment unit is used to dynamically adjust the identified dimensions to be optimized by coordinating the rule engine with historical fire protection data, and generate optimized and updated fire protection strategies.

[0078] The cloud management platform will also distribute the optimized and updated fire protection strategy to the station-level host;

[0079] The station-level host is also used to receive the optimized and updated fire protection strategy issued by the cloud management platform, and apply the optimized and updated fire protection strategy to the subsequent fire protection decisions of the battery energy storage equipment.

[0080] The cloud management platform also provides the following services: monitoring and alarm services, which display the safety status of each energy storage device in real time and present alarm information in the form of data, graphs, text, and sound; BI analysis, which provides data dashboards and reporting functions and supports multi-dimensional data analysis; remote dispatch services, which support remote intervention and control of lower-level station hosts; notification and announcement services, which are used to issue management instructions and announcements to station hosts; and system management services, which are responsible for the management of platform users, permissions, and basic information.

[0081] As an edge computing and control unit, the station-level host possesses protocol parsing, data processing, storage, and forwarding capabilities, undertaking data acquisition, command issuance, and local decision-making functions. The cloud management platform aggregates, analyzes, and optimizes data from multiple stations, forming a two-tier architecture of "edge execution + cloud optimization." Internally, the station-level host employs a dual-server architecture with one primary and one backup server. Automatic switching of video signals and services is achieved through a KVM switch and remote controller, ensuring high system availability. Through the collaborative work of these modules, this system not only achieves real-time fire monitoring and intelligent response but also significantly improves the accuracy and adaptability of fire response through a data-driven strategy optimization mechanism, effectively reducing the fire risk of energy storage power stations.

[0082] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A battery energy storage device fire prevention and control method, characterized in that, The method comprises the following steps: S1, collecting real-time fire characteristic parameters of the battery energy storage device; S2, making a fire-fighting decision based on the real-time fire characteristic parameters and a preset fire-fighting strategy and generating a fire-fighting instruction; S3, obtaining an effect feedback after the fire-fighting instruction is executed; S4, optimizing and updating the fire-fighting strategy based on the real-time fire characteristic parameters and the effect feedback; S5, applying the optimized and updated fire-fighting strategy to subsequent fire-fighting decisions of the battery energy storage device; Step S4 comprises: S41, generating a deviation evaluation index of actual fire extinguishing efficiency and an expected target based on the effect feedback after the fire-fighting instruction is executed and the change of the fire characteristic parameters; S42, inputting the deviation evaluation index and the real-time fire characteristic parameters into a self-adaptive analysis model to identify a dimension to be optimized in the fire-fighting strategy, wherein the dimension comprises a response trigger condition, a priority rule and a threshold parameter; S43, dynamically adjusting the identified dimension to be optimized through a rule engine in cooperation with historical fire-fighting data to generate an optimized and updated fire-fighting strategy; S41 specifically comprises: S411, taking the fire characteristic parameters at the time when the fire-fighting instruction is executed as an initial state, simulating the fire-fighting instruction through a digital twin of the battery energy storage device to generate an expected fire extinguishing efficiency curve, wherein the expected fire extinguishing efficiency curve is specifically a trajectory of the predicted fire characteristic parameters changing over time; S412, generating an actual fire extinguishing efficiency curve based on the initial state, the effect feedback after the fire-fighting instruction is executed and the time sequence change data of the fire characteristic parameters within a monitoring period; S413, calculating the fire characteristic parameter difference between the actual fire extinguishing efficiency curve and the expected fire extinguishing efficiency curve at each time point and integrating to generate the deviation evaluation index; The identification of the dimension to be optimized in the fire-fighting strategy specifically comprises: The self-adaptive analysis model extracts index features of the deviation evaluation index and time sequence data of the fire characteristic parameters, combines the index features and the time sequence data to obtain optimization weights of multiple dimensions, and defines a dimension whose optimization weight is greater than a preset weight threshold as a dimension to be optimized; S43 specifically comprises: S431, generating multiple candidate adjustment schemes for the target to be optimized from the rule engine based on the historical fire-fighting data of the battery energy storage device, wherein the target to be optimized is specifically a dimension to be optimized with the largest optimization weight; S432, inputting the candidate adjustment schemes into the digital twin of the battery energy storage device in sequence for simulation verification to obtain a simulation fire extinguishing efficiency curve corresponding to each candidate adjustment scheme; S433, calculating simulation deviation evaluation indexes between each simulation fire extinguishing efficiency curve and the expected fire extinguishing efficiency curve, and selecting a candidate adjustment scheme with the smallest simulation deviation evaluation index as the optimized and updated fire-fighting strategy.

2. The battery energy storage device fire prevention and control method of claim 1, wherein, The digital twin is specifically a digital twin of the battery energy storage device which can simulate the process of battery thermal runaway and the response behavior of the fire extinguishing system based on the physical structure, electrical characteristics and historical operation data of the battery energy storage device.

3. A battery energy storage device fire prevention and control system, characterized in that, The method is applied to the battery energy storage device fire prevention and control method of any one of claims 1-2, comprising a data acquisition module, an execution module, a station-level host and a cloud management platform. The data acquisition module is in communication connection with the station-level host, and is configured to acquire real-time fire characteristic parameters of the battery energy storage device and upload the real-time fire characteristic parameters to the station-level host; The station-level host is in communication connection with the execution module, and is configured to make a fire-fighting decision based on the real-time fire characteristic parameters and a preset fire-fighting strategy, generate a fire-fighting instruction, and send the fire-fighting instruction to the execution module; the station-level host is also configured to acquire an effect feedback after the fire-fighting instruction is executed, and upload the effect feedback and the real-time fire characteristic parameters to the cloud management platform; The execution module is configured to execute the fire-fighting instruction and return effect feedback data generated after the fire-fighting instruction is executed to the station-level host; The cloud management platform is in communication connection with the station-level host, and is configured to receive the real-time fire characteristic parameters and the effect feedback data, and perform the following optimization process: A deviation evaluation unit is configured to generate a deviation evaluation index of actual fire extinguishing efficiency and an expected target based on the effect feedback data and the change of the fire characteristic parameters after the fire-fighting instruction is executed; A dimension identification unit is configured to input the deviation evaluation index and the real-time fire characteristic parameters into a self-adaptive analysis model, and identify a dimension to be optimized in the fire-fighting strategy, the dimension including a response trigger condition, a priority rule and a threshold parameter; A rule adjustment unit is configured to dynamically adjust the identified dimension to be optimized by a rule engine in cooperation with historical fire-fighting data, and generate an updated fire-fighting strategy; The cloud management platform is also configured to send the updated fire-fighting strategy to the station-level host; The station-level host is also configured to receive the updated fire-fighting strategy sent by the cloud management platform, and apply the updated fire-fighting strategy to subsequent fire-fighting decisions of the battery energy storage device.

Images