A thermal runaway trend perception method for energy storage plants

Abstract

The application discloses a thermal runaway trend perception method for an energy storage power station, and relates to the technical field of safety monitoring of energy storage power stations.The method comprises the following steps: firstly, loading the facility distribution topology and the energy storage battery distribution position of the current energy storage power station; taking the energy storage battery distribution position as a thermal runaway starting point and the energy storage facility distribution topology as a scene constraint, searching for a thermal runaway transaction set and counting a spreading path; meanwhile, based on the high-frequency operation parameters of the energy storage battery distribution position, the environmental temperature threshold is determined through a temperature optimization process; finally, the energy storage battery distribution position, the spreading path and the environmental temperature threshold are stored in a thermal runaway trend mapping table in association, and thus the forward-looking thermal runaway trend perception is realized.The application combines the spatial layout and the operation parameters, realizes the accurate prediction of the thermal runaway spreading path and the triggering condition, and effectively improves the safety early warning capability of the energy storage power station.

Description

Technical Field

[0001] This invention relates to the field of safety monitoring technology for energy storage power stations, and specifically to a method for sensing thermal runaway trends in energy storage power stations. Background Technology

[0002] Thermal runaway is a serious safety hazard during the operation of energy storage power stations. It can be triggered by a single battery cell and spread between battery clusters or stacks as heat accumulates and is transferred, ultimately leading to a large-scale safety incident.

[0003] In existing technologies, monitoring of thermal runaway often relies on passive monitoring or real-time monitoring of the temperature or voltage of individual batteries. This approach suffers from significant lag and lacks a holistic assessment of the spatial propagation trend of thermal runaway. Meanwhile, ambient temperature is crucial to the thermal state of batteries, but current methods fail to effectively integrate the physical layout of energy storage facilities with historical operating parameters to accurately define the critical environmental conditions that trigger thermal runaway in battery cells at different locations under specific operating modes. Therefore, it is difficult to accurately perceive trends and provide early warnings of risks in the early stages of thermal runaway, failing to meet the forward-looking safety management requirements of large-scale energy storage power plants. Summary of the Invention

[0004] This invention addresses the technical problem of the lack of early and accurate perception and warning capabilities for thermal runaway propagation trends in existing technologies by providing a method for sensing thermal runaway trends in energy storage power plants.

[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:

[0006] This invention provides a method for sensing thermal runaway trends in energy storage power stations, comprising:

[0007] The energy storage facility distribution topology of the loaded energy storage power station includes the distribution locations of energy storage batteries;

[0008] Using the location of the energy storage battery distribution as the thermal runaway starting point and the topology of the energy storage facility distribution as the scenario constraint, the thermal runaway event set is retrieved and the propagation path is statistically analyzed.

[0009] Load the predefined high-frequency operating parameters of the energy storage battery distribution location, perform runaway environment temperature optimization, and obtain the environment temperature threshold;

[0010] The ambient temperature threshold, the propagation path, and the distribution location of the energy storage battery are associated and stored, and added to the thermal runaway trend mapping table;

[0011] Based on the thermal runaway trend mapping table, thermal runaway trend sensing is performed.

[0012] The beneficial effects of this invention are:

[0013] Compared to existing technologies, this invention first treats thermal runaway as a process with spatial propagation characteristics by combining the physical layout information of energy storage facilities. This allows for the statistical analysis of potential propagation paths based on historical data, enabling macroscopic prediction of thermal runaway propagation trends. Secondly, it introduces the analysis of high-frequency operating parameters of battery distribution locations and performs refined runaway environment temperature optimization, thereby determining more accurate and personalized thermal runaway environment temperature trigger thresholds for battery cells in specific locations. Thirdly, by constructing a trend mapping table that associates location, path, and temperature thresholds, it effectively integrates and structures multi-source information, providing a solid data foundation for trend perception. Finally, performing trend perception based on this comprehensive mapping table significantly improves the accuracy and foresight of early warnings, enabling earlier and more reliable identification and intervention of thermal runaway risks in energy storage power stations. Attached Figure Description

[0014] Figure 1 A flowchart illustrating a thermal runaway trend sensing method for an energy storage power station provided by the present invention;

[0015] Figure 2 This is a logical diagram illustrating a thermal runaway trend sensing method for energy storage power stations provided by the present invention. Detailed Implementation

[0016] 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.

[0017] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0018] In the description of this invention, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed herein.

[0019] Example 1, as Figure 1 , Figure 2 As shown, this embodiment of the invention provides a method for sensing thermal runaway trends in energy storage power stations, including:

[0020] S10: Load the energy storage facility distribution topology of the energy storage power station, wherein the energy storage facility distribution topology has the distribution location of energy storage batteries;

[0021] Specifically, the energy storage facility distribution topology is a structured data model used to characterize the overall physical architecture of an energy storage power station. It describes the physical layout and connection relationships between various energy storage facilities within the power station, and is a structured spatial location model. This energy storage facility distribution topology explicitly includes the specific distribution location information of the energy storage batteries, that is, the specific coordinates of each energy storage battery cell or cluster within the power station space. The process of loading the energy storage facility distribution topology of the energy storage power station typically involves reading and parsing the energy storage facility distribution topology data from a pre-set power station configuration database or digital design file, putting it in a ready state for subsequent steps. By obtaining the precise distribution location of the energy storage batteries, a necessary data foundation can be laid for spatial location-based thermal runaway analysis.

[0022] S20: Using the location of the energy storage battery distribution as the thermal runaway starting point and the topology of the energy storage facility distribution as the scenario constraint, retrieve the thermal runaway event set and count the propagation path;

[0023] First, taking the location of the energy storage batteries as the starting point of thermal runaway and the topology of the energy storage facility distribution as the scenario constraint, a set of thermal runaway events is retrieved, including:

[0024] Load the thermal runaway transaction to be analyzed, wherein the thermal runaway transaction to be analyzed includes the thermal runaway record origin and the energy storage facility distribution record topology;

[0025] When the energy storage facility distribution similarity between the energy storage facility distribution topology and the energy storage facility distribution record topology is greater than or equal to the energy storage facility distribution similarity threshold, and the thermal runaway initiation point is the same as the thermal runaway record initiation point, the thermal runaway transaction to be analyzed is added to the thermal runaway transaction set.

[0026] Specifically, taking the location of energy storage batteries as the starting point of thermal runaway and the topology of energy storage facility distribution as the scenario constraint, the thermal runaway event set is retrieved. This is a screening process based on spatial scenario similarity and starting point consistency, used to find thermal runaway event cases that are highly relevant to the current analysis scenario from historical records.

[0027] First, the thermal runaway events to be analyzed are loaded, which are historical thermal runaway events collected in the database. Each thermal runaway event contains two key pieces of information: the starting point of the thermal runaway record, which records the initial location of the energy storage battery where the thermal runaway event occurred in history; and the topology of the energy storage facility distribution record, which records the spatial layout of the energy storage power station at the time of the thermal runaway event. Loading the thermal runaway events to be analyzed can build a filterable historical event database, providing a data source for subsequent extraction of relevant cases based on similarity matching.

[0028] Secondly, a filtering and matching operation is performed. The currently loaded energy storage facility distribution topology is compared with the energy storage facility distribution record topology in the transaction to be analyzed, and the similarity in facility layout is calculated, i.e., the energy storage facility distribution similarity. Specifically, the energy storage facility distribution similarity can be calculated by comparing the positional relationships, connectivity relationships, and relative distances of corresponding facility nodes in the two topologies, and can be quantified using graph structure matching algorithms or spatial overlap algorithms. If the calculated energy storage facility distribution similarity reaches or exceeds the preset energy storage facility distribution similarity threshold, and the currently specified thermal runaway initiation point and the thermal runaway record in the historical transaction point to the same energy storage battery location, then the thermal runaway transaction to be analyzed is determined to be highly correlated with the current scenario. The preset energy storage facility distribution similarity threshold is set based on the requirements for scenario matching accuracy and combined with historical data analysis experience, such as 0.8 or 80%.

[0029] Finally, all thermal runaway events that meet the conditions of similarity in energy storage facility distribution and consistency in initiation point are included in the thermal runaway event set. This thermal runaway event set is a selected collection of historical thermal runaway events that are highly relevant to the current analysis scenario. It contains a series of historical thermal runaway event records with similar spatial layouts and the same ignition point location, which can serve as a reliable data foundation for subsequent statistical propagation paths.

[0030] Furthermore, the statistical spread path includes:

[0031] From the thermal runaway event set, extract the first propagation path and the distribution record topology of the first energy storage facility, wherein the first propagation path includes the tag number sequence of the first propagation device;

[0032] Extract the first spread device tag number sequence from the first spread location sequence of the first energy storage facility distribution record topology;

[0033] Based on the first spread location sequence, the second spread device tag sequence is extracted from the energy storage facility distribution topology and added to the initial spread path set;

[0034] Based on the initial set of propagation paths, the propagation paths are statistically analyzed.

[0035] The propagation path describes the spatial sequence of a thermal runaway event propagating among energy storage power plant facilities, representing the possible directions and chronological order of the fault spreading from its initial location to surrounding equipment. Statistical propagation path analysis is the process of summarizing and extracting typical propagation patterns from historical events, providing a basis for predicting the development trend of thermal runaway under the current power plant layout. Specifically, the steps for statistical propagation path analysis are as follows:

[0036] First, extract individual historical transaction data from the established thermal runaway transaction set, specifically the first propagation path and its corresponding first energy storage facility distribution record topology. The first propagation path records the sequence of devices affected by the thermal runaway phenomenon in this historical event, specifically a first propagation device tag number sequence containing multiple device tag numbers.

[0037] Secondly, based on the first propagation device tag number sequence, the actual physical location corresponding to each device tag number is determined in the topology of the first energy storage facility distribution record, thus obtaining the first propagation location sequence. This first propagation location sequence depicts the specific propagation trajectory of thermal runaway under a specific historical spatial layout.

[0038] Then, the obtained first spread location sequence is mapped to the current energy storage facility distribution topology. Through location matching, devices in the same or corresponding locations in the current energy storage facility distribution topology are identified, and the tag number sequence of these devices is recorded to form the second spread device tag number sequence. This second spread device tag number sequence is added to the initial spread path set as a possible spread path based on the adaptation result of the current energy storage facility distribution topology to historical propagation paths, thus completing the path transformation from historical spatial layout to current spatial layout.

[0039] Finally, based on the constructed initial spread path set containing multiple transformed paths, a specific statistical algorithm is used for analysis to extract the most representative or most probable spread path as the final analysis result.

[0040] Specifically, based on the initial set of propagation paths, the propagation paths are statistically analyzed, including:

[0041] Cluster analysis is performed on the initial spread path set to obtain multiple clusters of initial spread paths;

[0042] Based on the initial spread paths of the multiple clusters, delete the initial spread paths within the cluster whose number is less than or equal to the number threshold within the cluster to obtain the spread paths.

[0043] First, cluster analysis is performed on the initial spread path set to obtain multiple clusters of initial spread paths, including:

[0044] Extract the first initial spread path and the second initial spread path from the initial spread path set;

[0045] Align the starting points of the first initial spread path and the second initial spread path, and count the number of identical position numbers;

[0046] Extract the total number of longer path indices from the first initial spread path and the second initial spread path;

[0047] If the ratio of the number of corresponding serial numbers to the total number of serial numbers in the longer path is greater than or equal to the ratio threshold, the first initial spread path and the second initial spread path are added to the same cluster; otherwise, they are set to different clusters.

[0048] Specifically, the process of performing cluster analysis on the initial set of spreading paths to obtain multi-cluster paths is achieved by comparing the similarity between paths.

[0049] First, two paths are extracted from the initial spread path set for analysis: the first initial spread path and the second initial spread path. Second, the starting points of the first and second initial spread paths are aligned to ensure the comparison is based on the same starting position. The number of devices with the same tag number at the same position on both paths is counted, i.e., the number of identical tag numbers. Simultaneously, the total number of tags on the longer path is extracted as the comparison benchmark. Then, the ratio of the number of identical tag numbers to the total number of tags on the longer path is calculated; this ratio reflects the similarity between the two paths. When this ratio reaches or exceeds a preset ratio threshold, the two paths are considered highly similar and are grouped into the same cluster; if the ratio is below the preset threshold, the two paths are set as different clusters.

[0050] By comparing each pair of paths, the initial set of propagation paths is divided into multiple clusters, each containing paths with similar propagation characteristics, thus laying the foundation for subsequent selection of representative propagation paths.

[0051] Furthermore, based on the initial propagation paths of multiple clusters, initial propagation paths with a number less than or equal to a cluster number threshold are deleted to obtain the propagation paths. Specifically, the clustering results are filtered according to a preset cluster number threshold, eliminating initial propagation path categories with too few members. These categories typically represent atypical or accidental propagation patterns with low universality and reference value. The preset cluster number threshold is set based on the requirement for the significance of the path pattern, combined with the total amount and distribution characteristics of historical data; for example, it can be set to 5% of the total number of historical transaction sets. By deleting categories where the number of initial propagation paths within a cluster does not reach the threshold, the remaining set of initial propagation paths is the final statistically obtained propagation path. This processing method ensures that the final obtained propagation paths have significant statistical meaning and representativeness, accurately reflecting the most likely propagation pattern of thermal runaway under the current energy storage power station layout.

[0052] S30: Load the predefined high-frequency operating parameters of the energy storage battery distribution location, perform runaway environment temperature optimization, and obtain the environment temperature threshold;

[0053] Loading predefined high-frequency operating parameters for the distribution location of energy storage batteries is a process of calling representative operating condition data for a specific location from a pre-stored dataset. This process yields a set of parameters that characterize the most common or typical operating states of the energy storage battery distribution location, which can be used for subsequent precise analysis of the critical environmental conditions that trigger thermal runaway at the energy storage battery distribution location under different operating modes.

[0054] The predefined process for high-frequency operation parameters includes:

[0055] Load the historical operating parameter set of the energy storage batteries at the distribution locations of the energy storage batteries;

[0056] Cluster analysis is performed on the historical operating parameter set of the energy storage battery to obtain multiple clusters of historical operating parameters;

[0057] Traverse the multi-cluster historical operation parameters, select the centroid operation parameters, and add them to the high-frequency operation parameters.

[0058] First, load the historical operation parameter set of the energy storage battery corresponding to the location of the energy storage battery distribution. This historical operation parameter set contains a variety of operation parameters recorded during the historical operation of the energy storage battery distribution location.

[0059] Secondly, cluster analysis was performed on the historical operating parameter set of the energy storage battery. Specifically, the cluster analysis process automatically categorized operating parameters with similar patterns in the historical data based on the feature similarity between operating parameters, forming multiple sets of historical operating parameters with similar characteristics, where each cluster represents a typical operating mode.

[0060] The implementation approach of this clustering analysis is as follows: First, a preset deviation threshold for operational parameter attributes is set as the judgment criterion. This deviation threshold is set based on the characteristics of the energy storage battery's operating parameters and engineering experience, such as 10% to 20% of the normal fluctuation range of each parameter. The deviation values ​​of each corresponding attribute between any two sets of operational parameters are calculated. Further, the proportion of attributes with deviation values ​​less than or equal to the deviation threshold is statistically analyzed, and this proportion is defined as the similarity between the two sets of operational parameters. Finally, the similarity is compared with the preset similarity threshold to determine the affiliation relationship between the two sets of parameters: if the similarity is less than the similarity threshold, the two sets of operational parameters are determined to belong to the same cluster; otherwise, they are classified into different clusters. The similarity threshold is set based on the requirements for clustering accuracy and the distribution characteristics of historical data, such as 70%.

[0061] Finally, the historical operating parameters of the multi-cluster energy storage batteries obtained through clustering are traversed. For each cluster of historical operating parameters, its centroid operating parameter is determined using the following method: First, any set of operating parameters within the cluster is extracted, and its similarity to all other operating parameters within the cluster is calculated, forming a set of similarity values; then, the arithmetic mean of this similarity set is calculated, and this mean is defined as the centroid value of the operating parameter set; the above calculation process is repeated for all operating parameter sets within the cluster to obtain the centroid value of each parameter set; finally, the operating parameter with the largest centroid value is selected and determined as the centroid operating parameter of the cluster. The centroid operating parameters selected by this method for all clusters are collected to form a set of high-frequency operating parameters for subsequent temperature optimization analysis. This screening mechanism ensures that the obtained operating parameters have the highest representativeness within their respective clusters and can effectively cover the main historical operating condition characteristics of that location.

[0062] Further, predefined high-frequency operating parameters based on the distribution location of the energy storage batteries are loaded, and runaway environment temperature optimization is performed to obtain an environmental temperature threshold, including:

[0063] Extract the first set of high-frequency operation parameters from the high-frequency operation parameters;

[0064] Based on the distribution location of the energy storage batteries, historical monitoring values ​​of ambient temperature are loaded, and box-type diagram analysis is performed to obtain the ambient temperature of the box.

[0065] According to the preset temperature deviation, the minimum value of the ambient temperature of the chamber is extracted, and the temperature is divided into preset temperature values ​​from small to large to obtain the test temperature sequence.

[0066] Based on the test temperature sequence, runaway environment temperature optimization is performed for the first set of high-frequency operating parameters to obtain the first environment temperature threshold.

[0067] The first set of high-frequency operating parameters are associated with and stored with the first ambient temperature threshold, and then added to the ambient temperature threshold.

[0068] First, the first set of high-frequency operating parameters is extracted from the obtained parameters and used as the current analysis object. This set of parameters represents a typical operating mode of the energy storage battery at a specific distribution location. Based on the distribution location of the energy storage battery, the corresponding historical environmental temperature monitoring values ​​are loaded. These historical environmental temperature monitoring values ​​are continuously collected by temperature sensors deployed at the energy storage battery distribution location during historical operating cycles, representing the actual environmental temperature conditions at that location over a long period. Box plot analysis is performed on these historical environmental temperature monitoring values. By calculating the upper quartile, lower quartile, and interquartile range, the normal temperature distribution range can be determined, i.e., the temperature interval containing 50% of the core data after excluding outliers, thus obtaining the box ambient temperature characterizing the main temperature range.

[0069] Secondly, based on the preset temperature deviation value, the minimum value within the ambient temperature range of the enclosure is extracted as the starting point. Following the preset temperature gradient, the temperature is then divided upwards at equal intervals until the preset upper temperature limit is reached, thus generating an ordered test temperature sequence. The preset temperature deviation value is set according to the accuracy of ambient temperature monitoring and engineering safety margin requirements, such as 2℃; the preset temperature gradient is set according to the accuracy requirements of temperature optimization and computational efficiency, such as 1℃; and the preset upper temperature limit is set according to the safety operation standards of energy storage batteries and the temperature resistance characteristics of the materials, such as 60℃.

[0070] Based on this test temperature sequence, an optimization calculation for the runaway environment temperature is performed for the first set of high-frequency operating parameters. This optimization process iteratively tests the risk of thermal runaway under different temperature conditions, and finally determines a critical temperature value as the first ambient temperature threshold.

[0071] Specifically, based on the test temperature sequence, runaway environment temperature optimization is performed for the first set of high-frequency operating parameters to obtain a first ambient temperature threshold, including:

[0072] Extract the k-th test temperature from the test temperature sequence, where k is an integer, the initial value of k is equal to 1, and the length of the test temperature sequence is ≥ k ≥ 1;

[0073] Based on the distribution location of the energy storage batteries, the model of the energy storage batteries is extracted. Based on the model of the energy storage batteries, the operating temperature prediction model and the predefined thermal runaway temperature threshold are activated. The operating temperature prediction model is generated by machine learning training, with environmental temperature recording data and operating record parameters as constraints and operating monitoring temperature as output supervision.

[0074] The kth test temperature and the first group of high-frequency operation parameters are input into the operation temperature prediction model to obtain the operation temperature prediction value.

[0075] When the predicted operating temperature is greater than or equal to the thermal runaway temperature threshold, the kth test temperature is set as the first ambient temperature threshold.

[0076] otherwise:

[0077] When k equals the length of the test temperature sequence, the first ambient temperature threshold is configured to be empty;

[0078] When k is less than the length of the test temperature sequence, k is incremented by one, and the first ambient temperature threshold configuration loop is executed.

[0079] First, test temperatures are extracted sequentially from the test temperature sequence and identified by the sequence number k, with an initial value of 1. The value of k covers the entire test temperature sequence, and the length of the test temperature sequence is ≥ k ≥ 1. Second, based on the distribution location of the energy storage batteries currently being analyzed, the corresponding energy storage battery model is extracted. Based on this model information, two key components are activated: one is an operating temperature prediction model specifically developed for this battery model. This model uses historical ambient temperature records and operating record parameters as input features and the actual monitored operating temperature as the training target, and is trained using machine learning methods; the other is a predefined thermal runaway temperature threshold for this battery model, which is the critical value at which the internal temperature of the battery may trigger thermal runaway.

[0080] For example, since there is a highly nonlinear and complex correlation between ambient temperature, operating parameters and predicted battery operating temperature, and neural network models have significant advantages in multi-level feature abstraction and complex pattern recognition, a neural network model is chosen to construct the operating temperature prediction model.

[0081] Specifically, the operational temperature prediction model mainly consists of an input layer, a feature abstraction layer, and a temperature prediction output layer. The input layer receives a standardized feature vector, which contains input features from multiple dimensions, including environmental temperature records and operational parameters. The feature abstraction layer employs a multi-layer fully connected neural network structure. The number of hidden layer neurons is adaptively configured according to the dimensions of the input features. Each neural network layer uses the ReLU activation function to introduce non-linear transformation capabilities, and Dropout layers are embedded between network layers with a dropout rate set between 0.2 and 0.4 to effectively suppress model overfitting and improve its generalization performance. The output layer uses a linear activation function to map the final abstract features to a continuous numerical value, which serves as the predicted operational temperature.

[0082] During training, key hyperparameters included a learning rate of 0.001, 150 training epochs, and a batch size of 64. The learning rate was set to balance training stability and convergence speed; the number of training epochs ensured the model fully learned the thermal behavior patterns in the data; and the batch size balanced training efficiency with memory resource consumption. Specifically, a supervised learning approach was used. Feature vectors were collected from historical operational data of the energy storage battery as the input sample set, and corresponding actual operational monitoring temperatures were simultaneously acquired to form a sample label set. The input sample set and the corresponding label sample set were divided into training, validation, and test sets in a 7:2:1 ratio.

[0083] Furthermore, the feature vectors of the samples in the training set are used as input, and the corresponding actual monitored operating temperature is used as the supervision signal. The network weight parameters are iteratively optimized using the backpropagation algorithm and the Adam optimizer. The mean squared error loss function is used to measure the deviation between the predicted operating temperature and the actual monitored temperature. The training process is monitored using a validation set. When the value of the validation set loss function no longer decreases for several consecutive rounds and the model prediction error reaches a predetermined threshold, such as a mean absolute error below 2°C, training is terminated, resulting in a converged operating temperature prediction model. This operating temperature prediction model can effectively capture the complex nonlinear relationship between environmental parameters, operating parameters, and battery temperature, achieving accurate operating temperature prediction.

[0084] Then, the currently selected k-th test temperature and the first set of high-frequency operating parameters are input into the operating temperature prediction model. This operating temperature prediction model calculates and outputs the predicted operating temperature value. This predicted operating temperature value is compared with the thermal runaway temperature threshold: if the predicted operating temperature value reaches or exceeds the thermal runaway temperature threshold, then the currently tested k-th test temperature is determined to be the sought critical ambient temperature, and it is set as the first ambient temperature threshold, thus completing the optimization process.

[0085] If the predicted operating temperature does not reach the thermal runaway temperature threshold, it is necessary to determine whether the test is complete: If k has reached the total length of the test temperature sequence, it indicates that none of the preset test temperatures have triggered the risk of thermal runaway. At this time, the first ambient temperature threshold is set to null, which means that no clear critical point has been found under the current parameter combination. If k is less than the sequence length, the value of k is increased by 1, the next test temperature is selected, and the above prediction and comparison steps are repeated to form a loop iteration until the critical temperature is found or all test temperatures have been traversed.

[0086] Finally, the first set of high-frequency operation parameters are associated with the obtained first ambient temperature threshold, stored together, and this association is added to the ambient temperature threshold set to form a complete parameter-temperature mapping relationship.

[0087] S40: Associate and store the ambient temperature threshold, the propagation path, and the distribution location of the energy storage battery, and add them to the thermal runaway trend mapping table;

[0088] Specifically, the determined ambient temperature threshold, the statistically derived propagation path, and the corresponding distribution location of the energy storage batteries are correlated and added as a complete data record to the thermal runaway trend mapping table.

[0089] This thermal runaway trend mapping table is a structured dataset. Each record clearly describes the propagation path a thermal runaway event might follow when the ambient temperature reaches a specific threshold at a given energy storage battery location. This associative storage method integrates previously independent ambient temperature conditions, spatial location information, and propagation path sequences into a cohesive whole, forming a queryable and analyzable knowledge base. This provides direct and efficient data support and judgment criteria for the final execution of thermal runaway trend detection.

[0090] S50: Perform thermal runaway trend sensing based on the thermal runaway trend mapping table.

[0091] Specifically, by matching real-time monitoring data from energy storage power stations with pre-stored knowledge in a mapping table, dynamic assessment and early warning of thermal runaway risks are achieved. When monitoring the location of a specific energy storage battery, the first step is to acquire the real-time ambient temperature data for that location. This real-time ambient temperature is then compared with the ambient temperature threshold associated with that location in the thermal runaway trend mapping table. If the real-time temperature reaches or exceeds the threshold, it indicates that thermal runaway risk conditions have been triggered.

[0092] Simultaneously, one or more propagation paths associated with the location of the energy storage battery and the current triggering conditions are retrieved from the thermal runaway trend mapping table. These propagation paths directly indicate the possible spatial propagation direction and device sequence of the thermal runaway event from its starting location, thus completing the perception of the thermal runaway trend.

[0093] Finally, based on the above matching and prediction results, a comprehensive perception conclusion including the risk status and potential spread path is output, providing decision support for taking targeted prevention and intervention measures, thereby achieving proactive and accurate perception of thermal runaway trends.

[0094] In summary, the embodiments of this application have at least the following technical effects:

[0095] Compared to existing technologies, this invention enables precise perception and early warning of thermal runaway trends. First, by comprehensively analyzing the physical layout characteristics of energy storage facilities, potential spatial propagation paths of thermal runaway can be accurately identified, thereby predicting the propagation trend of the fault. Furthermore, by combining the historical operating characteristics of battery cells, the critical ambient temperature conditions that trigger thermal runaway at different locations under specific operating modes can be determined, significantly improving the accuracy of the warning threshold. Further, by constructing a trend mapping table that integrates location information, propagation paths, and temperature thresholds, effective integration and rapid retrieval of multi-dimensional safety information are achieved. Finally, based on this comprehensive mapping relationship, trend judgment can be made to issue timely warnings in the early stages of thermal runaway, buying valuable time for safety measures and effectively improving the overall safety level of energy storage power stations.

[0096] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0097] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0098] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A method for sensing thermal runaway trends in energy storage power stations, characterized in that, include: The energy storage facility distribution topology of the loaded energy storage power station includes the distribution locations of energy storage batteries; Using the location of the energy storage battery distribution as the thermal runaway starting point and the topology of the energy storage facility distribution as the scenario constraint, the thermal runaway event set is retrieved and the propagation path is statistically analyzed. Load the predefined high-frequency operating parameters of the energy storage battery distribution location, perform runaway environment temperature optimization, and obtain the environment temperature threshold; The ambient temperature threshold, the propagation path, and the distribution location of the energy storage battery are associated and stored, and added to the thermal runaway trend mapping table; Based on the thermal runaway trend mapping table, thermal runaway trend sensing is performed.

2. The method as described in claim 1, characterized in that, Using the location of the energy storage batteries as the thermal runaway starting point and the topology of the energy storage facility distribution as the scenario constraint, a thermal runaway event set is retrieved, including: Load the thermal runaway transaction to be analyzed, wherein the thermal runaway transaction to be analyzed includes the thermal runaway record origin and the energy storage facility distribution record topology; When the energy storage facility distribution similarity between the energy storage facility distribution topology and the energy storage facility distribution record topology is greater than or equal to the energy storage facility distribution similarity threshold, and the thermal runaway initiation point is the same as the thermal runaway record initiation point, the thermal runaway transaction to be analyzed is added to the thermal runaway transaction set.

3. The method as described in claim 1, characterized in that, Statistical spread paths include: From the thermal runaway event set, extract the first propagation path and the distribution record topology of the first energy storage facility, wherein the first propagation path includes the tag number sequence of the first propagation device; Extract the first spread device tag number sequence from the first spread location sequence of the first energy storage facility distribution record topology; Based on the first spread location sequence, the second spread device tag sequence is extracted from the energy storage facility distribution topology and added to the initial spread path set; Based on the initial set of propagation paths, the propagation paths are statistically analyzed.

4. The method as described in claim 3, characterized in that, Based on the initial set of propagation paths, the propagation paths are statistically analyzed, including: Cluster analysis is performed on the initial spread path set to obtain multiple clusters of initial spread paths; Based on the initial spread paths of the multiple clusters, delete the initial spread paths within the cluster whose number is less than or equal to the number threshold within the cluster to obtain the spread paths.

5. The method as described in claim 4, characterized in that, Cluster analysis is performed on the initial spread path set to obtain multiple clusters of initial spread paths, including: Extract the first initial spread path and the second initial spread path from the initial spread path set; Align the starting points of the first initial spread path and the second initial spread path, and count the number of identical position numbers; Extract the total number of longer path indices from the first initial spread path and the second initial spread path; If the ratio of the number of corresponding serial numbers to the total number of serial numbers in the longer path is greater than or equal to the ratio threshold, the first initial spread path and the second initial spread path are added to the same cluster; otherwise, they are set to different clusters.

6. The method as described in claim 1, characterized in that, The predefined process for high-frequency operation parameters includes: Load the historical operating parameter set of the energy storage batteries at the distribution locations of the energy storage batteries; Cluster analysis is performed on the historical operating parameter set of the energy storage battery to obtain multiple clusters of historical operating parameters; Traverse the multi-cluster historical operation parameters, select the centroid operation parameters, and add them to the high-frequency operation parameters.

7. The method as described in claim 1, characterized in that, Load predefined high-frequency operating parameters for the distribution locations of the energy storage batteries, perform runaway environment temperature optimization, and obtain the environmental temperature threshold, including: Extract the first set of high-frequency operation parameters from the high-frequency operation parameters; Based on the distribution location of the energy storage batteries, historical monitoring values ​​of ambient temperature are loaded, and box-type diagram analysis is performed to obtain the ambient temperature of the box. According to the preset temperature deviation, the minimum value of the ambient temperature of the chamber is extracted, and the temperature is divided into preset temperature values ​​from small to large to obtain the test temperature sequence. Based on the test temperature sequence, runaway environment temperature optimization is performed for the first set of high-frequency operating parameters to obtain the first environment temperature threshold. The first set of high-frequency operating parameters are associated with and stored with the first ambient temperature threshold, and then added to the ambient temperature threshold.

8. The method as described in claim 7, characterized in that, Based on the test temperature sequence, runaway environment temperature optimization is performed for the first set of high-frequency operating parameters to obtain a first ambient temperature threshold, including: Extract the k-th test temperature from the test temperature sequence, where k is an integer, the initial value of k is equal to 1, and the length of the test temperature sequence is ≥ k ≥ 1; Based on the distribution location of the energy storage batteries, the model of the energy storage batteries is extracted. Based on the model of the energy storage batteries, the operating temperature prediction model and the predefined thermal runaway temperature threshold are activated. The operating temperature prediction model is generated by training with machine learning, using environmental temperature recording data and operating record parameters as constraints and operating monitoring temperature as output supervision. The kth test temperature and the first group of high-frequency operation parameters are input into the operation temperature prediction model to obtain the operation temperature prediction value. When the predicted operating temperature is greater than or equal to the thermal runaway temperature threshold, the kth test temperature is set as the first ambient temperature threshold. otherwise: When k equals the length of the test temperature sequence, the first ambient temperature threshold is configured to be empty; When k is less than the length of the test temperature sequence, k is incremented by one, and the first ambient temperature threshold configuration loop is executed.

Images