Active safety fire extinguishing system of energy storage battery with aerogel

Abstract

The invention belongs to the technical field of energy storage battery safety fire fighting, and particularly relates to an energy storage battery active safety fire fighting system with aerogel, which comprises a sensing layer, a transmission layer, a platform layer and an execution layer, the sensing layer collects operation and environment data of the energy storage battery, and the transmission layer transmits the data to the platform layer; the platform layer comprises three modules: a data preprocessing module processes received data; the risk identification module evaluates the preprocessed data in real time according to a preset strategy by using two active identification algorithms to obtain a risk evaluation result; the fire-fighting strategy making module determines an optimal fire-fighting strategy accordingly; and the execution layer receives the strategy and carries out control and aerogel fire extinguishing operation on the energy storage battery. The problem that in the prior art, an energy storage battery lacks an effective technical means in the field of active fire fighting can be solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of energy storage battery safety fire protection, and in particular relates to an energy storage battery active safety fire protection system with aerogel. Background Art

[0002] Against the backdrop of an accelerating global energy transition, energy storage batteries are becoming increasingly important as a key enabler of new power systems. With the large-scale development and utilization of renewable energy sources like solar and wind, energy storage batteries can effectively mitigate the volatility and intermittency of renewable energy generation, ensuring a stable supply and efficient utilization of energy. They play an indispensable role in power systems, transportation, distributed energy, and other fields.

[0003] However, with the widespread use of energy storage batteries, their safety has gradually become a focus of industry attention. Lithium-ion batteries, for example, have been widely used in the energy storage field due to their high energy density and rapid charge and discharge speeds. However, they also carry the risk of thermal runaway. When lithium batteries are affected by factors such as overcharge, over-discharge, overheating, mechanical abuse, or internal short circuits, a series of complex chemical reactions occur within the battery, causing a rapid increase in temperature, which in turn triggers thermal runaway, potentially leading to serious safety accidents such as explosions.

[0004] In the existing technology, the safety identification of energy storage batteries is usually achieved by developing multiple risk identification algorithms, but this fails to fully utilize the unique advantages of active safety fire protection in energy storage batteries. At the same time, the developed risk identification algorithms are difficult to meet the needs of different types of energy storage batteries and complex operating environments. Therefore, this problem needs to be solved urgently. Summary of the Invention

[0005] The technical problem solved by the present invention is to provide an active safety fire protection system of an energy storage battery with aerogel, so as to solve the problem in the prior art of lack of effective technical means for energy storage batteries in the field of active fire protection.

[0006] The basic solution provided by the present invention is an active safety fire protection system for energy storage batteries with aerogel, comprising a perception layer, a transmission layer, a platform layer, and an execution layer. The perception layer is used to collect perception data of the energy storage battery operation process and operating environment. The transmission layer is used to transmit the perception data collected by the perception layer to the platform layer. The platform layer includes a data preprocessing module, a risk identification module, and a fire protection strategy formulation module.

[0007] The data preprocessing module is used to preprocess the received perception data to obtain preprocessed perception data;

[0008] The risk identification module deploys a first active risk identification algorithm and a second active risk identification algorithm, and the risk identification module calls the first active risk identification algorithm and the second active risk identification algorithm to perform real-time risk assessment on the pre-processed perception data according to a preset active risk identification strategy to obtain a risk assessment result of the energy storage battery;

[0009] The fire fighting strategy formulation module is used to formulate the best fire fighting strategy according to the risk assessment results of the energy storage battery;

[0010] The execution layer is used to receive the optimal fire fighting strategy and implement corresponding control and aerogel fire extinguishing operations on the energy storage battery according to the optimal fire fighting strategy.

[0011] The technical principle of the present invention is as follows: the technical solution of this application realizes active safety fire protection of energy storage batteries through a four-layer architecture, and the core logic is "data-driven, precise decision-making"; among them, the perception layer is responsible for collecting perception data of the operation process and operating environment of the energy storage battery, providing a basic data source for subsequent analysis; the transmission layer undertakes the data transmission function, accurately transmits the perception data collected by the perception layer to the platform layer, and ensures the smoothness of data flow; the platform layer is the core decision-making link, and the data preprocessing module first preprocesses the received perception data to obtain preprocessed perception data, and then calls the internally deployed first risk active identification algorithm and second risk active identification algorithm through the risk identification module, and performs real-time risk assessment on the preprocessed perception data according to the preset risk active identification strategy, outputs the energy storage battery risk assessment result, and finally the fire protection strategy formulation module formulates the optimal fire protection strategy based on the risk assessment result; the execution layer receives the optimal fire protection strategy issued by the platform layer, and implements corresponding control and aerogel fire extinguishing operations on the energy storage battery according to the strategy, forming a complete process from data collection to risk disposal.

[0012] The technical effect corresponding to the technical principle of this application: The technical solution of this application effectively improves the safety protection capability of energy storage batteries through clear layered collaboration and precise algorithm application:

[0013] 1. In the risk identification phase, relying on dual-algorithm collaborative evaluation and preset identification strategies, it is possible to conduct real-time analysis of energy storage battery operating data and accurately obtain risk assessment results, thus avoiding the problems of delayed or inaccurate risk judgment in traditional protection methods;

[0014] 2. In the fire disposal phase, the optimal firefighting strategy formulated based on the risk assessment results can ensure that the control and firefighting operations implemented by the execution layer are targeted, avoiding unnecessary disposal actions and responding promptly when risks occur, effectively addressing safety hazards through aerogel-related firefighting operations;

[0015] 3. From the overall system, the ordered cooperation of the perception layer, the transmission layer, the platform layer and the execution layer realizes the closed-loop management from data acquisition to fire fighting execution, ensures that the energy storage battery can be quickly identified and disposed through the adaptive fire fighting measures when a safety risk occurs, significantly reduces the probability of energy storage battery safety accidents, reduces the impact of accidents, and provides reliable protection for the active safety operation of the energy storage battery.

[0016] Further, the perception layer includes a first perception layer and a second perception layer, the first layer perception data collected by the first perception layer includes monitoring data of an environmental monitoring sensor, monitoring data of an energy storage battery operating parameter monitoring sensor, and aerogel fire fighting means data; the second layer perception data collected by the second perception layer includes aerogel layout state data, fire fighting risk area data, and energy storage battery combustion prediction data.

[0017] Beneficial effect: In the present application, the perception layer is divided into a first perception layer and a second perception layer. From the data acquisition dimension, the first perception layer is specially designed to collect monitoring data of an environmental monitoring sensor, monitoring data of an energy storage battery operating parameter monitoring sensor, and aerogel fire fighting means data, to ensure that the core parameters of battery operation (such as operating state indicators) and the basic environmental state can be mastered in real time, and to avoid risk misjudgment due to missing key operating and environmental data; the second perception layer focuses on aerogel layout state, fire fighting risk area and energy storage battery combustion prediction data, directly relates the state of the core components of the fire fighting system and the key information of risk evolution, and makes up for the short board of missing fire fighting supporting state, risk area positioning and combustion trend data of a single perception layer; the combination of the two types of perception data makes the perception data transmitted to the platform layer cover the full dimension of "battery operation-environmental state-fire fighting components-risk trend", so that the preprocessed data obtained by the data preprocessing module is more complete, and then when the risk identification module calls the double algorithm evaluation, the risk can be judged based on a more comprehensive data source, and the evaluation deviation caused by one-sided data is reduced.

[0018] Further, the risk identification module includes a first risk identification unit and a second risk identification unit, the first risk identification unit receives the monitoring data of the environmental monitoring sensor, the monitoring data of the energy storage battery operating parameter monitoring sensor, and the aerogel fire fighting means data in the first layer perception data, calls a first risk active identification algorithm for first risk calculation, and if the first risk does not satisfy a preset first risk threshold, generates a low risk evaluation result of the energy storage battery;

[0019] If the first risk satisfies the preset first risk threshold and does not satisfy a preset second risk threshold, a medium risk evaluation result of the energy storage battery is generated;

[0020] The second risk identification unit is configured to, when the first risk exceeds the second risk threshold, call a second risk active identification algorithm based on the aerogel layout state data, the fire risk area data, and the energy storage battery combustion prediction data to perform second risk judgment, and generate a high-risk evaluation result of the energy storage battery.

[0021] Beneficial effects: In the present application, the risk identification module is divided into first and second risk identification units, and different risk thresholds and associated perception data are used to call algorithms, which realizes hierarchical and accurate evaluation of the risk of energy storage batteries, effectively improves the logicality and accuracy of risk judgment, and provides more targeted basis for subsequent fire fighting strategy formulation.

[0022] Further, the construction process of the first risk active identification algorithm is as follows:

[0023] A fault scoring function is constructed based on the monitoring data of the environmental monitoring sensor in the first layer of perception data and the monitoring data of the energy storage battery operation parameter monitoring sensor, and the expression is as follows:

[0024]

[0025] Wherein, M represents the feature dimension of the monitoring data in the first layer of perception data, ReLU() represents a nonlinear activation function, λ m represents a feature weight vector, f m (t) represents a real-time feature value, μ m represents the historical mean of the feature, σ m represents the standard deviation of the feature, and Entropy(C(t)) represents the entropy value of the fault code sequence, C(t) represents the fault code set in a period of time, and a represents the entropy weight.

[0026] Based on the scoring result of the fault scoring function, a clustering identification algorithm is called to output the fault category, and the expression is as follows:

[0027]

[0028] X={S edge (t)|f1(t)、f2(t)……f m (t)}

[0029] ∈ k =0.08+0.0001×Cycle

[0030] Wherein, x i , x j represents a sample point in the feature vector X, dist() represents the distance calculation between sample points x i , x j , and ∈ kDynamic clustering radius, Cycle represents the number of energy storage battery cycles, represents ∈ k Sample set in the neighborhood, MinPts k represents the minimum number of neighborhood points

[0031] According to the scoring result of the fault scoring function and the clustering recognition result, output the first risk calculation result, the expression is:

[0032]

[0033] Where, β represents the fault scoring weight, K represents the fault type, γ k represents the fault severity coefficient, ClusterProb(k) represents the probability that the sample belongs to the kth fault.

[0034] Beneficial effects: The first risk active identification algorithm significantly improves the accuracy, dynamic adaptability and result reliability of the first risk calculation based on the first layer of perception data by constructing the complete process of the fault scoring function, the clustering recognition algorithm and the first risk calculation result expression, providing a scientific and rigorous calculation basis for the risk identification module to divide the low risk and medium risk evaluation results.

[0035] Further, the construction process of the second risk active identification algorithm is:

[0036] A thermal runaway early warning index model of the energy storage battery is constructed, a mechanism-data fusion model is adopted, and the expression is:

[0037]

[0038] Wherein, is the mechanism term, A represents the frequency factor, E a represents the activation energy, R represents the gas constant, T(t) represents the real-time core temperature of the energy storage battery, LSTM() represents the long short-term memory network with a hidden layer of 32, H(t) represents the time sequence feature, Sigmoid() represents the activation function, and Z(t) represents the battery expansion amount monitoring value.

[0039] A safety situation awareness model of the energy storage battery system is constructed, and the expression is:

[0040]

[0041] Wherein, GNN represents a graph neural network, G represents a topology graph of the energy storage battery pack, is the input of the graph neural network, W is the edge weight, represents the thermal conductivity coefficient between the energy storage battery modules; Couple(i,j) represents the fault coupling degree between the ith energy storage battery module and the jth energy storage battery module, δ represents the coupling coefficient, and N represents the number of adjacent energy storage battery modules.

[0042] The second risk is calculated based on the thermal runaway early warning index model and the energy storage battery system safety situation awareness model, and the expression is:

[0043] R2 = (η × P TR + (1 - η) × S cloud ) × Exp(t persist ) + κ × Damage(t)

[0044] wherein η represents the thermal runaway probability weight, Exp(t persist ) represents the continuous risk amplification term, t persist represents the risk duration in minutes, Damage(t) represents the estimated battery damage degree, and κ represents the damage weight.

[0045] Beneficial effects: The second risk active identification algorithm can accurately calculate the second risk by constructing the thermal runaway early warning index model, the energy storage battery system safety situation awareness model and the second risk calculation expression, and provide reliable basis for generating the energy storage battery high-risk evaluation result.

[0046] Further, the best fire fighting strategy is formulated according to the energy storage battery risk evaluation result in the fire fighting strategy formulation module, and the specific formulation is:

[0047] If the energy storage battery risk evaluation result is the energy storage battery low-risk evaluation result, a regular operation and maintenance reminder signal is generated;

[0048] If the energy storage battery risk evaluation result is the energy storage battery medium-risk evaluation result, an energy storage battery operation adjustment command is generated, and an energy storage battery operation strategy is generated;

[0049] If the energy storage battery risk evaluation result is the energy storage battery high-risk evaluation result, after calculating the cut-off delay time, a corresponding area energy storage battery cut-off operation command is generated and transmitted to the execution layer for execution, and the aerogel injection amount is calculated, an aerogel injection command for the corresponding area of the energy storage battery is generated and transmitted to the execution layer for execution.

[0050] Beneficial effects: The fire fighting strategy module formulates differentiated optimal fire fighting strategies according to different energy storage battery risk assessment results, can realize accurate response to different risk levels, effectively improves the pertinence and efficiency of energy storage battery safety management, first generates a regular operation and maintenance reminder signal for low risk assessment results, can avoid risk expansion through daily maintenance in advance, reduces unnecessary emergency disposal; secondly, the running adjustment command and running strategy are generated for the medium risk assessment results, which can control the risk development in time through adjusting the running state to avoid the risk upgrading to high risk; finally, the cut-off delay time is calculated and the corresponding area cut-off operation command is generated for the high risk assessment results, and the aerogel injection amount is calculated and the injection command is generated, which can quickly cut off the risk source and accurately implement the fire extinguishing operation, maximally reduce the damage of high risk to the energy storage battery, and overall ensure that there are adaptive disposal schemes under each risk level to ensure the safe operation of the energy storage battery.

[0051] Further, the calculation expression of the cut-off delay time is:

[0052] t cut = 0.8-0.005xR2

[0053] Wherein, R2 is the second risk calculation result;

[0054] The aerogel injection amount calculation expression is:

[0055]

[0056] Wherein, R2 is the second risk calculation result, Area represents the risk area, which is obtained by positioning the risk area in the energy storage battery group topology graph G, P TR represents the thermal runaway early warning index.

[0057] Beneficial effects: Through the calculation of the cut-off delay time and the aerogel injection amount, the fire fighting operation instruction corresponding to the high risk assessment result is more accurate and controllable, which further improves the scientificity and effectiveness of fire fighting disposal in high risk scenarios.

[0058] Further, the energy storage battery operation parameter monitoring sensor includes a main energy storage battery operation parameter monitoring sensor and a secondary energy storage battery operation parameter monitoring sensor, the main energy storage battery operation parameter monitoring sensor is arranged at a key node of the energy storage battery system, and the secondary energy storage battery operation parameter monitoring sensor is arranged at a secondary node position of the key node of the energy storage battery.

[0059] The operation parameters between the main energy storage battery operation parameter monitoring sensor at the key node and the secondary energy storage battery operation parameter monitoring sensor at the secondary node are associated through an association algorithm.

[0060] Beneficial effects: For the collection points of operating parameters in the energy storage battery system, key nodes and secondary nodes are designed. The data collected by sensors at key nodes are more accurate, but they are easy to be damaged and have a high destruction rate. Therefore, secondary nodes are selected near the key nodes, and the sensor collection data at the key nodes are associated with the sensor collection data at the secondary nodes. Even if the sensors at the technical key nodes are damaged, the energy storage battery operating parameters can still be collected from the sensors at the secondary nodes, ensuring the reliability of the perception layer. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 Schematic diagram of the system architecture of an embodiment of the present invention;

[0062] Figure 2 Schematic diagram of the layout of the energy storage battery system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0063] The following is further described in detail through specific implementation methods:

[0064] Lithium-ion battery energy storage systems, a key technology for modern energy storage and conversion, operate based on the reversible intercalation and deintercalation of lithium ions between the positive and negative electrodes. During the charging phase, electrical energy from an external power source causes lithium ions to escape from the positive electrode material's lattice, pass through the electrolyte, and then through the separator before being embedded in the negative electrode's lattice. This process converts electrical energy into chemical energy and stores it. During the discharge phase, lithium ions escape from the negative electrode, reverse through the electrolyte and separator, and re-embed into the positive electrode, where chemical energy is converted back into electrical energy and released to power the external load.

[0065] Lithium battery explosion accidents usually result from the interaction of a series of complex internal reactions and external factors, among which overcharging, over-discharging, short circuit and thermal runaway are the most critical disaster-causing factors. Based on this, traditional energy storage battery fire protection technology has many limitations when dealing with lithium battery explosions, and it is difficult to meet the growing safety needs of lithium battery energy storage systems. Aerogel mainly reduces the risk of lithium battery explosions through two key mechanisms: thermal insulation and flame retardancy, providing reliable safety protection for energy storage battery systems. How to make effective use of aerogel in the field of active safety fire protection of energy storage batteries? This application provides an energy storage battery active safety fire protection system with aerogel, and its embodiments are as follows:

[0066] Example 1:

[0067] Example 1 is basically as shown in the attached Figure 1 Figure 2 shows an active safety fire protection system for energy storage batteries with aerogel, which includes a perception layer, a transmission layer, a platform layer, and an execution layer. The perception layer is used to collect perception data about the operation process and operating environment of the energy storage battery.

[0068] In the present embodiment, the perception layer includes a first perception layer and a second perception layer, the first layer of perception data collected by the first perception layer includes monitoring data of environmental monitoring sensors, monitoring data of energy storage battery operation parameter monitoring sensors, aerogel fire fighting means data; wherein, as shown in Figure 2 The sensors for monitoring the environment mainly use temperature and humidity sensors, which monitor the environmental temperature of the energy storage battery cabin and the environmental temperature of the energy storage battery integrated area, and are deployed on the monitoring station; the monitoring of the energy storage battery operation parameters uses temperature sensors, voltage sensors, current sensors, etc., the temperature sensor uses a patch type temperature sensor to closely adhere to the battery shell to accurately measure the battery surface temperature, and the voltage sensor and the current sensor are arranged on the transmission circuit of the energy storage battery to monitor the charge and discharge voltage and current of the energy storage battery in real time; the aerogel fire fighting means data includes whether aerogel is used in the fire fighting means of the energy storage battery system.

[0069] The second layer of perception data collected by the second perception layer includes aerogel layout state data, fire risk area data, and energy storage battery combustion prediction data; in the present embodiment, the aerogel layout state data includes the layout state of the aerogel spray pipeline in the energy storage battery cabin and the energy storage battery area, the characteristics (type, structure, thickness) of the aerogel, and the fire risk area data is obtained by analyzing the spatial layout of the energy storage battery system, the connection mode between the battery modules, and the distribution of the aerogel, etc., as shown in Figure 2 The energy storage battery module cluster diagram, to obtain the area range that may be affected by fire; the energy storage battery combustion prediction includes energy storage battery combustion size prediction and heat prediction, the energy storage battery combustion size prediction considers the type, capacity, thermal runaway temperature, and combustible gas combustion rate of the energy storage battery, therefore, computational fluid dynamics is used to predict the size and development trend of the energy storage battery combustion flame; the heat prediction is obtained by analyzing the heat generation rate, heat conduction coefficient, and heat insulation performance of aerogel during the thermal runaway process of the energy storage battery, thereby predicting the heat distribution and change trend at different positions of the fire scene.

[0070] The transmission layer then transmits the perception data collected by the perception layer to the platform layer, in the present embodiment, the transmission layer uses a combination of wired and wireless transmission methods, for sensors with close distance and large data transmission volume, such as the data transmission between the battery management system and the platform layer, industrial Ethernet and other wired transmission methods are used to ensure the stability and real-time performance of data transmission; for some sensors that are distributed relatively dispersedly and difficult to wire, such as temperature sensors and gas sensors, wireless transmission methods are used, such as ZigBee wireless communication technology; these wireless sensor nodes transmit data to the sink node through the ad hoc network, and then the sink node uploads the data to the platform layer through wired network or 5G communication technology.

[0071] In the platform layer, a data preprocessing module, a risk identification module and a fire fighting strategy making module are included, wherein the data preprocessing module is configured to preprocess the received sensing data to obtain preprocessed sensing data; in this embodiment, the data preprocessing includes data cleaning, normalization processing, standardization processing and feature extraction processing, and the purpose of data cleaning is to remove noise and outliers in the data, since the sensor may be affected by electromagnetic interference, environmental changes and other factors during actual operation, and there may be some errors or unreasonable values in the collected data, therefore, the filtering algorithm is used to smooth the abnormal data and remove noise interference;

[0072] The normalization processing and the standardization processing are configured to unify the dimensions and distribution ranges of different types of data, so as to facilitate subsequent data analysis and model training; the feature extraction is configured to extract characteristic quantities reflecting the operating state of the energy storage battery, such as the charge and discharge rate of the battery, the temperature change rate, the gas concentration change, etc., which can intuitively reflect the operating trend and potential risks of the battery.

[0073] The risk identification module deploys a first risk active identification algorithm and a second risk active identification algorithm, and the risk identification module calls the first risk active identification algorithm and the second risk active identification algorithm to perform real-time risk assessment on the preprocessed sensing data according to a preset risk active identification strategy, to obtain a risk assessment result of the energy storage battery; wherein the risk identification module includes a first risk identification unit and a second risk identification unit, the first risk identification unit receives monitoring data of an environmental monitoring sensor and monitoring data of an energy storage battery operating parameter monitoring sensor in the first layer sensing data, calls the first risk active identification algorithm to perform first risk calculation, and if the first risk does not satisfy a preset first risk threshold, generates a low risk assessment result of the energy storage battery;

[0074] If the first risk satisfies the preset first risk threshold and does not satisfy a preset second risk threshold, a medium risk assessment result of the energy storage battery is generated;

[0075] The second risk identification unit is configured to, if the first risk exceeds the second risk threshold, call the second risk active identification algorithm based on aerogel layout state data, fire fighting risk area data and energy storage battery combustion prediction data to perform second risk judgment, and generate a high risk assessment result of the energy storage battery.

[0076] In this embodiment, the first risk active identification algorithm is a basic link of the entire energy storage battery active safety fire fighting system, which aims to comprehensively and accurately capture potential risks in the operating process of the energy storage battery, therefore, the construction process of the first risk active identification algorithm is as follows:

[0077] A fault scoring function is constructed based on the monitoring data of the environmental monitoring sensor in the first layer of perception data and the monitoring data of the energy storage battery operation parameter monitoring sensor, and the expression is:

[0078]

[0079] wherein M represents the feature dimension of the monitoring data in the first layer of perception data, in the embodiment, the feature dimension M = 6, including temperature deviation rate, voltage fluctuation coefficient, current ripple factor, internal resistance change rate, aerogel injection response delay, and humidity deviation degree, ReLU() represents a nonlinear activation function, λ m represents a feature weight vector, f m (t) represents a real-time feature value, μ m represents the historical mean value of the feature, σ m represents the standard deviation of the feature, and Entropy(C(t)) represents the entropy value of the fault code sequence, C(t) represents the fault code set in the last 5 minutes, and a represents the entropy weight.

[0080] Based on the scoring result of the fault scoring function, a clustering recognition algorithm is called to output the fault category, and the expression is:

[0081]

[0082] X = {S edge (t)|f1(t), f2(t) … f m (t)}

[0083] ∈ k = 0.08 + 0.0001 * Cycle

[0084] wherein x i , x j represents a sample point in the feature vector X, in the embodiment, the feature vector space X includes S edge (t) and 6-dimensional original features, dist() represents the distance calculation between sample points x i , x j , ∈ k represents a dynamic clustering radius, Cycle represents the cycle number of the energy storage battery, represents the sample set in the ∈ k neighborhood, MinPts k represents the minimum neighborhood point number, which is set to 8 for NCM cells and 6 for LFP cells according to the battery type;

[0085] According to the scoring result of the fault scoring function and the clustering recognition result, a first risk calculation result is output, and the expression is:

[0086]

[0087] Wherein, β represents the fault scoring weight, K represents the fault type. In this embodiment, the preset fault types include overtemperature, overshoot, internal short circuit, poor contact, and aerogel failure. k represents the fault severity coefficient. In this embodiment, the internal short circuit is set to 1.8, the overtemperature is set to 1.5, and the others are set to 1.0. ClusterProb(k) represents the probability that the sample belongs to the kth type of fault, which is updated through Bayesian estimation.

[0088] After the first risk is calculated by the first risk active identification algorithm, it is compared with the preset first risk threshold. If it is lower than the first risk threshold, it means that the energy storage battery system has no faults and risks, and it is a low-risk assessment result. It is only necessary to adopt a strategy of regularly pushing the system status; when the first risk exceeds the first risk threshold but does not exceed the second risk threshold, it means that there is a fault in the operation of the energy storage battery system, and these faults may pose a risk of fire or may just be a routine system fault. Regardless of which one, it is a risk assessment result in the energy storage battery. On the one hand, passive safety measures need to be taken, such as adjusting the energy storage battery operation strategy, starting the backup energy storage battery, etc. On the other hand, operation and maintenance personnel need to be arranged for timely maintenance.

[0089] When the first risk exceeds the preset second risk threshold, it means that there is a very high risk. At this time, it is necessary to combine new factors to generate a new risk judgment, that is, to adopt the second risk active identification algorithm.

[0090] Therefore, the first active risk identification algorithm of the present application adopts the technical means of active identification and implementation of passive safety measures to achieve the fire safety of energy storage batteries. At the same time, the first active risk identification algorithm of the present application is deployed on the edge side, which has the defect of computing power constraint. Based on this, the first active risk identification algorithm of the present application adopts the hierarchical collaboration of fault scoring function and clustering algorithm to solve the problem of computing power constraint on the edge side, forming a process of screening first and then classification. Specifically, the first step is to perform risk coarse screening through the fault scoring function, and then through S edge (t) Calculate the comprehensive risk analysis for each monitoring period, when S edge When (t) is less than the preset first risk threshold, it is directly judged as a fault-free contribution without entering the clustering process. edge (t) When the risk reaches or exceeds the preset first threshold, S edge (t) Instead of feeding its 6-dimensional feature dimension into the clustering algorithm, this step can filter out 80% of the low-risk samples, greatly reducing the computational complexity of the clustering algorithm.

[0091] Then the second step clustering algorithm is used to do fault subclassification, first using the dynamic clustering radius ∈ k and the minimum number of neighborhood points MinPtsk , classify samples with similar risk characteristics into the same category, such as over-temperature faults and internal short-circuit faults; then output the probability ClusterProb(k) of each sample belonging to each type of fault; therefore, the clustering algorithm solves the problem of relying solely on S edge (t) The problem of being unable to distinguish fault types, while the clustering algorithm can accurately locate the fault type through similarity analysis of the feature vector space X.

[0092] For the second risk active identification algorithm, its construction process is as follows:

[0093] A thermal runaway warning index model for energy storage batteries is constructed using a mechanism-data fusion model, and the expression is:

[0094]

[0095] in, is the mechanism term, A represents the frequency factor, E a represents activation energy, R represents gas constant, T(t) represents real-time core temperature of energy storage battery, LSTM() represents long short-term memory network with 32 hidden layers, H(t) represents time series feature, Sigmoid() represents activation function, and Z(t) represents battery expansion monitoring value;

[0096] Construct a safety situation awareness model for the energy storage battery system, and the expression is:

[0097]

[0098] Where GNN stands for graph neural network, G represents the energy storage battery group topology graph, which is the input of the graph neural network, W is the edge weight, and represents the thermal conductivity coefficient between energy storage battery modules; Couple(i,j) represents the fault coupling degree between the i-th energy storage battery module and the j-th energy storage battery module, δ represents the coupling coefficient, and N represents the number of adjacent energy storage battery modules.

[0099] The second risk is calculated based on the thermal runaway warning index model and the energy storage battery system safety situation awareness model. The expression is:

[0100] R2=(η×P TR +(1-η)×S cloud )×Exp(t persist )+κ×Damage(t)

[0101] Where η represents the probability weight of thermal runaway, Exp(t persist ) represents the continuous risk amplification term, t persist represents the risk duration in minutes, Damage(t) represents the estimated battery damage, and κ represents the damage weight.

[0102] In this embodiment, the second risk active identification algorithm is started, which means that there is a risk of fire in the energy storage battery. At this time, adjusting the operation strategy of the energy storage battery system has no effect, and the active fire prevention strategy needs to be implemented. Therefore, in addition to the factors of the energy storage battery operation parameters, environmental parameters, and aerogel fire fighting means that need to be considered in the first risk active identification algorithm, the factors of the fire range, size, and heat influence when the fire occurs also need to be considered. Based on this, in the second risk active identification algorithm of the present application, the strategy implemented is to upgrade from a single module early warning to a system-level prevention and control. Specifically, by P TR triggering-amplification of S cloud , the specific process is as follows:

[0103] First, P TR trigger local risk early warning. When the P TR of a certain energy storage battery module exceeds the preset second risk threshold value, it is determined that this module is a core risk source, and its position, P TR value, number, and other information are sent to the GNN model of S cloud .

[0104] Second, S cloud amplify the overall risk situation. The GNN model takes the core risk source as the starting point, combines Couple(i,j), calculates the risk diffusion path and system affected range, and amplifies the overall situation value through . Taking a specific example, the core risk source P TR = 2.0 (high risk), and there are three adjacent modules, then Couple(i,j) is corrected to 0.6, 0.5, and 0.4 respectively, δ = 0.08, ∑Couple(i,j) = 1.5, then 1 + 0.08x1.5 = 1.12, the basic situation value (such as 5.0) output by the GNN model is multiplied by 1.12, S cloud = 5.6, so the overall situation is improved by 12%, reflecting the diffusion influence of the core risk source.

[0105] Therefore, in the calculation process of the second risk active identification algorithm in this embodiment, the local high risk is quickly converted into a global situation early warning, avoiding the situation of ignoring the system-level chain risk by only focusing on a single module.

[0106] After further calculating the second risk, a high-risk evaluation result is generated according to the second risk and transmitted to the fire fighting strategy module to develop a corresponding fire fighting strategy.

[0107] Therefore, the second risk active identification algorithm in the present application uses the technical means of active identification and active fire fighting to improve the active safety and fire fighting capability of the energy storage battery.

[0108] The fire fighting strategy formulation module formulates the best fire fighting strategy based on the risk assessment results of the energy storage battery. Specifically:

[0109] If the energy storage battery risk assessment result is a low risk assessment result for the energy storage battery, an operation and maintenance reminder signal is generated regularly; after the execution layer receives the operation and maintenance reminder signal, it arranges the corresponding operation and maintenance personnel to regularly inspect and maintain the energy storage battery system.

[0110] If the energy storage battery risk assessment result is a medium risk assessment result for the energy storage battery, an energy storage battery operation adjustment command is generated, and an energy storage battery operation strategy is generated; after the execution layer receives the energy storage battery operation adjustment command, it automatically adjusts the energy storage battery operation strategy, such as adjusting the energy storage battery charge and discharge strategy, adjusting the ambient temperature of the energy storage battery compartment, and checking whether the aerogel equipment and facilities are normal. At the same time, the execution layer receives the energy storage battery operation strategy and transmits it to the operation and maintenance personnel, who can use it as a reference for actively adjusting the operation of the energy storage battery.

[0111] If the energy storage battery risk assessment result is a high-risk assessment result for the energy storage battery, after calculating the cut-off delay time, a cut-off operation command for the energy storage battery in the corresponding area is generated and transmitted to the execution layer for execution, and the aerogel injection amount is calculated, and an aerogel injection command for the corresponding area of ​​the energy storage battery is generated and transmitted to the execution layer for execution. In this embodiment, the high-risk assessment result means that there are serious safety hazards in the operating state of the energy storage battery, which may cause fires and other accidents at any time. Therefore, it is necessary to immediately activate active safety fire protection for the energy storage battery. In this regard, the measures taken by this application include cutting off the power supply of the relevant energy storage battery, activating aerogel fire-fighting equipment for fire extinguishing and heat insulation, and sending emergency alerts to the fire department and relevant management personnel to dispatch additional firefighters for support.

[0112] Therefore, it is particularly important to determine the delay time and the amount of aerogel sprayed during shutdown. The purpose of the shutdown delay time is to avoid false power outages caused by misjudgment, accelerate power outages in high-risk scenarios, and ensure the reliability of equipment operation. The calculation expression for the shutdown delay time is:

[0113] t cut =0.8-0.005×R2

[0114] R2 is the result of the second risk calculation. The basic delay term is set to 0.8s to ensure sufficient time to verify the authenticity of the risk signal. In the risk adjustment term -0.005×R2, the higher the R2, the shorter the delay time, achieving dynamic adjustment based on the more urgent the risk and the faster the power outage.

[0115] The calculation expression of aerogel injection amount is:

[0116]

[0117] Wherein, R2 is the second risk calculation result, Area represents the risk area area, which is obtained by positioning the risk area in the energy storage battery pack topology G, P TR represents the thermal runaway early warning index; in this application, the calculation of the aerogel injection amount needs to ensure that the amount of aerogel is sufficient to extinguish or suppress thermal runaway, and also needs to avoid excessive waste, therefore, the basic injection amount term 1.2 x R2 is used to match the system level risk severity, the higher R2 is, the larger the basic injection amount is; the dynamic compensation term The risk area is adapted to the thermal runaway emergency, the area size is balanced with the risk emergency, and the amount deviation caused by a single factor is avoided.

[0118] Finally, the execution layer receives the optimal fire fighting strategy and implements corresponding control and aerogel fire extinguishing operation on the energy storage battery according to the optimal fire fighting strategy.

[0119] Embodiment two:

[0120] The difference between embodiment two and embodiment one is that in embodiment two, the energy storage battery operating parameter monitoring sensor includes a main energy storage battery operating parameter monitoring sensor and a secondary energy storage battery operating parameter monitoring sensor, the main energy storage battery operating parameter monitoring sensor is arranged at a key node of the energy storage battery system, and the secondary energy storage battery operating parameter monitoring sensor is arranged at a secondary node position at the key node of the energy storage battery.

[0121] The operating parameters between the main energy storage battery operating parameter monitoring sensor at the key node and the secondary energy storage battery operating parameter monitoring sensor at the secondary node are associated through an association algorithm.

[0122] In this embodiment, the key node refers to a monitoring point in the energy storage battery system that has a decisive influence on the safety, performance and service life of the system; the sensor data of these nodes directly reflects the core state of the system and is crucial for the safe operation and fault diagnosis of the system.

[0123] The secondary node refers to a monitoring point in the energy storage battery system that has relatively small influence on the safety, performance and service life of the system; the sensor data of these nodes is mainly used for auxiliary verification, redundant backup and comprehensive perception of the system state.

[0124] In actual application process, the sensor at the key node can monitor the real-time operating parameters of the energy storage battery, but the sensor at the key node may be damaged, for example, being bitten by an animal, reaching the end of its service life, etc. Therefore, in order to enable the sensor at the key node to continue monitoring the operating parameters after damage, the secondary node is arranged near the key node, and the key node and the secondary node are associated, so that even if the key node is damaged, the operating parameter monitoring result at the key node can be obtained through the parameters monitored by the secondary node and the association algorithm.

[0125] The association algorithm adopted between the key node and the secondary node in the application can adopt different association algorithms according to different monitoring data, for example, a back-stepping algorithm based on voltage association, a temperature back-stepping algorithm based on heat conduction, etc., and the application does not make too many limitations.

[0126] The above is only an embodiment of the application, and common knowledge such as specific structures and characteristics in the scheme is not described in detail here. The person skilled in the art knows all the ordinary technical knowledge in the field of the application before the application date or the priority date, can know all the prior art in the field, and has the ability to apply conventional experimental means before that date. The person skilled in the art can improve and implement the scheme under the guidance of the application combined with their own ability. Some typical known structures or known methods should not be an obstacle for the person skilled in the art to implement the application. It should be noted that, for those skilled in the art, without departing from the structure of the application, a number of modifications and improvements can be made, which should also be considered within the protection scope of the application. These will not affect the implementation effect and practicality of the patent. The protection scope of the application should be subject to the content of its claims, and the specific implementation mode and the like in the specification can be used to explain the content of the claims.

Claims

1. An active safety fire protection system with an energy storage battery and aerogel, characterized by: It includes a perception layer, a transmission layer, a platform layer, and an execution layer. The perception layer is used to collect perception data of the energy storage battery operation process and operating environment. The transmission layer is used to transmit the perception data collected by the perception layer to the platform layer. The platform layer includes a data preprocessing module, a risk identification module, and a fire protection strategy formulation module. The data preprocessing module is used to preprocess the received perception data to obtain preprocessed perception data; The risk identification module deploys a first active risk identification algorithm and a second active risk identification algorithm, and the risk identification module calls the first active risk identification algorithm and the second active risk identification algorithm to perform real-time risk assessment on the pre-processed perception data according to a preset active risk identification strategy to obtain a risk assessment result of the energy storage battery; The fire fighting strategy formulation module is used to formulate the best fire fighting strategy according to the risk assessment results of the energy storage battery; The execution layer is used to receive the optimal fire fighting strategy and implement corresponding control and aerogel fire extinguishing operations on the energy storage battery according to the optimal fire fighting strategy.

2. The active safety fire protection system of the energy storage battery with aerogel according to claim 1 is characterized in that: The perception layer includes a first perception layer and a second perception layer. The first layer of perception data collected by the first perception layer includes monitoring data of environmental monitoring sensors, monitoring data of energy storage battery operation parameter monitoring sensors, and aerogel fire-fighting means data; the second layer of perception data collected by the second perception layer includes aerogel deployment status data, fire risk area data, and energy storage battery combustion prediction data.

3. The active safety fire protection system of the energy storage battery with aerogel according to claim 2, characterized in that: The risk identification module includes a first risk identification unit and a second risk identification unit. The first risk identification unit receives monitoring data from the environmental monitoring sensor, monitoring data from the energy storage battery operating parameter monitoring sensor, and aerogel firefighting means data in the first layer of perception data, calls a first risk active identification algorithm to perform a first risk calculation, and generates a low risk assessment result for the energy storage battery if the first risk does not meet a preset first risk threshold. If the first risk meets a preset first risk threshold but does not meet a preset second risk threshold, generating a risk assessment result for the energy storage battery; The second risk identification unit is used to call a second risk active identification algorithm to perform a second risk judgment based on aerogel deployment status data, fire risk area data, and energy storage battery combustion prediction data when the first risk exceeds a second risk threshold, and generate a high-risk assessment result for the energy storage battery.

4. The active safety fire protection system of the energy storage battery with aerogel according to claim 3 is characterized in that: The construction process of the first active risk identification algorithm is as follows: Based on the monitoring data of the environmental monitoring sensor and the monitoring data of the energy storage battery operating parameter monitoring sensor in the first layer of perception data, a fault scoring function is constructed, which is expressed as follows: Among them, M represents the feature dimension of the monitoring data in the first layer of perception data, ReLU() represents the nonlinear activation function, and λ m represents the feature weight vector, f m (t) represents the real-time characteristic value, μ m represents the historical mean of the feature, σ m represents the standard deviation of the feature, Entropy(C(t)) represents the entropy value of the fault code sequence, C(t) represents the set of fault codes for a period of time, and α represents the entropy value weight; Based on the scoring results of the fault scoring function, the clustering recognition algorithm is called to output the fault category. The expression is: X={S edge (t)|f1(t)、f2(t)……f m (t)} ∈ k =0.08+0.0001×Cycle Among them, x i 、x j Represents the sample point in the feature vector X, dist() represents the sample point x i 、x j The distance between ∈ k represents the dynamic clustering radius, Cycle represents the number of cycles of the energy storage battery, Represents ∈ k The sample set in the neighborhood, MinPts k Indicates the minimum number of neighborhood points; According to the scoring result of the fault scoring function and the cluster identification result, the first risk calculation result is output, and the expression is: Among them, β represents the fault scoring weight, K represents the fault type, and γ k represents the fault severity coefficient, and ClusterProb(k) represents the probability that the sample belongs to the kth type of fault.

5. The active safety fire protection system of the energy storage battery with aerogel according to claim 4 is characterized in that: The construction process of the second risk active identification algorithm is as follows: A thermal runaway warning index model for energy storage batteries is constructed using a mechanism-data fusion model, and the expression is: in, is the mechanism term, A represents the frequency factor, E a represents activation energy, R represents gas constant, T(t) represents real-time core temperature of energy storage battery, LSTM() represents long short-term memory network with 32 hidden layers, H(t) represents time series feature, Sigmoid() represents activation function, and Z(t) represents battery expansion monitoring value; Construct a safety situation awareness model for the energy storage battery system, and the expression is: Where GNN stands for graph neural network, G represents the energy storage battery group topology graph, which is the input of the graph neural network, W is the edge weight, and represents the thermal conductivity coefficient between energy storage battery modules; Couple(i,j) represents the fault coupling degree between the i-th energy storage battery module and the j-th energy storage battery module, δ represents the coupling coefficient, and N represents the number of adjacent energy storage battery modules. The second risk is calculated based on the thermal runaway warning index model and the energy storage battery system safety situation awareness model. The expression is: R2=(η×P TR +(1-n)×S cloud )×Exp(t persist )+κ×Damage(t) Where η represents the probability weight of thermal runaway, Exp(t persist ) represents the continuous risk amplification term, t persist represents the risk duration in minutes, Damage(t) represents the estimated battery damage, and κ represents the damage weight.

6. The active safety fire protection system of the energy storage battery with aerogel according to claim 3 is characterized in that: The fire fighting strategy formulation module formulates the optimal fire fighting strategy according to the risk assessment results of the energy storage battery as follows: If the energy storage battery risk assessment result is a low risk assessment result for the energy storage battery, an operation and maintenance reminder signal is generated regularly; If the energy storage battery risk assessment result is a medium risk assessment result for the energy storage battery, an energy storage battery operation adjustment command is generated, and an energy storage battery operation strategy is generated; If the energy storage battery risk assessment result is a high risk assessment result for the energy storage battery, after calculating the cut-off delay time, a cut-off operation command for the energy storage battery in the corresponding area is generated and transmitted to the execution layer for execution, and the aerogel injection amount is calculated, and an aerogel injection command for the energy storage battery corresponding area is generated and transmitted to the execution layer for execution.

7. The active safety fire protection system of the energy storage battery with aerogel according to claim 6, characterized in that: The calculation expression of the cut-off delay time is: t cut =0.8-0.005×R2 Among them, R2 is the second risk calculation result; The calculation expression of the aerogel injection amount is: Among them, R2 is the second risk calculation result, Area represents the area of ​​the risk area, which is obtained by locating the risk area in the energy storage battery group topology map G, and P TR Indicates the thermal runaway warning index.

8. The active safety fire protection system of the energy storage battery with aerogel according to claim 2, characterized in that: The energy storage battery operating parameter monitoring sensor includes a main energy storage battery operating parameter monitoring sensor and a secondary energy storage battery operating parameter monitoring sensor. The main energy storage battery operating parameter monitoring sensor is arranged at a key node of the energy storage battery system, and the secondary energy storage battery operating parameter monitoring sensor is arranged at a secondary node position at the key node of the energy storage battery. The operating parameters of the main energy storage battery operating parameter monitoring sensor at the key node and the auxiliary energy storage battery operating parameter monitoring sensor at the secondary node are associated through an association algorithm.

Images