Thermal management cascading safety control method for energy storage circuit

Abstract

The application discloses a thermal management hierarchical linkage safety control method of an energy storage loop, relates to the technical field of safety control of an electric energy storage system, and comprises the following steps: an energy storage loop control system continuously and synchronously monitors a multi-dimensional operation state parameter set of an energy storage battery plug-in box in a battery module in the energy storage loop, an operation characteristic vector of the energy storage battery plug-in box is constructed, correlation analysis processing is performed on the operation characteristic vector, abnormal state judgment information is obtained, the operation characteristic vector is subjected to abnormal state evolution analysis according to the abnormal state judgment information, a thermal runaway risk identifier is obtained, and hierarchical coordination control is executed on the battery module, after the hierarchical coordination control is executed, the operation characteristic vector is continuously monitored, a runaway signal of the battery module is generated, and then, electrical isolation processing is executed on the energy storage loop corresponding to the battery module, and after the electrical isolation processing is executed, directional immersion processing is executed according to an active fire-fighting logic, so that the thermal runaway diffusion probability is effectively reduced, and the overall operation safety of the energy storage system is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of safety control of electric energy storage systems, in particular to a thermal management hierarchical linkage safety control method of an energy storage loop. BACKGROUND

[0002] In an energy storage loop system, energy storage battery modules are arranged in a modular form, and the entire system needs to continuously perform high-power charging and discharging tasks to meet the application requirements of power grid regulation, peak load shifting, or micro-grid power supply. During operation, the system not only requires sustained and stable energy output capability, but also must ensure that each battery module maintains a safe and controllable working state under high-power load to avoid whole group or even system failure caused by single or local module abnormalities. Therefore, under high-power operating conditions, how to inhibit heat accumulation and reduce the risk of thermal runaway through a targeted and dynamically adjustable thermal management and control strategy is one of the core technical tasks in the design and operation of the energy storage loop system.

[0003] Under high load environmental conditions, rapid heat accumulation inside the battery module can cause a significant nonlinear amplification effect of the chemical reaction rate, and further cause abnormal phenomena such as gas expansion, heat spread, and electrolyte decomposition. Existing thermal management methods mainly rely on module-level centralized liquid cooling or local heat exchanger circulation, and trigger global or coarse-grained cooling operations through preset temperature thresholds. However, these methods have obvious control limitations: single threshold triggering often lags behind and cannot cope with rapidly evolving modules with heat accumulation; when cooling capacity is limited, it cannot ensure that high-risk modules are cooled first; more importantly, centralized liquid cooling circuits are prone to uneven heat dissipation between plug-in boxes, which can cause significant temperature differences and weaken the overall consistency and life stability of the battery module; local hot spots may develop into irreversible chemical reactions before the control strategy takes effect, resulting in a significant decline in safety control effectiveness. If only traditional and coarse temperature control cooling methods are used for safety control, once the abnormal heat exceeds the designed cooling capacity, the system may face risks such as continuous temperature rise, heat spread to adjacent modules, and internal electrical short circuit, which can further cause cell performance degradation, internal resistance rise, and internal short circuit problems. SUMMARY

[0004] To overcome the deficiencies of the prior art, the present application provides a thermal management hierarchical linkage safety control method of an energy storage loop, which can effectively solve the problems involved in the background art.

[0005] To achieve the above purpose, the present application realizes the following technical scheme: a thermal management hierarchical linkage safety control method of an energy storage loop, comprising S1: an energy storage loop control system continuously and synchronously monitors a set of multi-dimensional operating state parameters of an energy storage battery plug-in box in a battery module in the energy storage loop, and constructs an operating feature vector of the energy storage battery plug-in box.

[0006] S2: Perform correlation analysis on the operating feature vector of the energy storage battery pack to obtain abnormal state judgment information of the energy storage battery pack.

[0007] S3: The energy storage loop control system analyzes the abnormal state evolution of the energy storage battery pack based on the abnormal state judgment information of the energy storage battery pack, and obtains the thermal runaway risk identifier of the energy storage battery pack.

[0008] S4: The energy storage loop control system performs hierarchical coordinated control of the battery modules in the energy storage loop based on the thermal runaway risk identification of the energy storage battery pack.

[0009] S5: After performing hierarchical coordinated control, continuously monitor the operating characteristic vector of the energy storage battery box, generate the runaway signal of the battery module, and the energy storage circuit control system performs electrical isolation processing on the energy storage circuit corresponding to the battery module based on the runaway signal of the battery module. After performing electrical isolation processing, perform directional immersion processing according to the active fire protection logic.

[0010] Furthermore, the method for the energy storage loop control system to continuously and synchronously monitor the multi-dimensional operating status parameter set of the energy storage battery box in the battery module of the energy storage loop is as follows: collecting the multi-dimensional operating status parameter set of the energy storage battery box in the battery module of the energy storage loop. The multi-dimensional operating status parameter set of the energy storage battery box in the battery module includes electrical characteristic data, thermal characteristic data, gas characteristic data and operating characteristic data of the energy storage battery box.

[0011] Furthermore, the method for constructing the operating feature vector of the energy storage battery box is as follows: analyze the thermal characteristic data of the energy storage battery box in the same battery module to obtain the maximum temperature difference of the energy storage battery box, form an analysis window with a preset number of continuous operating cycles, record the maximum temperature difference of the energy storage battery box in the analysis window, and form a temperature difference numerical sequence of the energy storage battery box.

[0012] The slope of the linear regression of the temperature difference numerical sequence of the energy storage battery compartment is taken as the rate of change of the temperature difference of the energy storage battery compartment.

[0013] The standard deviation of the temperature difference sequence of the energy storage battery compartment is used as the temperature difference fluctuation range of the energy storage battery compartment.

[0014] Based on the temperature difference numerical sequence, temperature difference change rate, and temperature difference fluctuation amplitude of the energy storage battery box, temperature difference evolution characteristic data of the energy storage battery box are constructed.

[0015] The electrical, thermal, gaseous, temperature difference evolution, and operational characteristic data of the energy storage battery pack are structurally combined to obtain the operational characteristic vector of the energy storage battery pack.

[0016] Furthermore, the method for performing correlation analysis on the operating feature vector of the energy storage battery pack is as follows: input the operating feature vector of the energy storage battery pack into a pre-trained abnormal state judgment model, and output the abnormal risk probability value, abnormal feature identifier, and feature data subset corresponding to the abnormal feature identifier of the energy storage battery pack.

[0017] The abnormal feature identifiers include electrical features, thermal features, gas features, temperature difference evolution features, and operational features; the feature data subsets corresponding to the abnormal feature identifiers include electrical feature data corresponding to electrical features, thermal feature data corresponding to thermal features, gas feature data corresponding to gas features, temperature difference evolution feature data corresponding to temperature difference evolution features, and operational feature data corresponding to operational features.

[0018] Furthermore, the method for obtaining the abnormal state judgment information of the energy storage battery box is as follows: extract the abnormal risk probability value of the energy storage battery box, and map the abnormal risk probability value to the corresponding abnormal state judgment interval based on the predefined abnormal state judgment interval division rule table to obtain the abnormal level of the energy storage battery box.

[0019] The abnormal state judgment information of the energy storage battery box is obtained by encapsulating the abnormal risk probability value, abnormal level, abnormal feature identifier, and the feature data subset corresponding to the abnormal feature identifier.

[0020] Furthermore, the method for performing abnormal state evolution analysis on the operating characteristic vector of the energy storage battery pack is as follows:

[0021] The system receives abnormal state judgment information from the energy storage battery pack and serializes it in chronological order to form a time evolution sequence of the abnormal state of the energy storage battery pack.

[0022] Based on the time evolution sequence of abnormal states of the energy storage battery pack, the thermal runaway risk of the energy storage battery pack is classified and determined, and the thermal runaway risk label of the energy storage battery pack is obtained.

[0023] The thermal runaway risk indicators for the energy storage battery pack include warning level, intervention level, and emergency level.

[0024] Furthermore, the method for performing hierarchical coordinated control on the battery modules in the energy storage circuit is as follows:

[0025] Step a1: Analyze the feature data subset corresponding to the abnormal feature identifier of the energy storage battery box to obtain the comprehensive deviation index of the energy storage battery box. The comprehensive deviation index of the energy storage battery box is used to quantify the degree to which the current operating state of the energy storage battery box deviates from the historical stable operating benchmark.

[0026] Step a2: Extract the thermal runaway risk identifier of the energy storage battery pack. When the thermal runaway risk identifier of the energy storage battery pack is at the warning level, perform warning-level coordinated control processing on the battery module to which the energy storage battery pack belongs based on the abnormal characteristic identifier and comprehensive deviation index of the energy storage battery pack. This includes sending the pack risk priority sorting instruction, power dynamic constraint instruction and warning-level cooling control instruction to the energy storage loop control system.

[0027] Step a3: When the thermal runaway risk of the energy storage battery pack is identified as intervention level, intervention-level coordinated control processing is performed on the battery module to which the energy storage battery pack belongs, based on the abnormal characteristic identification and comprehensive deviation index of the energy storage battery pack. This includes sending intervention-level power dynamic constraint commands, intervention-level topology isolation protection commands, and intervention-level cooling enhancement control commands to the energy storage loop control system.

[0028] Step a4: When the thermal runaway risk of the energy storage battery box is identified as an emergency level, emergency-level coordinated control processing is performed on the battery module to which the energy storage battery box belongs, including: sending emergency-level power dynamic constraint commands, emergency-level electrical isolation control commands, and emergency-level safety protection switching commands to the energy storage circuit control system.

[0029] Furthermore, the method for generating the runaway signal of the battery module is as follows: continuously construct the operating feature vector of the energy storage battery box, and analyze it to obtain the thermal runaway risk indicator of the energy storage battery box in the battery module; when the thermal runaway risk indicator of the energy storage battery box is continuously marked as an emergency level, it is determined that the thermal runaway risk indicator of the energy storage battery box meets the condition of continuous establishment of the emergency level.

[0030] The operating characteristic vector of the energy storage battery pack in the corresponding battery module is acquired simultaneously to determine whether the thermal runaway evolution trend meets the irreversible condition.

[0031] A battery module runaway signal is generated when the thermal runaway risk indicator meets the conditions for the emergency level to remain in effect and the thermal runaway evolution trend meets the conditions for irreversibility; otherwise, no battery module runaway signal is generated.

[0032] Furthermore, the method for performing electrical isolation processing on the energy storage circuit corresponding to the battery module is as follows: after generating the runaway signal of the battery module, an emergency isolation control command is sent to switch the power output control state of the corresponding battery module to the zero power state, and a disconnection control command is sent to the main DC contactor controlling the corresponding battery module.

[0033] Furthermore, the method for performing directional immersion processing according to active fire protection logic after performing electrical isolation processing is as follows: extract the operating feature vectors of all energy storage battery boxes in the battery module, arrange them in order of physical location of the energy storage battery boxes, and form the module state vector of the battery module.

[0034] Based on the module state vector of the corresponding battery module in the analysis window, joint judgment processing is performed on the smoke characteristic state and temperature characteristic state, and then directional immersion processing is performed.

[0035] The present invention has the following beneficial effects:

[0036] This invention constructs a multi-dimensional operational feature vector covering electrical, thermal, gas, temperature difference evolution, and operational status, and combines it with an abnormal state judgment model and abnormal state time evolution analysis to achieve early identification and trend determination of thermal runaway risks in energy storage battery packs. Compared with the static criteria of relying on a single temperature, voltage, or gas threshold in the prior art, this invention can cross-validate anomalies from multiple physical dimensions, and further analyze the risk evolution momentum through the trend analysis of anomaly risk probability changes, thereby improving the ability to identify early hidden anomalies and rapidly deteriorating risks, effectively reducing the probability of runaway caused by delayed early warning, and enhancing the safety redundancy and predictive protection capabilities of energy storage systems during operation.

[0037] This invention normalizes and fuses multiple feature data sub-items corresponding to anomaly identifiers, and drives a hierarchical coordinated control strategy accordingly. This enables refined and adaptive linkage of power regulation, topology isolation, and cooling control at the energy storage loop level. Compared with the fixed intensity or single action control methods in the prior art, this invention can dynamically match the power constraint range, cooling intensity, and isolation strategy according to different stages of anomaly severity, avoiding the impact of overprotection on system performance. At the same time, it can quickly increase the control intensity when the risk escalates, thereby achieving a better balance between safety and system availability, and improving the overall stability and reliability of the energy storage loop operation.

[0038] This invention constructs a reliable thermal runaway determination and response triggering mechanism by combining continuous emergency state determination with multi-dimensional irreversible physical evolution conditions during the runaway signal generation and subsequent safety handling stages. Based on this, an active fire suppression strategy combining electrical isolation and directional immersion is implemented, effectively avoiding false triggering caused by transient anomalies or sensor disturbances. This ensures that the highest level of safety handling is only initiated when there is a genuine risk of irreversible thermal runaway, thereby reducing unnecessary system downtime and fire suppression medium consumption. At the same time, through directional and graded fire suppression response methods, the precise suppression capability of local risks is enhanced, reducing the possibility of thermal runaway spreading to adjacent battery modules and energy storage circuits, and comprehensively improving the safety level of the energy storage system. Attached Figure Description

[0039] Figure 1 This is a schematic diagram of the method flow of the present invention.

[0040] Figure 2This is a schematic diagram of the method for performing hierarchical coordinated control of battery modules in an energy storage circuit according to the present invention. Detailed Implementation

[0041] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0042] Please see Figure 1 As shown, this embodiment of the invention provides a technical solution: a hierarchical linkage safety control method for thermal management of energy storage circuits, including S1: the energy storage circuit control system continuously and synchronously monitors the multi-dimensional operating status parameter set of the energy storage battery box in the battery module of the energy storage circuit, and constructs the operating feature vector of the energy storage battery box.

[0043] The method for the energy storage loop control system to continuously and synchronously monitor the multi-dimensional operating status parameter set of the energy storage battery pack within the energy storage loop is as follows:

[0044] The energy storage loop control system collects multi-dimensional operating status parameter sets of the energy storage battery boxes in the battery module in the energy storage loop in parallel. The multi-dimensional operating status parameter set of each energy storage battery box in the battery module includes electrical characteristic data, thermal characteristic data, gas characteristic data and operating characteristic data of the energy storage battery box.

[0045] It should be added that the energy storage loop control system is the core control unit that monitors, analyzes and controls the operating status of the energy storage loop. It communicates with the battery management system, power conversion device and loop protection device to receive multi-dimensional operating status parameters from the energy storage battery box and battery module, and performs status assessment, risk judgment and control command generation based on the acquired operating data, thereby realizing unified scheduling and safe control of the power output status, power change rate and loop connection status of the energy storage loop power supply branch.

[0046] It should be noted that the electrical characteristic data of the energy storage battery pack includes the battery terminal voltage, voltage fluctuation rate, internal resistance variation difference, power output value, and rated power value; the thermal characteristic data of the energy storage battery pack includes the liquid cooling flow rate, liquid cooling circuit inlet temperature, liquid cooling circuit outlet temperature, and surface temperature; the gas characteristic data of the energy storage battery pack includes the hydrogen concentration, carbon monoxide concentration, and ethylene concentration; the operational characteristic data of the energy storage battery pack includes the heat dissipation power and energy storage battery load; and through sensors deployed on the energy storage battery pack, the energy storage circuit control system synchronously collects and updates a multi-dimensional set of operational status parameters according to a preset operating cycle.

[0047] In this embodiment, by synchronously monitoring multi-dimensional operating status parameters, the one-sidedness and lag of traditional single temperature or voltage threshold monitoring are reduced, providing a complete data foundation for early identification of composite anomalies in the battery pack. The principle is to cross-verify the battery status from multiple physical dimensions such as energy input, heat generation and transfer, by-product release and system load, rather than simply listing data.

[0048] The method for constructing the operational feature vector of the energy storage battery pack is as follows:

[0049] After obtaining the multi-dimensional operating status parameter set of the energy storage battery pack, the energy storage loop control system analyzes the thermal characteristic data of the energy storage battery pack in the same battery module. In each operating cycle, the maximum and minimum surface temperatures of all energy storage battery packs in the same battery module are counted, and the absolute difference between the maximum and minimum surface temperatures is calculated as the maximum temperature difference of the energy storage battery pack.

[0050] In each operating cycle, the average surface temperature of all energy storage battery boxes in the same battery module is calculated. In the physical layout of the battery module, the combination formed by two directly adjacent energy storage battery boxes is defined as an energy storage battery box pair. The absolute difference between the average surface temperatures of the energy storage battery box pairs is recorded as the absolute surface temperature difference of the energy storage battery box pair.

[0051] An analysis window is formed by a preset number of continuous operating cycles. The analysis window ends at the current operating cycle and traces back a preset number of historical operating cycles. Together with the current operating cycle, they form a continuous operating cycle interval. The maximum temperature difference of the energy storage battery pack within the analysis window is recorded, forming a numerical sequence of temperature differences of the energy storage battery pack.

[0052] The slope of the linear regression of the temperature difference numerical sequence of the energy storage battery compartment is taken as the rate of change of the temperature difference of the energy storage battery compartment.

[0053] The standard deviation of the temperature difference sequence of the energy storage battery compartment is used as the temperature difference fluctuation range of the energy storage battery compartment.

[0054] Based on the temperature difference numerical sequence, temperature difference change rate, and temperature difference fluctuation amplitude of the energy storage battery box, temperature difference evolution characteristic data of the energy storage battery box is constructed. The temperature difference evolution characteristic is used to characterize the trend and stability characteristics of the maximum temperature difference of the energy storage battery box changing with the operating cycle.

[0055] The electrical, thermal, gaseous, temperature difference evolution, and operational characteristic data of the energy storage battery pack are structurally combined to obtain the operational characteristic vector of the energy storage battery pack.

[0056] The method for structured combination processing of electrical characteristic data, thermal characteristic data, gas characteristic data, temperature difference evolution characteristic data, and operational characteristic data of energy storage battery packs is as follows:

[0057] The values ​​of each sub-item in the electrical characteristic data, thermal characteristic data, gas characteristic data, temperature difference evolution characteristic data, and operational characteristic data are sequentially spliced ​​together to form a numerical array that can comprehensively reflect the thermal state and operational state of the energy storage battery pack. This numerical array is the operational characteristic vector of the energy storage battery pack.

[0058] S2: Perform correlation analysis on the operating feature vector of the energy storage battery pack to obtain abnormal state judgment information of the energy storage battery pack.

[0059] The method for performing correlation analysis on the operating feature vector of the energy storage battery pack is as follows:

[0060] The operational feature vector of the energy storage battery pack is input into a pre-trained abnormal state judgment model, and the output is the abnormal risk probability value of the energy storage battery pack, the abnormal feature identifier, and the feature data subset corresponding to the abnormal feature identifier.

[0061] In this embodiment, the model is used to analyze the complex nonlinear correlations and temporal dependencies in high-dimensional feature vectors that are difficult to express by manual rules, thereby achieving early and quantitative assessment of thermal runaway risk. It not only outputs the risk probability, but also traces the main risk characteristics through attention mechanisms and other technologies, making the early warning results interpretable and enabling targeted control. This reduces the limitation of traditional threshold methods that only issue alarms but do not diagnose.

[0062] The electrical characteristic data of the energy storage battery pack directly reflects its voltage, current, and power state, and is an important indicator of the battery's energy supply and instantaneous load capacity. For example, abnormal fluctuations in current and voltage can increase the probability of anomalies and may trigger anomaly indicators in the electrical characteristics. Thermal characteristic data characterizes the temperature distribution and heat accumulation of the energy storage battery pack. By affecting the chemical reaction rate and changes in the battery's internal impedance, it directly influences the degree of deviation from the risk of thermal runaway. For example, sustained high temperatures or rapid temperature rises increase the probability of anomalies and trigger anomaly indicators in the thermal characteristics. Gas characteristic data describes the energy storage... Changes in the concentration of combustible gases inside or outside the battery compartment reflect byproducts released in the early stages of thermal runaway. An abnormally high gas concentration directly increases the probability of anomalies and identifies abnormal gas characteristics. Temperature gradient evolution data captures temperature gradient changes inside the battery compartment or between adjacent cells. By reflecting localized overheating or uneven heating trends, it plays a crucial role in determining the probability of anomalies and identifying abnormal temperature gradient evolution characteristics. Operational characteristic data characterizes the overall operating status of the energy storage battery compartment within the module, such as load changes, SOC fluctuations, and cycling modes. This comprehensively reflects the impact of abnormal battery system operation on the evolution of thermal runaway. For example, abnormal operational characteristics can lead to an increased probability of anomalies and trigger anomaly identification.

[0063] The operational feature vector constructed by combining the above various feature data, after being processed by the abnormal state judgment model, outputs an abnormal risk probability value to quantify the degree to which the current operating state of the energy storage battery pack has deviated from the thermal runaway evolution state. The output abnormal feature identifier is used to indicate the main risk source category. The abnormal feature identifier includes electrical features, thermal features, gas features, temperature difference evolution features, and operational features. At the same time, the specific feature data extracted from the current operational feature vector on which the abnormal feature identifier is triggered by the abnormal state judgment model outputs constitutes the feature data subset corresponding to the abnormal feature identifier, thereby realizing a comprehensive assessment of the abnormal state of the energy storage battery pack. The feature data subset corresponding to the output abnormal feature identifier includes electrical feature data corresponding to electrical features, thermal feature data corresponding to thermal features, gas feature data corresponding to gas features, temperature difference evolution feature data corresponding to temperature difference evolution features, and operational feature data corresponding to operational features.

[0064] The training method for the abnormal state judgment model is as follows:

[0065] A bidirectional LSTM network based on an attention mechanism is selected as the model architecture for processing operational feature vectors. An abnormal state judgment model for judging abnormal states of energy storage battery packs is constructed. The abnormal state judgment model includes a feature encoding layer, a feature fusion layer, and a risk output layer connected in sequence. The feature encoding layer includes multiple parallel feature encoding sub-networks, which are used to encode electrical feature data, thermal feature data, gas feature data, temperature difference evolution feature data, and operational feature data in the operational feature vector of the energy storage battery pack, respectively, to obtain the corresponding high-dimensional feature vectors. The feature fusion layer is used to perform weighted fusion of the high-dimensional feature vectors to generate a fused feature vector. The risk output layer includes an abnormal risk probability output branch and an abnormal feature identification output branch. The abnormal risk probability output branch outputs the abnormal risk probability value of the energy storage battery pack, and the abnormal feature identification output branch outputs the abnormal feature identification of the energy storage battery pack. The abnormal feature identification output branch synchronously outputs the feature data subset index information corresponding one-to-one with the abnormal feature identification to realize the accurate mapping relationship between the abnormal feature identification and the specific feature data. The abnormal risk probability value is a value between 0 and 1.

[0066] A training sample set for an abnormal state judgment model is constructed. Multi-dimensional operating state parameter sets for the stable operation phase of the energy storage battery pack are extracted from historical operating data to generate normal state samples, and corresponding low-abnormal-risk probability values ​​are labeled for these samples. Multi-dimensional operating state parameter sets for the thermal runaway evolution phase are extracted from historical records of thermal runaway events to generate abnormal state samples, and corresponding high-abnormal-risk probability values ​​and abnormal feature identifiers are labeled for these samples. Simultaneously, the operating feature vector sub-items that play a dominant role in judging abnormal feature identifiers in the abnormal state samples are labeled to train the output capability of the feature data subset corresponding to the abnormal feature identifiers, thus obtaining the training sample set.

[0067] It should be added that the anomaly risk probability value can be determined using rule-based or statistical methods, specifically including:

[0068] An anomaly detection index system is constructed based on a multi-dimensional set of operating state parameters during the operation of the energy storage loop. This multi-dimensional set of operating state parameters is converted into an operating feature vector, and a corresponding safety threshold range is set for each feature data in the operating feature vector. The safety threshold range includes an upper threshold and a lower threshold, which are determined based on parameter distribution ranges obtained from historical normal operation data. The real-time observed values ​​of each feature data are compared with the corresponding safety threshold ranges. When the real-time observed value is greater than the upper threshold, the difference between the real-time observed value and the upper threshold is calculated as a positive deviation; when the real-time observed value is less than the lower threshold, the difference between the lower threshold and the real-time observed value is calculated as a negative deviation; when the real-time observed value is less than the lower threshold, the difference between the lower threshold and the real-time observed value is calculated as a negative deviation. When the observed value is within the safe threshold range, the deviation is set to zero. The deviation of the real-time observed value of each feature data from the safe threshold range is statistically analyzed to obtain the anomaly index of each feature data. The anomaly index of each feature data is then normalized. The normalization process adopts an interval mapping method based on the statistical range of the historical deviation of each feature data, so that the anomaly index of each feature data is mapped to a uniform scale range. A comprehensive anomaly index is obtained based on a weighted fusion method. The weight of each feature data anomaly index is preset based on the degree of influence of the corresponding feature data on the thermal state of the energy storage battery pack, and remains consistent within the analysis window. The comprehensive anomaly index is normalized to the range of 0 to 1 and recorded as the anomaly risk probability value.

[0069] In the above implementation, the anomaly risk probability value essentially reflects the degree of deviation of the current operating state from the normal state; the larger the value, the higher the anomaly risk. Based on this, a bidirectional LSTM model based on an attention mechanism is introduced, serving only as an optimization of the aforementioned rules or statistical methods. By learning the complex nonlinear relationships in historical operating data, it provides a more accurate estimate of the anomaly risk probability value.

[0070] Supervised training is performed on the abnormal state judgment model based on the training sample set. The difference between the abnormal risk probability value, abnormal feature identifier, and the feature data subset corresponding to the abnormal feature identifier output by the abnormal state judgment model and the corresponding label in the sample set is used as the training objective. The model parameters are iteratively updated through backpropagation until the mean squared error of the abnormal risk probability value prediction output by the model is lower than the preset mean squared error threshold of the abnormal risk probability value prediction, and the classification accuracy of the abnormal feature identifier reaches more than 95%. The model training is then judged to have met the convergence condition, and this model is recorded as the trained abnormal state judgment model. The setting of the mean squared error threshold of the abnormal risk probability value prediction ensures that the model prediction has high numerical accuracy. The reason for the value is that too low an error threshold may lead to training difficulty or overfitting, while too high an error threshold cannot meet the accuracy requirements required for risk quantification.

[0071] The method for obtaining abnormal state judgment information of the energy storage battery pack is as follows:

[0072] The abnormal state judgment model extracts the abnormal risk probability value, abnormal feature identifier, and the feature data subset corresponding to the abnormal feature identifier of the energy storage battery box in the current operating cycle.

[0073] Based on a predefined rule table for dividing abnormal state judgment intervals, the probability value of abnormal risk is mapped to the corresponding abnormal state judgment interval. The abnormal state judgment interval is divided into multiple continuous and non-overlapping interval ranges in order of risk level from low to high. Each abnormal state judgment interval uniquely corresponds to an abnormal level. The abnormal levels of the energy storage battery box include normal, low abnormal, medium abnormal and high abnormal.

[0074] An example of the rule table for dividing the abnormal state determination interval is shown in Table 1.

[0075] Table 1. Rules for Dividing Abnormal State Judgment Intervals

[0076] Abnormal state determination interval Abnormality level [0，0.2) Normal [0.2，0.5) Low abnormality [0.5，0.8) Medium abnormality [0.8，1] High abnormality

[0077] In one specific embodiment, the abnormal risk probability value is mapped to the abnormal state judgment interval. When the abnormal risk probability value is within the range of a certain abnormal state judgment interval, the abnormal state judgment information of the energy storage battery box is determined to correspond to the abnormal level corresponding to the abnormal state judgment interval, which is recorded as the abnormal level of the energy storage battery box.

[0078] The abnormal risk probability value, abnormal level, abnormal feature identifier, and corresponding feature data subset of the energy storage battery pack are encapsulated according to a unified data field format to obtain the abnormal state judgment information of the energy storage battery pack. The abnormal state judgment information is structured data, which serves as the direct input basis for subsequent thermal runaway risk identifier generation and hierarchical coordinated control processing, thereby realizing the data connection between the abnormal identification result and the hierarchical linkage safety control process of thermal management.

[0079] In this embodiment, encapsulating data into structured data ensures the standardization and reliability of information transmission, facilitates data exchange and parsing between different modules, and is a necessary technical step in building an automated safety control system.

[0080] S3: The energy storage loop control system analyzes the abnormal state evolution of the energy storage battery pack based on the abnormal state judgment information of the energy storage battery pack, and obtains the thermal runaway risk identifier of the energy storage battery pack. The thermal runaway risk identifier of the energy storage battery pack includes the warning level, intervention level and emergency level.

[0081] The method for the energy storage loop control system to perform abnormal state evolution analysis on the operating characteristic vector of the energy storage battery pack based on the abnormal state judgment information of the energy storage battery pack is as follows:

[0082] The system receives abnormal state judgment information of the energy storage battery pack, which includes an abnormal risk probability value, an abnormal level, an abnormal feature identifier, and a subset of feature data corresponding to the abnormal feature identifier. The abnormal state judgment information in the analysis window is then serialized and arranged in chronological order to form a time evolution sequence of the abnormal state of the energy storage battery pack.

[0083] Based on the time evolution sequence of abnormal states of energy storage battery packs, the thermal runaway risk of energy storage battery packs is classified and determined to obtain thermal runaway risk labels. Specifically, when the abnormality level of the energy storage battery pack is normal, the absolute difference of the surface temperature of the current energy storage battery pack pair is compared with a preset absolute difference threshold. If the absolute difference of the surface temperature of the current energy storage battery pack pair is higher than the preset absolute difference threshold, the thermal runaway risk label of the energy storage battery pack pair with the higher average surface temperature is marked as a warning level; otherwise, no thermal runaway risk label is generated. When the abnormality level of the energy storage battery pack is low, the thermal runaway risk label of the energy storage battery pack is marked as a warning level. When the abnormality level is medium, the thermal runaway risk label of the energy storage battery pack is marked as an intervention level.

[0084] When the anomaly level is high and the rate of increase of the anomaly risk probability value within the analysis window exceeds the preset threshold for the rate of increase of the anomaly risk probability value, the thermal runaway risk of the energy storage battery pack is marked as an emergency level. When the anomaly level is high but the rate of increase of the anomaly risk probability value within the analysis window does not exceed the preset threshold for the rate of increase of the anomaly risk probability value, the thermal runaway risk of the energy storage battery pack is marked as an intervention level. The thermal runaway risk of the energy storage battery pack serves as the direct decision-making basis for setting the power regulation priority and generating the thermal management execution sequence in the subsequent thermal management hierarchical linkage safety control. The preset threshold for the rate of increase of the anomaly risk probability value is determined by those skilled in the art based on historical data statistical analysis. By analyzing multiple historical cases of emergency intervention or thermal runaway of the energy storage battery pack, the growth rate of the anomaly risk probability value within the most recent three analysis windows in these cases is statistically analyzed, and the average value is taken as the threshold for the rate of increase of the anomaly risk probability value.

[0085] In this embodiment, abnormal state evolution analysis of the operating characteristic vector of the energy storage battery pack helps to transform static, single-point anomaly judgment into dynamic, trend-based risk assessment. Traditional methods only make judgments based on the current threshold exceeding the standard, which has serious lag. This invention constructs an abnormal state time evolution sequence and analyzes key indicators, such as the rate of change of the abnormal risk probability value, to identify high-risk states that are deteriorating rapidly in advance. This provides the system with a valuable early warning and intervention window before the thermal runaway reaches the point of physical irreversibility. This process is not a simple time series observation, but rather quantifies the evolution momentum of risk by analyzing the growth rate of the risk probability in the sequence. The judgment criteria are clear and quantifiable, achieving a leap in early warning capability from anomalies to potential rapid deterioration.

[0086] S4: The energy storage loop control system performs hierarchical coordinated control of the battery modules in the energy storage loop based on the thermal runaway risk identification of the energy storage battery pack.

[0087] like Figure 2 As shown, the method for implementing hierarchical coordinated control of battery modules in the energy storage circuit is as follows:

[0088] Step a1: Analyze the subset of feature data corresponding to the abnormal feature identifiers of the energy storage battery box to obtain the comprehensive deviation index of the energy storage battery box.

[0089] Feature data items are extracted from the feature data subset corresponding to the abnormal feature identifiers of the energy storage battery pack. For each extracted feature data item, a reference benchmark value and a deviation judgment threshold are obtained. The reference benchmark value is taken from the statistical feature value of the energy storage battery pack during the historical stable operation phase. The deviation judgment threshold is preset based on the data dispersion characteristics of the historical stable operation phase. The absolute difference between the actual value of the feature data item and the corresponding reference benchmark value is recorded as the deviation of the feature data item. The ratio of the deviation of the feature data item to the corresponding deviation judgment threshold is recorded as the deviation index of the feature data item. The deviation index of the feature data item is used to characterize the deviation intensity of the current feature data item relative to the stable operation state. The reference benchmark value is taken from the statistical average value of each feature data item in the multi-dimensional operation state parameter set collected during the historical long-term stable operation phase of the energy storage battery pack. The deviation judgment threshold is preset by those skilled in the art based on the data dispersion characteristics of the historical stable operation phase. Specifically, it is the calculated standard deviation of each feature data item, and three times the standard deviation is set as the deviation judgment threshold of the feature data item.

[0090] Feature data sub-items refer to single parameter fields that constitute the operating feature vector of the energy storage battery pack. They are used to represent the numerical information of a specific operating state indicator. Each feature data sub-item corresponds to one dimension of the operating feature vector. For example, parameters such as battery terminal voltage, internal resistance change difference, liquid cooling flow rate, surface temperature, hydrogen concentration, maximum temperature difference, temperature difference change rate, and energy storage battery load are all feature data sub-items.

[0091] The deviation indices of all feature data items extracted from the feature data subset corresponding to the abnormal feature identifier are averaged to obtain the comprehensive deviation index of the energy storage battery box, which is used to quantify the degree to which the operating status of the energy storage battery box, as represented by the current abnormal feature data subset, deviates from its historical stable operating benchmark.

[0092] In this embodiment, the calculation of the comprehensive deviation index helps to normalize and integrate multi-dimensional and multidimensional abnormal feature data into a unified and comparable quantitative scale. This enables the system to achieve an overall assessment of the severity of the anomaly. The index directly reflects the distance of the current state from the healthy baseline, providing an accurate and adaptive decision-making basis for the measures taken in subsequent graded control. It reduces the potential deficiencies or overreactions of fixed intensity response in traditional methods. The principle is to define the normal fluctuation range through statistical methods and normalize and integrate abnormal deviations to quantify the comprehensive intensity of the anomaly.

[0093] Step a2: Extract the thermal runaway risk identifier of the energy storage battery pack. When the thermal runaway risk identifier of the energy storage battery pack is at the warning level, perform warning-level coordinated control processing on the battery module to which the energy storage battery pack belongs based on the abnormal characteristic identifier and comprehensive deviation index of the energy storage battery pack. This includes sending the pack risk priority sorting instruction, power dynamic constraint instruction and warning-level cooling control instruction to the energy storage loop control system.

[0094] The battery pack risk priority sorting instruction sorts the battery packs according to the timestamps of the thermal runaway risk indicators, thus obtaining the priority sorting of the battery packs.

[0095] The warning-level power dynamic constraint command dynamically limits the maximum allowable output power and power change rate of the battery module to which the energy storage battery box belongs. Based on the abnormal characteristics and comprehensive deviation index of the energy storage battery box, the warning ratio range of the rated power and the warning ratio range of the original power change rate are analyzed and obtained. The maximum allowable output power of the battery module to which it belongs is dynamically limited to the warning ratio range of the rated power, and the power change rate is limited to the warning ratio range of the original power change rate.

[0096] The method for analyzing and obtaining the warning ratio range of rated power and the warning ratio range of the allowable rate of change of original power based on the abnormal characteristic identification and comprehensive deviation index of the energy storage battery pack is as follows:

[0097] Based on the abnormal feature identifiers and comprehensive deviation index of the energy storage battery pack, the warning ratio range of the rated power and the warning ratio range of the original power change rate allowable value of the energy storage battery pack are obtained from the pre-constructed warning abnormal feature-power constraint matching rule table. The pre-constructed warning abnormal feature-power constraint matching rule table predefines the warning ratio range of the rated power and the warning ratio range of the original power change rate allowable value corresponding to different abnormal feature identifiers and comprehensive deviation index intervals.

[0098] The warning-level cooling control command, that is, according to the priority of the energy storage battery pack, calls the refrigerant storage to perform liquid cooling circulation according to the preset refrigerant flow rate, so as to suppress the heat accumulation rate of the energy storage battery pack and delay the evolution of abnormal state. The preset refrigerant flow rate is a benchmark cooling flow rate value determined by those skilled in the art through thermal balance based on the thermal design power of the battery module and the rated heat dissipation capacity of the cooling system.

[0099] The refrigerant storage method is as follows: During the operation of the energy storage circuit, the operating segment identifier corresponding to the energy storage battery box and the thermal load status parameters of the battery module are continuously acquired. When the operating segment identifier is low-load charging operation, the refrigerant storage mode is entered. The heat generated by the energy storage battery box during charging and discharging is continuously transferred to the intelligent refrigerant box through the independent heat exchanger built into the energy storage battery box, and the refrigerant is cooled. In the refrigerant storage mode, the refrigerant flows in a circulating manner in the intelligent refrigerant box, and the corresponding refrigerant callable capacity parameters are recorded. When the operating segment identifier switches to the high-load discharge operation range, the active cooling action is stopped, and the refrigerant that has completed refrigerant storage is called to participate in the liquid cooling circulation of the energy storage battery box. By driving the refrigerant to flow through the heat exchange path in the energy storage battery box according to the preset refrigerant flow rate, the heat inside the energy storage battery box is continuously absorbed and transferred, thereby suppressing the heat accumulation rate of the energy storage battery box and delaying the evolution of abnormal states without increasing the cooling energy consumption, thus providing a basis for extending battery life.

[0100] In this embodiment, the early warning-level coordinated control process helps to adopt a conservative strategy of reducing load and stabilizing temperature in the early stages of a risk. Through limited and reversible power constraints and enhanced heat dissipation by utilizing pre-stored cooling capacity, it aims to bring the abnormal state back to normal without seriously affecting the overall output performance of the system. It breaks the initial stage of the positive feedback loop from increased heat generation to temperature rise, reduces heat generation sources through power constraints, and enhances heat dissipation capacity by calling on stored cooling capacity, thus buying time for the battery module to recover or wait for further maintenance. Priority sorting ensures that limited cooling resources are given priority to the unit that first shows signs of risk, optimizing the allocation of safety resources.

[0101] Step a3: When the thermal runaway risk of the energy storage battery pack is identified as intervention level, intervention-level coordinated control processing is performed on the battery module to which the energy storage battery pack belongs, based on the abnormal characteristic identification and comprehensive deviation index of the energy storage battery pack. This includes sending intervention-level power dynamic constraint commands, intervention-level topology isolation protection commands, and intervention-level cooling enhancement control commands to the energy storage loop control system.

[0102] The intervention-level power dynamic constraint command, based on the early warning level coordinated control, dynamically limits the maximum allowable output power and power change rate of the battery modules belonging to the energy storage battery box. According to the abnormal characteristic identification and comprehensive deviation index of the energy storage battery box, the intervention ratio range of rated power and the intervention ratio range of the original power change rate are analyzed and obtained. The maximum allowable output power of the battery module is dynamically limited to the intervention ratio range of rated power, and the power change rate is limited to the intervention ratio range of the original power change rate.

[0103] Based on the abnormal characteristics and comprehensive deviation index of the energy storage battery pack, the method for analyzing and obtaining the intervention ratio range of the rated power and the intervention ratio range of the allowable value of the original power change rate is as follows:

[0104] Based on the abnormal feature identifiers and comprehensive deviation index of the energy storage battery pack, the intervention ratio range of the rated power and the intervention ratio range of the original power change rate allowable value of the energy storage battery pack are obtained from the pre-constructed intervention abnormal feature-power constraint matching rule table. The pre-constructed intervention abnormal feature-power constraint matching rule table predefines the corresponding intervention ratio range of the rated power and the intervention ratio range of the original power change rate allowable value for different abnormal feature identifiers and comprehensive deviation index intervals.

[0105] The specific setting rules for the warning anomaly feature—power constraint matching rule table and the intervention anomaly feature—power constraint matching rule table are as follows:

[0106] In the warning anomaly feature-power constraint matching rule table, the anomaly feature identifier and the comprehensive deviation index range are used as a joint index item. For each combination, a corresponding warning ratio range for rated power and a warning ratio range for the original power change rate are pre-set. The warning ratio range for rated power is used to limit the maximum allowable output power range of the battery module during the warning phase, and its value follows a monotonically decreasing principle: the larger the deviation index, the lower the allowable power limit. The warning ratio range for the original power change rate is used to limit the dynamic amplitude of power increase or decrease, and its value follows a principle: the larger the deviation index, the lower the change rate limit. The stricter the setting principle, in a specific embodiment, when the abnormal feature is identified as a thermal feature and the comprehensive deviation index is in the range of [1.0, 2.0), the warning ratio range of the rated power is set to [70%, 80%]; when the abnormal feature is identified as an electrical feature and the comprehensive deviation index is in the range of [2.0, 3.0), the warning ratio range of the rated power is set to [50%, 60%]. The specific values ​​of each ratio range are determined by those skilled in the art based on historical stable operation data, abnormal evolution data during the warning stage, and the response capability of the thermal management system through statistical analysis, and can be adaptively adjusted according to different battery models and application scenarios.

[0107] Intervention-level topology isolation protection command, that is, to perform topology isolation control on the energy storage circuit where the battery module is located, the topology isolation relay will disconnect the corresponding battery module from the energy storage circuit, thereby limiting the propagation path of abnormal state in the energy storage circuit, so that the battery module has controlled isolation and linkage safety protection capabilities.

[0108] The intervention-level cooling enhancement control command adjusts the liquid cooling flow rate to the preset maximum liquid cooling flow rate of the corresponding energy storage battery pack, thereby improving the circulation intensity of the cooling medium. The preset maximum liquid cooling flow rate is determined by those skilled in the art based on the operation data of the corresponding energy storage battery pack during the historical stable operation phase and abnormal evolution phase. By statistically processing the collected liquid cooling flow rate, temperature drop rate, and thermal characteristic change trend data, the average level of liquid cooling flow rate under the condition of suppressing continuous temperature rise and not causing abnormal cooling system load is selected as the maximum liquid cooling flow rate.

[0109] In this embodiment, the intervention-level coordinated control process helps to adopt a more stringent combination strategy of source control, isolation, and strong cooling when early warning measures fail to prevent the risk from escalating. The further constraint of power aims to significantly reduce the heat generation of the abnormal module itself; topology isolation physically disconnects it from the healthy circuit electrically, effectively blocking the path of fault propagation, which is a key step to prevent the situation from escalating into a system-level accident; and cooling enhancement is a key cooling of the isolated module, attempting to suppress the heat accumulation process inside it. This series of measures constitutes a precise handling of medium- and high-risk objects. Its principle is to sacrifice the operating capability of a module locally in exchange for the safety and stability of the entire energy storage circuit.

[0110] Step a4: When the thermal runaway risk of the energy storage battery box is identified as an emergency level, emergency-level coordinated control processing is performed on the battery module to which the energy storage battery box belongs, including: sending emergency-level power dynamic constraint commands, emergency-level electrical isolation control commands, and emergency-level safety protection switching commands to the energy storage circuit control system.

[0111] The emergency-level power dynamic constraint command means that the energy storage circuit control system switches the power output control state of the corresponding battery module to the power suppression state, so that the actual output power of the battery module is limited to the preset emergency-level power suppression range, thereby reducing the energy release rate inside the battery module. In a specific embodiment, the preset emergency-level power suppression range is a rated power range of (0%, 10%) set by those skilled in the art through analysis of historical data, which aims to minimize the energy release rate of the battery module.

[0112] The emergency-level electrical isolation control command switches the electrical connection between the corresponding battery module and the energy storage circuit to an isolated state, disconnecting the corresponding battery module from the energy storage circuit and thus blocking the energy exchange path between the corresponding battery module and the energy storage circuit.

[0113] The emergency-level safety protection switching command closes the insulating coolant circuit valve of the corresponding battery module and opens the backup solenoid valve of the fire-fighting medium passage of the corresponding battery module. This ensures that there is a control basis for immediate execution of mandatory safety measures in emergency-level abnormal conditions and suppresses the risk of the abnormal condition spreading to adjacent battery modules.

[0114] In this embodiment, the emergency-level coordinated control process helps to implement a defense centered on energy cutoff and safety protection preparation when the risk of thermal runaway is determined to be extremely high and may rapidly evolve into an irreversible event. Its working principle is: by suppressing the power to an extremely low level and performing electrical isolation, the energy input and output of the faulty module are completely cut off; at the same time, the cooling circuit is switched to a fire-fighting standby state to prepare immediate fire-fighting measures for possible thermal runaway. The core objective at this stage has shifted from controlling the anomaly to isolating the risk and preparing to deal with the worst-case scenario, aiming to control potential hazards to the smallest possible physical area, individual modules, and protect the safety of personnel and main equipment. The entire hierarchical linkage process reflects a progressive safety defense concept from early warning and delay to intervention control and then to emergency isolation and protection, rather than a traditional single-action response mode.

[0115] S5: After performing hierarchical coordinated control, continuously monitor the operating characteristic vector of the energy storage battery box, generate the runaway signal of the battery module, and the energy storage circuit control system performs electrical isolation processing on the energy storage circuit corresponding to the battery module based on the runaway signal of the battery module. After performing electrical isolation processing, perform directional immersion processing according to the active fire protection logic.

[0116] The method for generating the runaway signal of the battery module is as follows:

[0117] After step S4 is executed, the operating feature vector of the energy storage battery box is continuously constructed through step S1, and the thermal runaway risk identifier of the energy storage battery box in the battery module during continuous operation cycle is obtained by analysis.

[0118] When the thermal runaway risk indicator of the energy storage battery pack is continuously marked as emergency level in the analysis window, it is determined that the thermal runaway risk indicator of the energy storage battery pack meets the condition for continuous emergency level establishment.

[0119] The operating characteristic vector of the energy storage battery box in the corresponding battery module is acquired synchronously with the analysis window range. If the hydrogen concentration and carbon monoxide concentration in the energy storage battery box increase monotonically and are higher than the preset emergency threshold of the corresponding gas concentration, and the temperature difference change rate is continuously positive and is higher than the preset temperature difference change rate threshold, the thermal runaway evolution trend is determined to meet the irreversible condition.

[0120] In this embodiment, the method for setting the emergency threshold for gas concentration is as follows: Those skilled in the art, based on background gas concentration data released during the historical normal operation of the energy storage battery module of this model, take the statistical average of the background concentration plus three times the standard deviation as the emergency threshold for the corresponding gas concentration. For example, if the average background concentration of hydrogen is 5 ppm and the standard deviation is 15 ppm, then the emergency threshold for hydrogen concentration is set to 50 ppm; if the average background concentration of carbon monoxide is 2 ppm and the standard deviation is 6 ppm, then the emergency threshold for carbon monoxide concentration is set to 20 ppm. Those skilled in the art can adaptively adjust this threshold according to the specific battery chemistry system and operating environment. The method for setting the temperature difference change rate threshold is as follows: Based on the thermal runaway acceleration experimental data of this model of energy storage battery module, the critical rate of change at which the temperature gradient begins to deteriorate nonlinearly before thermal runaway occurs is selected as the temperature difference change rate threshold. For example, through experimental analysis, the temperature difference change rate threshold is set to 0.8℃ / min. This temperature difference change rate threshold can be adaptively adjusted according to the thermal design parameters and heat dissipation conditions of the battery module.

[0121] When the thermal runaway risk indicator meets the conditions for the emergency level to remain in effect within the analysis window, and the thermal runaway evolution trend meets the conditions for irreversibility, a runaway signal for the battery module is generated. The runaway signal is then used as the trigger for subsequent mandatory safety measures in the energy storage circuit. Otherwise, no runaway signal for the battery module is generated, and the hierarchical coordinated control state continues to be executed.

[0122] In this embodiment, the runaway signal generated by the battery module is not triggered by a single instantaneous indicator, but rather by constructing an AND gate logic that simultaneously satisfies a continuous emergency state and a multi-dimensional irreversible physical evolution trend. This design effectively filters false alarms caused by instantaneous sensor interference or fluctuations in a single parameter, improving the accuracy and reliability of the runaway signal generation. It ensures that the highest level of safety measures are only triggered when thermal runaway truly develops to a stage where the physicochemical reactions are intense and difficult to reverse through conventional cooling, thereby avoiding unnecessary system downtime and property damage. The principle behind this is that thermal runaway is an evolutionary process from quantitative change to qualitative change. The emergency level determined by a simple algorithm may be biased due to model uncertainties, while the monotonous increase in gas concentration and the accelerated expansion of temperature difference are direct physical evidence of the chain reaction of thermal runaway and the occurrence of thermal propagation. The combination of these two factors constitutes a more reliable cross-validation of the qualitative change point of thermal runaway.

[0123] The method for performing electrical isolation treatment on the energy storage circuit corresponding to the battery module is as follows:

[0124] After generating a runaway signal for the battery module, the energy storage loop control system sends an emergency isolation control command. Upon receiving the emergency isolation control command, the energy storage loop control system switches the power output control state of the corresponding battery module to zero power state and sends a disconnection control command to the main DC contactor controlling the corresponding battery module. The main DC contactor of the corresponding battery module refers to a high-power electromagnetic switch device installed at the output terminal of the corresponding battery module, used to physically disconnect the electrical connection between the battery module and the main DC bus of the energy storage system under an emergency command.

[0125] The disconnection control command indicates that the electrical isolation process of the energy storage circuit is complete when the main DC contactor of the corresponding battery module is in the open state and the voltage at the terminal of the corresponding battery module drops to a preset safe isolation judgment range. The safe isolation judgment voltage range is set to [0V, 5% of the rated voltage]. The upper limit of this range is set to take into account the superimposed effects of the residual induced voltage of the module after the storage circuit is disconnected, the distributed capacitance voltage to ground, and the noise voltage of the measurement circuit, to ensure that safe isolation has been achieved from an electrical measurement perspective. Those skilled in the art can adaptively adjust this percentage according to the rated voltage level of the system and the measurement accuracy.

[0126] In this embodiment, electrical isolation processing establishes the final safety boundary from logic control to physical isolation. By forcibly reducing the power output to zero and driving the physical contactor to disconnect, the energy interaction path between the faulty module and the energy storage circuit is completely cut off, effectively preventing the further expansion of the fault scale and electrical impact on other parts of the system. This creates a safe electrical environment for subsequent fire-fighting operations. The dual confirmation mechanism of status feedback and voltage verification reduces the risk that a single command signal may fail due to actuator failure, ensuring the final effectiveness of the isolation operation and the verifiability of the system's safe state.

[0127] The method for performing directional immersion treatment according to the active fire suppression logic after electrical isolation is as follows:

[0128] Extract the operating feature vectors of all energy storage battery boxes within the battery module, and arrange them in order of their physical location to form the module state vector of the battery module.

[0129] Based on the module state vector of the corresponding battery module within the analysis window, the specific process of performing joint judgment processing on smoke characteristic state and temperature characteristic state, and then performing directional immersion processing, is as follows: The smoke concentration and temperature of the corresponding battery module are collected. When the smoke concentration is higher than the preset smoke alarm threshold and the temperature is lower than the high-temperature trigger threshold, the fire extinguishing medium injection valve of the corresponding battery module is opened. The determination of fire extinguishing termination is based on whether the liquid level reaches the preset upper limit or the injection duration reaches the preset termination time, as measured by the liquid level sensor. The smoke alarm threshold is determined by those skilled in the art based on the sensitivity specifications of the selected smoke detector and the typical background smoke concentration in the energy storage chamber; for example, it can be set to 3%obs / m. The high-temperature trigger threshold is set by those skilled in the art based on the starting temperature at which the battery electrolyte begins to decompose and produce gas in large quantities; for example, it can be set to 120℃. The preset upper limit of the liquid level is set by those skilled in the art based on 70% of the effective volume of the module shell to ensure that the fire extinguishing medium can effectively immerse the battery. The preset termination time is obtained by those skilled in the art based on the design flow rate of the fire pipeline and the volume of the module shell; for example, it can be set to 60 seconds.

[0130] When the temperature exceeds the high-temperature trigger threshold and the thermal characteristics and temperature difference evolution characteristics in the module's state vector meet the structural safety response trigger conditions, the corresponding battery module's enhanced fire extinguishing medium injection valve is opened. The determination of fire extinguishing termination is based on whether the liquid level measured by the level sensor reaches the enhanced liquid level upper limit or the enhanced injection duration reaches the preset enhanced termination time. The enhanced liquid level upper limit is set to 90% of the effective volume of the module's outer shell. The preset enhanced termination time is 50% higher than the preset termination time, for example, it can be set to 90 seconds. Those skilled in the art can adaptively adjust the above liquid level and duration parameters according to the specific module size, fire extinguishing medium type, and pipeline design.

[0131] It should be noted that when the surface temperature of any energy storage battery box in the battery module is higher than the preset high temperature trigger threshold, or the temperature difference change rate of the liquid cooling circuit is higher than zero and the temperature difference change rate of the liquid cooling circuit exceeds the preset temperature difference change rate threshold, the temperature difference evolution characteristics are determined to meet the structural safety response trigger condition. The preset temperature difference change rate threshold is set to 2.0℃ / min. The preset temperature difference change rate threshold is set based on the maximum temperature rise gradient rate that may occur between adjacent energy storage battery boxes or different areas in the module during the thermal runaway propagation process. It is obtained through thermal runaway propagation experiments. Those skilled in the art can fine-tune this threshold according to the battery packing density and thermal resistance characteristics.

[0132] In this embodiment, the directional immersion treatment implements differentiated firefighting strategies based on different stages of thermal runaway development and the uneven heat distribution within the module. Through combined smoke-temperature judgment and structural safety response trigger condition assessment, it achieves precise matching of firefighting intensity from suppression to extinguishment and deep cooling. This not only improves firefighting efficiency and reduces extinguishing medium consumption, but more importantly, it initiates enhanced immersion to rapidly reduce the battery body temperature below the reaction termination point for risks of structural thermal spread. The high heat capacity of the liquid medium isolates the heat source, effectively preventing the deep spread of thermal runaway within the module and its propagation to adjacent modules. Its principle breaks away from the traditional reliance on a single threshold in firefighting systems. Through multi-dimensional state vector analysis, it correlates the firefighting response with the actual physical damage caused by thermal runaway, achieving refined and cost-effective safety management.

[0133] This invention provides holistic constraints and layered prevention and control of thermal runaway risks in energy storage systems. By sensing the state, assessing risks, and executing control measures for three levels of objects—energy storage battery boxes, battery modules, and energy storage circuits—a clear hierarchical runaway risk control architecture is formed. This architecture organically integrates functions such as anomaly identification, risk classification, coordinated control, electrical isolation, and proactive fire suppression into a unified decision-making logic chain. This allows thermal runaway risks to be intercepted and reduced step by step along a preset safety path, avoiding the protection failures or repetitive actions caused by the independent operation and disconnected responses of various safety subsystems in traditional systems.

[0134] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0135] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention.

Claims

1. A graded linkage safety control method for thermal management of energy storage loops, characterized in that, include: S1: The energy storage loop control system continuously and synchronously monitors the multi-dimensional operating status parameter set of the energy storage battery box in the battery module of the energy storage loop, and constructs the operating feature vector of the energy storage battery box. The thermal characteristic data of the energy storage battery boxes in the same battery module are analyzed to obtain the maximum temperature difference of the energy storage battery boxes. An analysis window is formed by a preset number of continuous operating cycles. The maximum temperature difference of the energy storage battery boxes in the analysis window is recorded to form a numerical sequence of temperature difference of the energy storage battery boxes. The slope of the linear regression of the temperature difference numerical sequence of the energy storage battery compartment is taken as the rate of change of the temperature difference of the energy storage battery compartment. The standard deviation of the temperature difference numerical sequence of the energy storage battery compartment is used as the temperature difference fluctuation range of the energy storage battery compartment. Based on the temperature difference numerical sequence, temperature difference change rate, and temperature difference fluctuation amplitude of the energy storage battery box, temperature difference evolution characteristic data of the energy storage battery box are constructed. The electrical, thermal, gaseous, temperature difference evolution, and operational characteristic data of the energy storage battery pack are structurally combined to obtain the operational characteristic vector of the energy storage battery pack. S2: Perform correlation analysis on the operating feature vector of the energy storage battery pack to obtain abnormal state judgment information of the energy storage battery pack; Extract the abnormal risk probability value of the energy storage battery box, and map the abnormal risk probability value to the corresponding abnormal state judgment interval based on the predefined abnormal state judgment interval table to obtain the abnormal level of the energy storage battery box. The abnormal risk probability value, abnormal level, abnormal feature identifier, and the feature data subset corresponding to the abnormal feature identifier of the energy storage battery box are encapsulated to obtain the abnormal state judgment information of the energy storage battery box. S3: The energy storage loop control system analyzes the abnormal state evolution of the energy storage battery pack based on the abnormal state judgment information of the energy storage battery pack, and obtains the thermal runaway risk identification of the energy storage battery pack. Receive abnormal state judgment information of energy storage battery packs, and serialize and arrange them in chronological order to form an abnormal state time evolution sequence of energy storage battery packs. Based on the time evolution sequence of abnormal states of the energy storage battery pack, the thermal runaway risk of the energy storage battery pack is classified and determined to obtain the thermal runaway risk label of the energy storage battery pack. The thermal runaway risk indicators for the energy storage battery pack include warning level, intervention level, and emergency level; S4: The energy storage loop control system performs hierarchical coordinated control on the battery modules in the energy storage loop based on the thermal runaway risk identification of the energy storage battery pack; S5: After performing hierarchical coordinated control, continuously monitor the operating characteristic vector of the energy storage battery box, generate the runaway signal of the battery module, and the energy storage circuit control system performs electrical isolation processing on the energy storage circuit corresponding to the battery module based on the runaway signal of the battery module. After performing electrical isolation processing, perform directional immersion processing according to the active fire protection logic.

2. The thermal management hierarchical linkage safety control method for energy storage circuits according to claim 1, characterized in that, The method for the energy storage loop control system to continuously and synchronously monitor the multi-dimensional operating status parameter set of the energy storage battery pack within the energy storage loop is as follows: The system collects a multi-dimensional set of operating status parameters for the energy storage battery pack within the battery module of the energy storage circuit. These parameters include electrical characteristic data, thermal characteristic data, gas characteristic data, and operational characteristic data of the energy storage battery pack.

3. The thermal management hierarchical linkage safety control method for energy storage circuits according to claim 1, characterized in that, The method for performing correlation analysis on the operating feature vector of the energy storage battery pack is as follows: The operating feature vector of the energy storage battery pack is input into the pre-trained abnormal state judgment model, and the output is the abnormal risk probability value, abnormal feature identifier, and feature data subset corresponding to the abnormal feature identifier of the energy storage battery pack. The abnormal feature identifiers include electrical features, thermal features, gas features, temperature difference evolution features, and operational features; the feature data subsets corresponding to the abnormal feature identifiers include electrical feature data corresponding to electrical features, thermal feature data corresponding to thermal features, gas feature data corresponding to gas features, temperature difference evolution feature data corresponding to temperature difference evolution features, and operational feature data corresponding to operational features.

4. The thermal management hierarchical linkage safety control method for energy storage circuits according to claim 1, characterized in that, The method for performing hierarchical coordinated control of battery modules in the energy storage circuit is as follows: Step a1: Analyze the feature data subset corresponding to the abnormal feature identifier of the energy storage battery box to obtain the comprehensive deviation index of the energy storage battery box. The comprehensive deviation index of the energy storage battery box is used to quantify the degree to which the current operating state of the energy storage battery box deviates from the historical stable operating benchmark. Step a2: Extract the thermal runaway risk identifier of the energy storage battery box. When the thermal runaway risk identifier of the energy storage battery box is at the warning level, perform warning-level coordinated control processing on the battery module to which the energy storage battery box belongs based on the abnormal feature identifier and comprehensive deviation index of the energy storage battery box. This includes sending the battery box risk priority sorting instruction, power dynamic constraint instruction and warning-level cooling control instruction to the energy storage loop control system. Step a3: When the thermal runaway risk of the energy storage battery pack is identified as intervention level, based on the abnormal characteristic identification and comprehensive deviation index of the energy storage battery pack, intervention-level coordinated control processing is performed on the battery module to which the energy storage battery pack belongs, including: sending intervention-level power dynamic constraint command, intervention-level topology isolation protection command and intervention-level cooling enhancement control command to the energy storage loop control system; Step a4: When the thermal runaway risk of the energy storage battery box is identified as an emergency level, emergency-level coordinated control processing is performed on the battery module to which the energy storage battery box belongs, including: sending emergency-level power dynamic constraint commands, emergency-level electrical isolation control commands, and emergency-level safety protection switching commands to the energy storage circuit control system.

5. The thermal management hierarchical linkage safety control method for energy storage circuits according to claim 1, characterized in that, The method for generating the runaway signal of the battery module is as follows: The operation feature vector of the energy storage battery box is continuously constructed, and the thermal runaway risk indicator of the energy storage battery box in the battery module is obtained by analysis. When the thermal runaway risk indicator of the energy storage battery box is continuously marked as the emergency level, it is determined that the thermal runaway risk indicator of the energy storage battery box meets the condition of continuous establishment of the emergency level. Simultaneously acquire the operating characteristic vector of the energy storage battery pack in the corresponding battery module to determine whether the thermal runaway evolution trend meets the irreversible condition; A battery module runaway signal is generated when the thermal runaway risk indicator meets the conditions for the emergency level to remain in effect and the thermal runaway evolution trend meets the conditions for irreversibility; otherwise, no battery module runaway signal is generated.

6. The thermal management hierarchical linkage safety control method for energy storage circuits according to claim 5, characterized in that, The method for performing electrical isolation processing on the energy storage circuit corresponding to the battery module is as follows: After generating a runaway signal for the battery module, an emergency isolation control command is sent to switch the power output control state of the corresponding battery module to zero power state and send a disconnect control command to the main DC contactor controlling the corresponding battery module.

7. The thermal management hierarchical linkage safety control method for energy storage circuits according to claim 6, characterized in that, The method for performing directional immersion treatment according to active fire protection logic after performing electrical isolation treatment is as follows: Extract the operating feature vectors of all energy storage battery boxes within the battery module, and arrange them in order of their physical location to form the module state vector of the battery module. Based on the module state vector of the corresponding battery module, a joint determination process is performed on the smoke characteristic state and the temperature characteristic state, and then a directional immersion process is performed.

Images