Energy storage power station fire monitoring method and system based on multi-modal data

Abstract

This invention discloses a method and system for fire monitoring in energy storage power stations based on multimodal data. The method includes: acquiring power station sensor data and module sensor data of any energy storage module within the energy storage power station; predicting power station fire information based on the power station sensor data using a fire risk identification model; predicting corresponding module fire information based on the module sensor data using a risk identification model corresponding to the energy storage module; and determining the probability of a fire in the energy storage power station based on a matching calculation between the power station fire information and the module fire information. Therefore, this invention can achieve accurate fire risk assessment based on multi-scale modal data fusion and matching verification, improving the accuracy and reliability of fire early warning for energy storage power stations and reducing the risk of missed fire reports due to local anomalies not being correlated with the overall situation.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and system for fire monitoring in energy storage power stations based on multimodal data. Background Technology

[0002] With the rapid growth of large-scale and high-density deployment of energy storage power stations, power companies and operators are increasingly focusing on improving power station operational safety through accurate fire early warning. Among these challenges, accurately and efficiently monitoring the fire risk of energy storage power stations has become a key technical issue. Existing technologies typically collect sensor data from individual energy storage modules within the power station, using fixed threshold analysis or a single fire identification model to predict fire risk, and then issue early warnings based on standard rules to ensure power station safety. However, existing solutions lack the ability to fuse and predict multi-scale modal data at both the power station and module levels, as well as the matching and verification of the results. This makes it difficult to accurately assess the probability of a real fire, and it fails to effectively correlate local anomalies with overall risk, resulting in insufficient accuracy in fire early warning. Furthermore, it is prone to missed warnings due to local anomalies not being fully recognized, thus limiting the reliability and response efficiency of fire prevention and control in energy storage power stations. Therefore, existing technologies have shortcomings that urgently need to be addressed. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a fire monitoring method and system for energy storage power stations based on multimodal data, which can realize accurate fire risk assessment based on multi-scale modal data fusion and matching verification, improve the accuracy and reliability of fire early warning for energy storage power stations, and reduce the risk of missed fire reports due to local anomalies not being associated with the whole.

[0004] To address the aforementioned technical problems, the first aspect of this invention discloses a fire monitoring method for energy storage power stations based on multimodal data, the method comprising:

[0005] Acquire the power station sensor data and the module sensor data of any one of the energy storage modules in the energy storage power station;

[0006] Based on the fire risk identification model, the fire information of the energy storage power station is predicted according to the power station sensor data.

[0007] Based on the risk identification model corresponding to the energy storage module, and according to the sensor data of the module, the corresponding module fire information is predicted;

[0008] The probability of a fire at the energy storage power station is determined by matching the fire information of the power station and the fire information of the module.

[0009] As an optional implementation, in the first aspect of the present invention, the energy storage module is an energy storage battery area, an energy storage control area, an energy storage mechanical structure area, or an energy storage maintenance channel area in the energy storage power station.

[0010] As an optional implementation, in the first aspect of the present invention, the power station sensing data is obtained by sensors installed in the energy storage power station and is used to characterize the sensing characteristics of the entire energy storage power station; the module sensing data is obtained by sensors installed in the energy storage module and is used to characterize the sensing characteristics within the energy storage module; the sensors include at least one of a temperature sensor, a humidity sensor, a photoelectric reflection sensor, an image sensor, and a sound sensor.

[0011] As an optional implementation, in the first aspect of the present invention, the step of predicting the power station fire information of the energy storage power station based on the power station sensor data according to the fire risk identification model includes:

[0012] All the power station sensor data, corresponding sensor locations, and the three-dimensional structural model of the energy storage power station are input into a trained fire risk identification model to obtain the power station fire information of the energy storage power station. The power station fire information includes the fire occurrence probability and fire cause corresponding to each model location in the three-dimensional structural model. The fire risk identification model is trained using a training dataset that includes multiple training power station sensor data and corresponding sensor location labels, three-dimensional structural model location labels, and fire information labels.

[0013] As an optional implementation, in the first aspect of the present invention, the step of predicting corresponding module fire information based on the risk identification model corresponding to the energy storage module and according to the module sensor data includes:

[0014] Determine the module parameters corresponding to the energy storage module;

[0015] Based on the module parameters, and using a similarity algorithm, the target risk identification model is determined from multiple candidate risk identification models.

[0016] The module sensing data is input into the target risk identification model to obtain the module fire information corresponding to the energy storage module; the module fire information includes the fire occurrence probability and fire cause of the energy storage module.

[0017] As an optional implementation, in the first aspect of the present invention, determining the probability of a fire in the energy storage power station based on the matching calculation between the power station fire information and the module fire information includes:

[0018] From all the energy storage modules of the energy storage power station, modules whose fire hazard level is greater than a preset threshold are selected to obtain multiple hazardous energy storage modules;

[0019] Calculate the matching degree between the fire information of all the hazardous energy storage modules and the fire information of the power plant;

[0020] Calculate the average fire occurrence probability of all the hazardous energy storage modules to obtain the probability parameter;

[0021] The matching degree and the probability parameters are input into the trained fire occurrence probability prediction model to obtain the possible fire probability of the energy storage power station; the fire occurrence probability prediction model is trained by a training dataset that includes multiple training matching degrees, corresponding training probability parameter labels, and corresponding fire probability labels.

[0022] As an optional implementation, in the first aspect of the invention, the degree of danger of the module fire information is calculated through the following steps:

[0023] For each of the energy storage modules, calculate the proportion of the fire cause of that energy storage module among all the fire causes of all the energy storage modules;

[0024] Among all the other energy storage modules, identify the module with the same cause of fire as the energy storage module to obtain multiple associated modules;

[0025] Calculate the weighted summation average of the fire occurrence probabilities of the energy storage module and all the associated modules to obtain the probability average parameter; wherein, the calculation weight of the fire occurrence probability of the associated module is less than the calculation weight of the energy storage module, and the calculation weight of the fire occurrence probability of the associated module is inversely proportional to the module location distance between the associated module and the energy storage module.

[0026] The product of the probability average parameter and the proportional weight is calculated to obtain the degree of danger of the fire information of the energy storage module; the proportional weight is proportional to the occurrence ratio.

[0027] As an optional implementation, in the first aspect of the invention, calculating the matching degree between the module fire information of all the hazardous energy storage modules and the power plant fire information includes:

[0028] The causes of fires in all the aforementioned hazardous energy storage modules are combined into a first set of causes;

[0029] The fire causes corresponding to all model locations in the power plant fire information are combined into a second set of causes;

[0030] The fire occurrence probability of all the hazardous energy storage modules and their corresponding module locations are combined into a first probability set;

[0031] The model locations and corresponding fire occurrence probabilities in the power plant fire information are combined into a second probability set;

[0032] Calculate the first similarity between the first set of causes and the second set of causes;

[0033] Calculate the second similarity between the first probability set and the second probability set;

[0034] The product of the first similarity and the second similarity is calculated to obtain the matching degree between the module fire information of all the hazardous energy storage modules and the power plant fire information.

[0035] A second aspect of this invention discloses a fire monitoring system for energy storage power stations based on multimodal data, the system comprising:

[0036] The acquisition module is used to acquire the power station sensing data of the energy storage power station and the module sensing data of any one of the energy storage modules.

[0037] The first prediction module is used to predict the fire information of the energy storage power station based on the fire risk identification model and the power station sensor data.

[0038] The second prediction module is used to predict the corresponding fire information of the energy storage module based on the risk identification model corresponding to the energy storage module and the sensor data of the module.

[0039] The matching module is used to determine the probability of a fire at the energy storage power station based on the matching calculation between the power station fire information and the module fire information.

[0040] As an optional implementation, in a second aspect of the present invention, the energy storage module is an energy storage battery area, an energy storage control area, an energy storage mechanical structure area, or an energy storage maintenance channel area in the energy storage power station.

[0041] As an optional implementation, in a second aspect of the invention, the power station sensing data is obtained by sensors installed in the energy storage power station and is used to characterize the sensing characteristics of the entire energy storage power station; the module sensing data is obtained by sensors installed in the energy storage module and is used to characterize the sensing characteristics within the energy storage module; the sensors include at least one of a temperature sensor, a humidity sensor, a photoelectric reflection sensor, an image sensor, and a sound sensor.

[0042] As an optional implementation, in a second aspect of the invention, the specific method by which the first prediction module predicts the power station fire information of the energy storage power station based on the power station sensor data, according to the fire risk identification model, includes:

[0043] All the power station sensor data, corresponding sensor locations, and the three-dimensional structural model of the energy storage power station are input into a trained fire risk identification model to obtain the power station fire information of the energy storage power station. The power station fire information includes the fire occurrence probability and fire cause corresponding to each model location in the three-dimensional structural model. The fire risk identification model is trained using a training dataset that includes multiple training power station sensor data and corresponding sensor location labels, three-dimensional structural model location labels, and fire information labels.

[0044] As an optional implementation, in a second aspect of the invention, the second prediction module predicts the corresponding module fire information based on the risk identification model corresponding to the energy storage module and according to the module's sensor data, in the following specific manner:

[0045] Determine the module parameters corresponding to the energy storage module;

[0046] Based on the module parameters, and using a similarity algorithm, the target risk identification model is determined from multiple candidate risk identification models.

[0047] The module sensing data is input into the target risk identification model to obtain the module fire information corresponding to the energy storage module; the module fire information includes the fire occurrence probability and fire cause of the energy storage module.

[0048] As an optional implementation, in a second aspect of the invention, the specific method by which the matching module determines the probability of a fire at the energy storage power station based on the matching calculation between the power station fire information and the module fire information includes:

[0049] From all the energy storage modules of the energy storage power station, modules whose fire hazard level is greater than a preset threshold are selected to obtain multiple hazardous energy storage modules;

[0050] Calculate the matching degree between the fire information of all the hazardous energy storage modules and the fire information of the power plant;

[0051] Calculate the average fire occurrence probability of all the hazardous energy storage modules to obtain the probability parameter;

[0052] The matching degree and the probability parameters are input into the trained fire occurrence probability prediction model to obtain the possible fire probability of the energy storage power station; the fire occurrence probability prediction model is trained by a training dataset that includes multiple training matching degrees, corresponding training probability parameter labels, and corresponding fire probability labels.

[0053] As an optional implementation, in a second aspect of the invention, the degree of danger of the module fire information is calculated through the following steps:

[0054] For each of the energy storage modules, calculate the proportion of the fire cause of that energy storage module among all the fire causes of all the energy storage modules;

[0055] Among all the other energy storage modules, identify the module with the same cause of fire as the energy storage module to obtain multiple associated modules;

[0056] Calculate the weighted summation average of the fire occurrence probabilities of the energy storage module and all the associated modules to obtain the probability average parameter; wherein, the calculation weight of the fire occurrence probability of the associated module is less than the calculation weight of the energy storage module, and the calculation weight of the fire occurrence probability of the associated module is inversely proportional to the module location distance between the associated module and the energy storage module.

[0057] The product of the probability average parameter and the proportional weight is calculated to obtain the degree of danger of the fire information of the energy storage module; the proportional weight is proportional to the occurrence ratio.

[0058] As an optional implementation, in a second aspect of the invention, the specific method by which the matching module calculates the matching degree between the module fire information of all the hazardous energy storage modules and the power plant fire information includes:

[0059] The causes of fires in all the aforementioned hazardous energy storage modules are combined into a first set of causes;

[0060] The fire causes corresponding to all model locations in the power plant fire information are combined into a second set of causes;

[0061] The fire occurrence probability of all the hazardous energy storage modules and their corresponding module locations are combined into a first probability set;

[0062] The model locations and corresponding fire occurrence probabilities in the power plant fire information are combined into a second probability set;

[0063] Calculate the first similarity between the first set of causes and the second set of causes;

[0064] Calculate the second similarity between the first probability set and the second probability set;

[0065] The product of the first similarity and the second similarity is calculated to obtain the matching degree between the module fire information of all the hazardous energy storage modules and the power plant fire information.

[0066] A third aspect of this invention discloses another fire monitoring system for energy storage power stations based on multimodal data, the system comprising:

[0067] Memory containing executable program code;

[0068] A processor coupled to the memory;

[0069] The processor calls the executable program code stored in the memory to execute some or all of the steps in the fire monitoring method for energy storage power stations based on multimodal data disclosed in the first aspect of the present invention.

[0070] The fourth aspect of the present invention discloses a computer storage medium storing computer instructions, which, when invoked, are used to execute some or all of the steps in the fire monitoring method for energy storage power stations based on multimodal data disclosed in the first aspect of the present invention.

[0071] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0072] This invention acquires sensor data from the energy storage power station and sensor data from any energy storage module, and predicts fire information for both the power station and the module based on a fire risk identification model. By matching the two data, the probability of a fire in the energy storage power station is determined. This enables accurate fire risk assessment based on multi-scale modal data fusion and matching verification, improving the accuracy and reliability of fire early warning for energy storage power stations and reducing the risk of missed fire reports due to local anomalies not being correlated with the overall situation. Attached Figure Description

[0073] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0074] Figure 1 This is a flowchart illustrating a fire monitoring method for energy storage power stations based on multimodal data, as disclosed in an embodiment of the present invention.

[0075] Figure 2 This is a schematic diagram of the structure of a fire monitoring system for an energy storage power station based on multimodal data, as disclosed in an embodiment of the present invention.

[0076] Figure 3 This is a schematic diagram of another fire monitoring system for energy storage power stations based on multimodal data disclosed in an embodiment of the present invention. Detailed Implementation

[0077] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. 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.

[0078] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0079] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0080] This invention discloses a fire monitoring method and system for energy storage power stations based on multimodal data. By acquiring the power station's sensor data and the sensor data of any energy storage module, it predicts power station fire information and module fire information based on a fire risk identification model, respectively. The probability of a fire in the energy storage power station is determined by matching the two data points. This enables accurate fire risk assessment based on multi-scale modal data fusion and matching verification, improving the accuracy and reliability of fire early warning for energy storage power stations and reducing the risk of missed fire reports due to local anomalies not being correlated with the overall situation. Detailed explanations follow.

[0081] Example 1: Please refer to Figure 1 , Figure 1 This is a flowchart illustrating a fire monitoring method for energy storage power stations based on multimodal data, as disclosed in an embodiment of the present invention. Figure 1 The described fire monitoring method for energy storage power stations based on multimodal data can be applied to data processing systems / data processing equipment / data processing servers (wherein, the server includes a local processing server or a cloud processing server). For example... Figure 1 As shown, the fire monitoring method for energy storage power stations based on multimodal data may include the following operations:

[0082] 101. Obtain the power station sensor data and the module sensor data of any one of the energy storage modules in the energy storage power station.

[0083] Optionally, the power plant sensing data may include ambient temperature, smoke concentration, VOC concentration, hydrogen concentration, CO concentration, overall vibration, power plant-level current and voltage, or video monitoring data; this invention does not impose any limitations.

[0084] Optionally, the sensing data of this module may include individual battery voltage, temperature, pressure, internal resistance, gas sensor data, thermal runaway warning signals, or module-level smoke detection data; this invention does not impose any limitations.

[0085] Optionally, this acquisition process can be implemented based on real-time data collection from a power plant SCADA system, BMS system, fire protection IoT platform, or edge gateway; this invention does not limit this process.

[0086] 102. Based on the fire risk identification model, predict the fire information of the energy storage power station according to the power station sensor data.

[0087] Optionally, the fire risk identification model can be a 3D-CNN model, a Graph Neural Network model, a Transformer spatial-temporal model, or a multimodal fusion model; this invention does not impose any limitations.

[0088] Optionally, the fire information of the power plant may include the global fire probability, the regional thermal runaway probability, the fire spread path prediction, or the ranking of key disaster-causing factors, which is not limited in this invention.

[0089] 103. Based on the risk identification model corresponding to the energy storage module, predict the corresponding module fire information according to the module sensor data.

[0090] Optionally, the fire information of this module may include the probability of thermal runaway of a single module, the location of the ignition point, the classification of the cause of the fire (such as overcharging, internal short circuit, mechanical damage) or the warning level, which is not limited in this invention.

[0091] Optionally, the risk identification model can be a module-specific model, a battery type-specific model, or an aging state adaptive model; this invention does not impose any limitations.

[0092] 104. Based on the matching calculation between the power station fire information and the module fire information, determine the probability of fire in the energy storage power station.

[0093] Optionally, the process of determining the probability of the fire can be based on multi-scale information fusion, consistency verification, or Bayesian probability integration, and this invention does not limit it.

[0094] Optionally, the probability result may trigger fire alarm linkage, power load reduction, emergency shutdown, or personnel evacuation; this invention does not limit this.

[0095] As can be seen, the above-described embodiments of the invention acquire power station sensor data and module sensor data of any energy storage module, predict power station fire information and module fire information based on a fire risk identification model, and determine the probability of fire in the energy storage power station by matching the two. This enables accurate fire risk assessment based on multi-scale modal data fusion and matching verification, improves the accuracy and reliability of fire early warning for energy storage power stations, and reduces the risk of missed fire reports due to local anomalies not being associated with the whole.

[0096] As an optional embodiment, in the above steps, the energy storage module is the energy storage battery area, energy storage control area, energy storage mechanical structure area, or energy storage maintenance channel area in the energy storage power station.

[0097] As can be seen, the above optional embodiments limit the types of energy storage modules, so that the technical solution of the present invention can be applied to different energy storage power station areas, helping to improve the accuracy and reliability of fire early warning in energy storage power stations and reduce the risk of fire underreporting caused by local anomalies not being associated with the whole.

[0098] As an optional embodiment, in the above steps, the power station sensing data is obtained by sensors installed in the energy storage power station and is used to characterize the sensing characteristics of the entire energy storage power station; the module sensing data is obtained by sensors installed in the energy storage module and is used to characterize the sensing characteristics within the energy storage module; the sensors include at least one of temperature sensors, humidity sensors, photoelectric reflection sensors, image sensors, and sound sensors.

[0099] As can be seen, the above optional embodiments define the content and differences between power station sensing data and module sensing data, as well as the corresponding sensor types, so that multimodal sensing data can effectively characterize the real-time operating characteristics of the power station as a whole or individual modules, assist in realizing accurate fire risk assessment based on multi-scale modal data fusion and matching verification, improve the accuracy and reliability of fire early warning for energy storage power stations, and reduce the risk of missed fire reports due to local anomalies not being associated with the whole.

[0100] As an optional embodiment, the step above, predicting power station fire information based on the fire risk identification model and power station sensor data, includes:

[0101] All power station sensor data, corresponding sensor locations, and the three-dimensional structural model of the energy storage power station are input into a trained fire risk identification model to obtain power station fire information of the energy storage power station.

[0102] Optionally, the power plant fire information includes the probability of fire occurrence and the cause of fire at each model location in the three-dimensional structural model.

[0103] Optionally, the fire risk identification model is trained using a training dataset that includes multiple training power plant sensor data and corresponding sensor location labels, 3D structural model location labels, and fire information labels.

[0104] Optionally, the three-dimensional structural model can be a BIM model, a point cloud model, or a gridded spatial model; this invention does not impose any limitations.

[0105] Optionally, the probability of fire occurrence can be at the voxel level, cabin level, or equipment level resolution; this invention does not limit this.

[0106] Optionally, the cause of the fire may include electrical failure, thermal runaway propagation, cooling failure, or external ignition source; this invention does not limit such causes.

[0107] Optionally, the training dataset can be derived from real fire reproduction experiments, historical accident data, high-fidelity simulations, or enhanced generated data; this invention does not impose any limitations.

[0108] Optionally, the fire risk identification model can be trained using supervised learning, weakly supervised learning, or contrastive learning strategies; this invention does not impose any limitations.

[0109] As can be seen, through the above optional embodiments, by inputting the power plant sensor data, sensor location, and three-dimensional structural model into the trained fire risk identification model, the probability and cause of fire occurrence at each location in the three-dimensional structural model are output. Thus, based on accurate power plant fire information prediction, the level of refinement and visualization of fire risk assessment is improved through spatial location correlation and three-dimensional modeling, providing high-quality global data support for module fire information matching and reducing the risk of fire location deviation caused by missing spatial information.

[0110] As an optional embodiment, the step above, predicting the corresponding module fire information based on the risk identification model corresponding to the energy storage module and according to the module's sensor data, includes:

[0111] Determine the module parameters corresponding to the energy storage module;

[0112] Based on the module parameters and a similarity algorithm, the target risk identification model is determined from multiple candidate risk identification models.

[0113] The module sensor data is input into the target risk identification model to obtain the module fire information corresponding to the energy storage module.

[0114] Optionally, the module fire information includes the probability of fire occurrence and the cause of fire in the energy storage module.

[0115] Optionally, the module parameters may include battery type (such as lithium iron phosphate, ternary lithium), capacity, state of charge (SOC), state of harmonics (SOH), number of cycles, production batch, or installation location. This invention does not impose any limitations on these parameters.

[0116] Optionally, the module parameters can be read from the BMS database, asset management system, or QR code identifier; this invention does not impose any limitations.

[0117] Optionally, the similarity algorithm can be cosine similarity, Mahalanobis distance, embedding space distance, or a hybrid algorithm of rule and distance; this invention does not impose any limitations.

[0118] Optionally, the candidate risk identification model can be pre-trained and stored according to battery chemistry system, capacity range, and aging stage; this invention does not limit this.

[0119] Optionally, the similarity between the module parameters in the training data of each candidate risk identification model and the module parameters of the energy storage module can be calculated, and the model with the highest similarity can be selected as the target risk identification model.

[0120] As can be seen, through the above optional embodiments, by selecting a target risk identification model based on a similarity algorithm according to the energy storage module parameters and inputting module sensor data to predict module fire information, the prediction is more targeted and accurate based on the accurate determination of module fire information, and reliable local risk data is provided for the calculation of the probability of fire, reducing the risk of misjudgment of module fire due to the generality of the model.

[0121] As an optional embodiment, the step above, determining the probability of a fire at the energy storage power station based on the matching calculation between the power station fire information and the module fire information, includes:

[0122] From all energy storage modules in the energy storage power station, modules whose fire hazard level exceeds a preset threshold are selected to obtain multiple hazardous energy storage modules;

[0123] Calculate the matching degree between the module fire information and the power plant fire information of all hazardous energy storage modules;

[0124] Calculate the average fire probability of all hazardous energy storage modules to obtain probability parameters;

[0125] The matching degree and probability parameters are input into the trained fire probability prediction model to obtain the possible fire probability of the energy storage power station. The fire probability prediction model is trained using a training dataset that includes multiple training matching degrees, corresponding training probability parameter labels, and corresponding fire probability labels.

[0126] Optionally, the threshold can be a fixed threshold, an adaptive threshold, or a threshold adjusted based on the size of the power plant; this invention does not impose any limitations.

[0127] Optionally, the hazardous energy storage module can be marked as a high-risk area, a key monitoring target, or trigger local isolation; this invention does not impose any limitations.

[0128] Optionally, the matching degree can be a comprehensive index of causal consistency and spatial probability distribution similarity, which is not limited in this invention.

[0129] Optionally, the average value can be a simple average, a weighted average, or a truncated average; this invention does not limit the type of average.

[0130] Optionally, this probability parameter can be used as the overall threat level of a group of dangerous modules, and this invention does not limit it.

[0131] Optionally, the fire occurrence probability prediction model can be a small neural network, gradient boosting tree, Bayesian network, or a rule + regression hybrid model; this invention does not limit the model.

[0132] Optionally, the fire occurrence probability prediction model achieves deep fusion of multi-scale information and final decision-making, which is not limited in this invention.

[0133] Optionally, the training dataset can be derived from multi-scale data pairs of historical fire events, simulation experiments, or expert-enhanced annotations; this invention does not impose any limitations.

[0134] Optionally, the training of the fire occurrence probability prediction model may employ supervised learning or sequential supervised learning to ensure the reliability of the fusion logic; this invention does not impose any limitations on this.

[0135] Optionally, the fire probability prediction model can be updated online to adapt to new battery types or protective measures, which is not limited in this invention.

[0136] As can be seen, through the above optional embodiments, by screening dangerous energy storage modules whose hazard level exceeds the threshold and calculating the matching degree between their module fire information and power station fire information, and combining the probability parameters input into the fire occurrence probability prediction model to obtain the final possible fire probability, the sensitivity and scientific nature of the assessment are improved by screening dangerous modules and multi-dimensional matching on the basis of accurate fire probability prediction. This provides forward-looking support for fire early warning of energy storage power stations and reduces the risk of delayed early warning caused by the failure to amplify the overall risk due to local high risk.

[0137] As an optional embodiment, the hazard level of the module fire information in the above steps is calculated through the following steps:

[0138] For each energy storage module, calculate the proportion of fire causes occurring in that energy storage module compared to all fire causes in all energy storage modules.

[0139] Among all other energy storage modules, identify the module with the same fire cause as the energy storage module to obtain multiple associated modules;

[0140] Calculate the weighted summation average of the fire occurrence probabilities of the energy storage module and all associated modules to obtain the probability average parameter; optionally, the calculation weight of the fire occurrence probability of the associated modules is less than the calculation weight of the energy storage module, and the calculation weight of the fire occurrence probability of the associated modules is inversely proportional to the module location distance between the associated modules and the energy storage module.

[0141] The risk level of a fire in the energy storage module is obtained by multiplying the probability average parameter and the proportional weight; the proportional weight is proportional to the occurrence rate.

[0142] Optionally, the occurrence ratio can be frequency, conditional probability, or weighted ratio, and the present invention does not limit it.

[0143] Optionally, the determination of the same reason can be an exact match or a semantic similarity match, and the present invention does not limit it.

[0144] Optionally, the weight of the energy storage module can be 0.6-0.9, and the total weight of the associated modules can be 0.1-0.4. This invention does not impose any limitations.

[0145] Optionally, the proportionality between the weight and the occurrence ratio can be a linear, exponential, or piecewise function, and this invention does not impose any limitations.

[0146] As can be seen, through the above optional embodiments, the degree of danger of fire information of the module is determined by calculating the proportion of fire causes and combining the weighted probability average and proportional weight of the weight of the associated module corresponding to the location distance. Thus, on the basis of accurate screening of dangerous modules, the comprehensiveness and spatial correlation of the degree of danger assessment are improved by cause association and distance weighted analysis, providing accurate quantitative basis for the identification of high-risk modules and reducing the risk of misjudgment of degree of danger due to neglect of location relationship.

[0147] As an optional embodiment, the step above, calculating the matching degree between the module fire information of all hazardous energy storage modules and the power plant fire information, includes:

[0148] The causes of all fires in hazardous energy storage modules are combined into a primary cause set;

[0149] Combine the fire causes corresponding to all model locations in the power plant fire information into a second set of causes;

[0150] The probability of fire occurrence of all hazardous energy storage modules and the corresponding module locations are combined into a first probability set;

[0151] The model locations and corresponding fire occurrence probabilities in the power plant fire information are combined into a second probability set;

[0152] Calculate the first similarity between the first set of causes and the second set of causes;

[0153] Calculate the second similarity between the first probability set and the second probability set;

[0154] Calculate the product of the first similarity and the second similarity to obtain the matching degree between the module fire information of all hazardous energy storage modules and the power plant fire information.

[0155] Optionally, the first or second set of causes may be a list of cause categories, a probability distribution, or an embedding vector, and this invention does not impose any limitations.

[0156] Optionally, the first similarity can be Jaccard similarity, cosine similarity, inverse KL divergence, or Spearman correlation coefficient of causal ranking, and this invention does not limit it.

[0157] Optionally, the first probability set or the second probability set can be a (location, probability) point set or a spatial probability field, which is not limited in this invention.

[0158] Optionally, the second similarity can be spatial correlation coefficient, inverse Earth Mover's Distance, SSIM, or IoU in high-risk regions; this invention does not limit the specific similarity.

[0159] As can be seen, through the above optional embodiments, by constructing the fire cause set and probability-location set of the hazardous energy storage module and the power station as a whole, respectively, and calculating the product of cause similarity and probability similarity as the matching degree, the accuracy and robustness of the matching assessment are improved by fusing cause and probability dimensions on the basis of accurate matching degree calculation. This provides high-quality matching features for predicting the probability of fire and reduces the risk of misjudging the overall fire probability due to single-dimensional bias.

[0160] Example 2: Please refer to Figure 2 , Figure 2 This is a schematic diagram of the structure of a fire monitoring system for an energy storage power station based on multimodal data, as disclosed in an embodiment of the present invention. Figure 2 The described fire monitoring system for energy storage power stations based on multimodal data can be applied to data processing systems / data processing equipment / data processing servers (wherein, the server includes a local processing server or a cloud processing server). For example... Figure 2 As shown, the fire monitoring system for energy storage power stations based on multimodal data may include:

[0161] The acquisition module 201 is used to acquire the power station sensing data of the energy storage power station and the module sensing data of any one of the energy storage modules.

[0162] The first prediction module 202 is used to predict the fire information of the energy storage power station based on the fire risk identification model and the power station sensor data.

[0163] The second prediction module 203 is used to predict the corresponding fire information of the energy storage module based on the risk identification model corresponding to the energy storage module and the module's sensor data.

[0164] Matching module 204 is used to determine the probability of fire in the energy storage power station based on the matching calculation between the power station fire information and the module fire information.

[0165] As can be seen, the above-described embodiments of the invention acquire power station sensor data and module sensor data of any energy storage module, predict power station fire information and module fire information based on a fire risk identification model, and determine the probability of fire in the energy storage power station by matching the two. This enables accurate fire risk assessment based on multi-scale modal data fusion and matching verification, improves the accuracy and reliability of fire early warning for energy storage power stations, and reduces the risk of missed fire reports due to local anomalies not being associated with the whole.

[0166] As an optional embodiment, the energy storage module can be an energy storage battery area, an energy storage control area, an energy storage mechanical structure area, or an energy storage maintenance channel area in an energy storage power station.

[0167] As can be seen, the above optional embodiments limit the types of energy storage modules, so that the technical solution of the present invention can be applied to different energy storage power station areas, helping to improve the accuracy and reliability of fire early warning in energy storage power stations and reduce the risk of fire underreporting caused by local anomalies not being associated with the whole.

[0168] As an optional embodiment, the power station sensing data is obtained by sensors installed in the energy storage power station and is used to characterize the sensing characteristics of the entire energy storage power station; the module sensing data is obtained by sensors installed in the energy storage module and is used to characterize the sensing characteristics within the energy storage module; the sensors include at least one of temperature sensors, humidity sensors, photoelectric reflection sensors, image sensors, and sound sensors.

[0169] As can be seen, the above optional embodiments define the content and differences between power station sensing data and module sensing data, as well as the corresponding sensor types, so that multimodal sensing data can effectively characterize the real-time operating characteristics of the power station as a whole or individual modules, assist in realizing accurate fire risk assessment based on multi-scale modal data fusion and matching verification, improve the accuracy and reliability of fire early warning for energy storage power stations, and reduce the risk of missed fire reports due to local anomalies not being associated with the whole.

[0170] As an optional embodiment, the first prediction module, based on a fire risk identification model and according to power station sensor data, predicts specific methods for power station fire information of the energy storage power station, including:

[0171] All power station sensor data, corresponding sensor locations, and the 3D structural model of the energy storage power station are input into a trained fire risk identification model to obtain power station fire information. Optionally, the power station fire information includes the probability of fire occurrence and the cause of fire corresponding to each model location in the 3D structural model. The fire risk identification model is trained using a training dataset that includes multiple training power station sensor data and corresponding sensor location annotations, 3D structural model location annotations, and fire information annotations.

[0172] As can be seen, through the above optional embodiments, by inputting the power plant sensor data, sensor location, and three-dimensional structural model into the trained fire risk identification model, the probability and cause of fire occurrence at each location in the three-dimensional structural model are output. Thus, based on accurate power plant fire information prediction, the level of refinement and visualization of fire risk assessment is improved through spatial location correlation and three-dimensional modeling, providing high-quality global data support for module fire information matching and reducing the risk of fire location deviation caused by missing spatial information.

[0173] As an optional embodiment, the second prediction module predicts the corresponding module fire information based on the risk identification model corresponding to the energy storage module and the module's sensor data in the following specific ways:

[0174] Determine the module parameters corresponding to the energy storage module;

[0175] Based on the module parameters and a similarity algorithm, the target risk identification model is determined from multiple candidate risk identification models.

[0176] The module sensor data is input into the target risk identification model to obtain the module fire information corresponding to the energy storage module; the module fire information includes the probability of fire occurrence and the cause of fire in the energy storage module.

[0177] As can be seen, through the above optional embodiments, by selecting a target risk identification model based on a similarity algorithm according to the energy storage module parameters and inputting module sensor data to predict module fire information, the prediction is more targeted and accurate based on the accurate determination of module fire information, and reliable local risk data is provided for the calculation of the probability of fire, reducing the risk of misjudgment of module fire due to the generality of the model.

[0178] As an optional embodiment, the matching module determines the specific method by which it calculates the probability of a fire at the energy storage power station based on the matching calculation between the power station fire information and the module fire information, including:

[0179] From all energy storage modules in the energy storage power station, modules whose fire hazard level exceeds a preset threshold are selected to obtain multiple hazardous energy storage modules;

[0180] Calculate the matching degree between the module fire information and the power plant fire information of all hazardous energy storage modules;

[0181] Calculate the average fire probability of all hazardous energy storage modules to obtain probability parameters;

[0182] The matching degree and probability parameters are input into the trained fire probability prediction model to obtain the possible fire probability of the energy storage power station. The fire probability prediction model is trained using a training dataset that includes multiple training matching degrees, corresponding training probability parameter labels, and corresponding fire probability labels.

[0183] As can be seen, through the above optional embodiments, by screening dangerous energy storage modules whose hazard level exceeds the threshold and calculating the matching degree between their module fire information and power station fire information, and combining the probability parameters input into the fire occurrence probability prediction model to obtain the final possible fire probability, the sensitivity and scientific nature of the assessment are improved by screening dangerous modules and multi-dimensional matching on the basis of accurate fire probability prediction. This provides forward-looking support for fire early warning of energy storage power stations and reduces the risk of delayed early warning caused by the failure to amplify the overall risk due to local high risk.

[0184] As an optional embodiment, the hazard level of the module fire information is calculated through the following steps:

[0185] For each energy storage module, calculate the proportion of fire causes occurring in that energy storage module compared to all fire causes in all energy storage modules.

[0186] Among all other energy storage modules, identify the module with the same fire cause as the energy storage module to obtain multiple associated modules;

[0187] Calculate the weighted summation average of the fire occurrence probabilities of the energy storage module and all associated modules to obtain the probability average parameter; optionally, the calculation weight of the fire occurrence probability of the associated modules is less than the calculation weight of the energy storage module, and the calculation weight of the fire occurrence probability of the associated modules is inversely proportional to the module location distance between the associated modules and the energy storage module.

[0188] The risk level of a fire in the energy storage module is obtained by multiplying the probability average parameter and the proportional weight; the proportional weight is proportional to the occurrence rate.

[0189] As can be seen, through the above optional embodiments, the degree of danger of fire information of the module is determined by calculating the proportion of fire causes and combining the weighted probability average and proportional weight of the weight of the associated module corresponding to the location distance. Thus, on the basis of accurate screening of dangerous modules, the comprehensiveness and spatial correlation of the degree of danger assessment are improved by cause association and distance weighted analysis, providing accurate quantitative basis for the identification of high-risk modules and reducing the risk of misjudgment of degree of danger due to neglect of location relationship.

[0190] As an optional embodiment, the specific method by which the matching module calculates the matching degree between the module fire information of all hazardous energy storage modules and the power plant fire information includes:

[0191] The causes of all fires in hazardous energy storage modules are combined into a primary cause set;

[0192] Combine the fire causes corresponding to all model locations in the power plant fire information into a second set of causes;

[0193] The probability of fire occurrence of all hazardous energy storage modules and the corresponding module locations are combined into a first probability set;

[0194] The model locations and corresponding fire occurrence probabilities in the power plant fire information are combined into a second probability set;

[0195] Calculate the first similarity between the first set of causes and the second set of causes;

[0196] Calculate the second similarity between the first probability set and the second probability set;

[0197] Calculate the product of the first similarity and the second similarity to obtain the matching degree between the module fire information of all hazardous energy storage modules and the power plant fire information.

[0198] As can be seen, through the above optional embodiments, by constructing the fire cause set and probability-location set of the hazardous energy storage module and the power station as a whole, respectively, and calculating the product of cause similarity and probability similarity as the matching degree, the accuracy and robustness of the matching assessment are improved by fusing cause and probability dimensions on the basis of accurate matching degree calculation. This provides high-quality matching features for predicting the probability of fire and reduces the risk of misjudging the overall fire probability due to single-dimensional bias.

[0199] Example 3: Please refer to Figure 3 , Figure 3 This is another fire monitoring system for energy storage power stations based on multimodal data disclosed in the embodiments of the present invention. Figure 3The described multimodal data-based fire monitoring system for energy storage power stations is applied in a data processing system / data processing equipment / data processing server (wherein, the server includes a local processing server or a cloud processing server). For example... Figure 3 As shown, the fire monitoring system for energy storage power stations based on multimodal data may include:

[0200] Memory 301 storing executable program code;

[0201] Processor 302 coupled to memory 301;

[0202] The processor 302 calls the executable program code stored in the memory 301 to execute the steps of the fire monitoring method for energy storage power stations based on multimodal data described in Embodiment 1.

[0203] Example 4: This embodiment of the invention discloses a computer read storage medium that stores a computer program for electronic data exchange, wherein the computer program causes a computer to execute the steps of the fire monitoring method for energy storage power stations based on multimodal data described in Example 1.

[0204] Example 5: This embodiment of the invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to perform the steps of the fire monitoring method for energy storage power stations based on multimodal data described in Example 1.

[0205] The foregoing has described specific embodiments of this specification; other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims may be performed in a different order than those shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily have to follow the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0206] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0207] For ease of description, the above devices are described in terms of function, divided into various units. Of course, in implementing this specification, the functions of each unit can be implemented in one or more software and / or hardware components.

[0208] Those skilled in the art will understand that the embodiments of this specification can be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the embodiments of this specification can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0209] This specification is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this specification. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0210] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0211] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0212] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0213] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0214] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0215] It should also be noted that 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 a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0216] This specification can be described in the general context of computer-executable instructions that are executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This specification can also be practiced in distributed computing environments, where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0217] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0218] Finally, it should be noted that the fire monitoring method and system for energy storage power stations based on multimodal data disclosed in the embodiments of the present invention are merely preferred embodiments of the present invention and are only used to illustrate the technical solutions of the present invention, not to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A fire monitoring method for energy storage power stations based on multimodal data, characterized in that, The method includes: Acquire the power station sensor data and the module sensor data of any one of the energy storage modules in the energy storage power station; Based on the fire risk identification model, the fire information of the energy storage power station is predicted according to the power station sensor data. Based on the risk identification model corresponding to the energy storage module, and according to the sensor data of the module, the corresponding module fire information is predicted; Based on the matching calculation between the power station fire information and the module fire information, the probability of a fire at the energy storage power station is determined, including: From all the energy storage modules of the energy storage power station, modules whose fire hazard level is greater than a preset threshold are selected to obtain multiple hazardous energy storage modules; Calculate the matching degree between the fire information of all the hazardous energy storage modules and the fire information of the power plant; Calculate the average fire occurrence probability of all the hazardous energy storage modules to obtain the probability parameter; The matching degree and the probability parameters are input into the trained fire occurrence probability prediction model to obtain the possible fire probability of the energy storage power station; the fire occurrence probability prediction model is trained by a training dataset that includes multiple training matching degrees, corresponding training probability parameter labels, and corresponding fire probability labels.

2. The fire monitoring method for energy storage power stations based on multimodal data according to claim 1, characterized in that, The energy storage module is the energy storage battery area, energy storage control area, energy storage mechanical structure area, or energy storage maintenance channel area in the energy storage power station.

3. The fire monitoring method for energy storage power stations based on multimodal data according to claim 1, characterized in that, The power station sensing data is obtained through sensors installed in the energy storage power station and is used to characterize the sensing characteristics of the entire energy storage power station; the module sensing data is obtained through sensors installed in the energy storage module and is used to characterize the sensing characteristics within the energy storage module; the sensors include at least one of temperature sensors, humidity sensors, photoelectric reflection sensors, image sensors, and sound sensors.

4. The fire monitoring method for energy storage power stations based on multimodal data according to claim 1, characterized in that, The fire risk identification model, based on the power station sensor data, predicts the power station fire information of the energy storage power station, including: All the power station sensor data, corresponding sensor locations, and the three-dimensional structural model of the energy storage power station are input into a trained fire risk identification model to obtain the power station fire information of the energy storage power station. The power station fire information includes the fire occurrence probability and fire cause corresponding to each model location in the three-dimensional structural model. The fire risk identification model is trained using a training dataset that includes multiple training power station sensor data and corresponding sensor location labels, three-dimensional structural model location labels, and fire information labels.

5. The fire monitoring method for energy storage power stations based on multimodal data according to claim 4, characterized in that, The risk identification model based on the energy storage module predicts the corresponding module fire information based on the module's sensor data, including: Determine the module parameters corresponding to the energy storage module; Based on the module parameters, and using a similarity algorithm, the target risk identification model is determined from multiple candidate risk identification models. The module sensing data is input into the target risk identification model to obtain the module fire information corresponding to the energy storage module; the module fire information includes the fire occurrence probability and fire cause of the energy storage module.

6. The fire monitoring method for energy storage power stations based on multimodal data according to claim 5, characterized in that, The severity level of the fire information in the module is calculated through the following steps: For each of the energy storage modules, calculate the proportion of the fire cause of that energy storage module among all the fire causes of all the energy storage modules; Among all the other energy storage modules, identify the module with the same cause of fire as the energy storage module to obtain multiple associated modules; Calculate the weighted summation average of the fire occurrence probabilities of the energy storage module and all the associated modules to obtain the probability average parameter; wherein, the calculation weight of the fire occurrence probability of the associated module is less than the calculation weight of the energy storage module, and the calculation weight of the fire occurrence probability of the associated module is inversely proportional to the module location distance between the associated module and the energy storage module. The product of the probability average parameter and the proportional weight is calculated to obtain the degree of danger of the fire information of the energy storage module; the proportional weight is proportional to the occurrence ratio.

7. The fire monitoring method for energy storage power stations based on multimodal data according to claim 5, characterized in that, The calculation of the matching degree between the fire information of all the hazardous energy storage modules and the fire information of the power plant includes: The causes of fires in all the aforementioned hazardous energy storage modules are combined into a first set of causes; The fire causes corresponding to all model locations in the power plant fire information are combined into a second set of causes; The fire occurrence probability of all the hazardous energy storage modules and their corresponding module locations are combined into a first probability set; The model locations and corresponding fire occurrence probabilities in the power plant fire information are combined into a second probability set; Calculate the first similarity between the first set of causes and the second set of causes; Calculate the second similarity between the first probability set and the second probability set; The product of the first similarity and the second similarity is calculated to obtain the matching degree between the module fire information of all the hazardous energy storage modules and the power plant fire information.

8. A fire monitoring system for energy storage power stations based on multimodal data, characterized in that, The system includes: The acquisition module is used to acquire the power station sensing data of the energy storage power station and the module sensing data of any one of the energy storage modules. The first prediction module is used to predict the fire information of the energy storage power station based on the fire risk identification model and the power station sensor data. The second prediction module is used to predict the corresponding fire information of the energy storage module based on the risk identification model corresponding to the energy storage module and the sensor data of the module. The matching module is used to determine the probability of a fire at the energy storage power station based on the matching calculation between the power station fire information and the module fire information, including: From all the energy storage modules of the energy storage power station, modules whose fire hazard level is greater than a preset threshold are selected to obtain multiple hazardous energy storage modules; Calculate the matching degree between the fire information of all the hazardous energy storage modules and the fire information of the power plant; Calculate the average fire occurrence probability of all the hazardous energy storage modules to obtain the probability parameter; The matching degree and the probability parameters are input into the trained fire occurrence probability prediction model to obtain the possible fire probability of the energy storage power station; the fire occurrence probability prediction model is trained by a training dataset that includes multiple training matching degrees, corresponding training probability parameter labels, and corresponding fire probability labels.

9. A fire monitoring system for energy storage power stations based on multimodal data, characterized in that, The system includes: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the fire monitoring method for energy storage power stations based on multimodal data as described in any one of claims 1-7.

Images