Energy storage cabinet fire early warning method and system based on multi-sensor fusion

Abstract

The invention relates to the technical field of energy storage cabinet safety early warning, and provides an energy storage cabinet fire early warning method and system based on multi-sensor fusion, and the method comprises the steps: obtaining the temperature data and gas data in an energy storage cabinet, and the operation state information of a cooling fan; in the ideal working period, a reference response mode of the temperature sensor and the gas sensor to the running state change is established; acquiring a real-time response mode during daily work; the real-time response mode of the temperature sensor is compared with the reference response mode, and when a preset deviation exists, it is judged that a heat transfer path of the temperature sensor is abnormal; the real-time response mode of the gas sensor is compared with the reference response mode, and when a preset deviation exists, it is judged that the gas sensor has an abnormal response to fan disturbance; and when the abnormity is judged to exist at the same time, early fire early warning is output. The method has the effect of improving the accuracy, sensitivity and reliability of early fire warning of the energy storage cabinet.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of energy storage cabinet safety early warning, and particularly relates to an energy storage cabinet fire early warning method and system based on multi-sensor fusion. BACKGROUND

[0002] In modern society, the demand for energy is growing, and the internal safe operation of the energy storage cabinet as a key energy infrastructure is particularly important. In order to ensure the safety of the energy storage cabinet, a set of precise fire early warning system is usually deployed to monitor the internal conditions in real time through various sensors, so as to discover and handle the fire at the initial stage. However, in actual operation, these early warning systems may face some unexpected challenges, which greatly reduces the effectiveness of early warning.

[0003] At the initial stage of the battery module thermal runaway, the single temperature sensor is delayed in response due to the aging of the heat conduction medium, so that the data cannot timely trigger the temperature threshold. At the same time, the gas sensor is disturbed by the internal wind field, so that the early gas concentration is diluted, and the data is in the normal range for a long time. Moreover, the failure elimination logic of the early warning system shields the real early abnormal signal as invalid data. The complex situation of multiple factors interweaving makes the traditional early warning method relying on a single information source or simple logic judgment completely invalid, and cannot accurately identify and issue high-precision early fire warning when the early thermal runaway of the battery module in the energy storage cabinet occurs.

[0004] In view of the above problems, the prior art needs to be improved. SUMMARY

[0005] The application discloses an energy storage cabinet fire early warning method and system based on multi-sensor fusion, which aims to solve the technical problem that the traditional early warning method cannot accurately identify and issue high-precision early fire warning when the early thermal runaway of the battery module occurs, due to the interweaving of multiple factors such as sensor response delay, gas dilution and system failure elimination logic misjudgment during the long-term operation of the energy storage cabinet.

[0006] The technical scheme of the application is as follows: In a first aspect, the application discloses an energy storage cabinet fire early warning method based on multi-sensor fusion, which specifically comprises: obtaining temperature data of a temperature sensor, gas data of a gas sensor and running state information of a heat dissipation fan in the energy storage cabinet; During the ideal working period of the energy storage cabinet in normal operation, the running state of the heat dissipation fan is controlled to change, the temperature data and the gas data responding to the change of the running state are monitored, the response mode of the temperature sensor and the gas sensor to the change of the running state of the heat dissipation fan is recorded, so as to establish the baseline response mode of the temperature sensor and the gas sensor to the change of the running state; During the daily work of the energy storage cabinet, the real-time response mode of the temperature sensor and the gas sensor to the change of the running state of the heat dissipation fan under the actual working environment is obtained in real time; The real-time response mode corresponding to the temperature sensor is compared with the baseline response mode to obtain a temperature response mode comparison result, and when the temperature response mode comparison result indicates that there is a preset deviation, it is determined that the temperature sensor has an abnormal heat transfer path; The real-time response mode corresponding to the gas sensor is compared with the baseline response mode to obtain a gas response mode comparison result, and when the gas response mode comparison result indicates that there is a preset deviation, it is determined that the gas sensor has an abnormal response to the disturbance of the fan; When it is simultaneously determined that the temperature sensor has an abnormal heat transfer path and the gas sensor has an abnormal response to the disturbance of the fan, an early fire warning is output.

[0007] Through the technical scheme, the present application can effectively solve the problems of not timely and inaccurate early warning caused by single sensor failure or environmental interference in the prior art, and through multi-sensor fusion and dynamic baseline comparison, the accuracy and reliability of early fire warning of the energy storage cabinet are significantly improved. Especially in the complex working conditions of sensor response lag and gas dilution, high-precision early warning can still be realized.

[0008] Further, on the basis of the above, the present application further proposes that during the ideal working period of the energy storage cabinet in normal operation, the running state of the heat dissipation fan is controlled to change, the temperature data and the gas data responding to the change of the running state are monitored, the response mode of the temperature sensor and the gas sensor to the change of the running state of the heat dissipation fan is recorded, so as to establish the baseline response mode of the temperature sensor and the gas sensor to the change of the running state, which comprises the following steps: Obtain the external environment parameters and the internal battery load parameters of the energy storage cabinet; During the ideal working period of the energy storage cabinet in normal operation, the running state of the heat dissipation fan is controlled to change, the temperature data and the gas data responding to the change of the running state are monitored, the response mode of the temperature sensor and the gas sensor to the change of the running state of the heat dissipation fan is recorded, and the influence characteristics of the external environment parameters and the internal battery load parameters on the response mode are characterized according to the external environment parameters and the internal battery load parameters; According to the external environment parameters and the internal battery load parameters, and the influence characteristics, the dynamic expected response mode of the temperature sensor and the gas sensor to the change of the running state is calculated; The real-time response mode corresponding to the temperature sensor is compared with the dynamic expected response mode, and when there is a preset deviation between the real-time response mode and the dynamic expected response mode, it is determined that the temperature sensor has an abnormal heat transfer path; The real-time response mode corresponding to the gas sensor is compared with the dynamic expected response mode, and when there is a preset deviation between the real-time response mode and the dynamic expected response mode, it is determined that the gas sensor has an abnormal response to fan disturbance; When it is determined that the temperature sensor has an abnormal heat transfer path and the gas sensor has an abnormal response to fan disturbance at the same time, an early fire warning is output.

[0009] Through the technical scheme, the application can dynamically adjust the reference response mode according to external environment and internal load parameters, so that the warning system can adapt to the operation change of the energy storage cabinet under different working conditions, thereby avoiding false alarms or missed alarms caused by fixed reference mode, and further improving the accuracy and robustness of the warning.

[0010] Further, on the basis of the above, the application further proposes that, during the ideal working period of the normal operation of the energy storage cabinet, the running state change of the heat dissipation fan is controlled, the temperature data and gas data responding to the running state change are monitored, the response mode of the temperature sensor and the gas sensor to the running state change of the heat dissipation fan is recorded, and the influence characteristics of external environment parameters and internal battery load parameters on the response mode are characterized according to the external environment parameters and the internal battery load parameters. The steps include: During the ideal working period of the normal operation of the energy storage cabinet, the running situation of the energy storage cabinet is identified according to the external environment parameters and the internal battery load parameters; For each running situation, a feature set of the expected response mode of the temperature sensor and the gas sensor to the running state change under the running situation is maintained; The real-time response mode is compared with the feature set of the corresponding expected response mode under the running situation to obtain an expected comparison result; According to the expected comparison result, the feature set of the corresponding expected response mode under the running situation is adjusted.

[0011] Through the technical scheme, the application can identify different running situations and maintain the corresponding feature set, so that the system can more finely manage and adjust the reference response mode, thereby still maintaining high-precision abnormal detection capability in complex and variable operating environments.

[0012] Preferably, on the basis of the above, the application further proposes that the step of adjusting the feature set of the corresponding expected response mode under the running situation according to the expected comparison result includes: The real-time response mode is subjected to a moving average processing to obtain an average deviation value; When the average deviation value is still stable above the preset adjustment threshold after continuously experiencing a preset number of periods, the feature set of the corresponding expected response mode under the running situation is triggered to be adjusted.

[0013] Through the technical solution, the application can effectively filter out transient noise and short-term fluctuations through moving average processing and continuous stability judgment, ensure that only continuous and significant deviations will trigger the adjustment of the reference mode, thereby avoiding frequent and unnecessary adjustments, and improving the stability and reliability of the system.

[0014] In some preferred embodiments, the application further proposes that during the daily work of the energy storage cabinet, the step of obtaining the real-time response mode of the running state change of the heat dissipation fan under the actual working environment by the temperature sensor and the gas sensor in real time comprises: continuously monitoring the start and stop and speed of the heat dissipation fan, and generating a disturbance event when the speed change exceeds a preset threshold; setting a monitoring time window before and after the disturbance event, and recording the temperature change curve of the temperature sensor and the concentration change curve of the gas sensor in the monitoring time window at a preset sampling frequency; extracting feature parameters from the temperature change curve and the concentration change curve; aligning the feature parameters with the running state change in time to form a temperature response mode and a gas response mode as a real-time response mode; storing the real-time response mode.

[0015] Through the technical solution, the application can accurately monitor the fan disturbance event, set a monitoring time window before and after it, and combine feature parameter extraction and time alignment to ensure that the acquisition of the real-time response mode has high precision and high timeliness, thereby providing a reliable data basis for subsequent abnormality determination.

[0016] On the basis of the above, the application further proposes that when the average deviation value is still stable above the preset adjustment threshold after continuously experiencing a preset number of periods, the feature set of the corresponding expected response mode under the running situation is triggered to be adjusted, comprising: determining whether the average deviation value is still stable above the preset adjustment threshold after continuously experiencing a preset number of periods to obtain an average deviation determination result; determining whether the overall running state of the energy storage cabinet is stable and has no early fire auxiliary indication to obtain a running state determination result; when the average deviation determination result and the running state determination result are both yes, triggering the adjustment of the feature set of the corresponding expected response mode under the running situation; adopting a gradual weighted average method to integrate the average deviation value into the feature set of the expected response mode; limiting the amplitude of each adjustment to be not more than a preset maximum value.

[0017] Through this technical solution, this application can ensure that the adjustment of the baseline mode is carried out under the premise of the overall safety and stability of the energy storage cabinet by introducing overall operation status judgment and auxiliary indication. By adopting progressive weighted averaging and amplitude limitation, it can effectively avoid excessive adjustment of the baseline mode due to misjudgment or drastic changes, thereby improving the system's adaptability and safety.

[0018] More specifically, in some implementation schemes, this application also proposes that the steps for determining whether the overall operating status of the energy storage cabinet is stable and without early fire indicators, and obtaining the operating status determination result, include: Acquire real-time data of various auxiliary indicator parameters within the energy storage cabinet; Based on real-time data, calculate the fluctuation range of each auxiliary indicator parameter within a preset time window; The fluctuation range is compared with a preset range threshold to obtain the fluctuation range comparison result; When the fluctuation range comparison result indicates that the fluctuation range exceeds the preset range threshold, the real-time data of the corresponding auxiliary indicator parameter will be removed from the overall operation status judgment. When all auxiliary indicator parameters are in a stable state after the removal operation is completed, and there are no auxiliary indicators of early fire, determine whether the overall operating status of the energy storage cabinet is stable and there are no auxiliary indicators of early fire, and obtain the operating status judgment result.

[0019] Through this technical solution, this application can effectively avoid the interference of a single or a few abnormal auxiliary indicator parameters on the overall operating status judgment by judging the fluctuation range of auxiliary indicator parameters and eliminating anomalies, thereby ensuring the accuracy and reliability of the overall operating status judgment.

[0020] Based on the above, this application further proposes that when the fluctuation range comparison result indicates that the fluctuation range exceeds a preset range threshold, the step of removing the real-time data of the corresponding auxiliary indicator parameter from the overall operating status judgment includes: The average deviation value is compared with the preset group deviation threshold. When the average deviation value exceeds the group deviation threshold, it is determined that the corresponding auxiliary indicator sensor has a group offset. When a collective offset is detected in the corresponding auxiliary indicator sensor, the real-time data of the auxiliary indicator parameter corresponding to the auxiliary indicator sensor is retained as a potential early abnormal signal, and further anomaly analysis is triggered.

[0021] Through this technical solution, this application can identify and retain auxiliary indicator sensor data with group deviation by introducing a group deviation threshold, and use it as a potential early abnormal signal for further analysis, thereby avoiding misjudging real abnormal signals as invalid data and improving the sensitivity of early fire warning.

[0022] In some preferred embodiments, this application further proposes that when a collective shift is determined to exist in the corresponding auxiliary indicator sensor, the real-time data of the auxiliary indicator parameter corresponding to the auxiliary indicator sensor is retained as a potential early anomaly signal, and further anomaly analysis is triggered. This includes the following steps: Obtain real-time data on the voltage, current, temperature, and internal resistance parameters of the battery modules involved in the group offset; By cross-comparing and time-series analyzing voltage, current, temperature, and internal resistance parameters, preliminary results of anomaly type judgment are obtained. Based on the preliminary anomaly type assessment, an infrared thermal imaging scan of the battery module was initiated to obtain an image of the surface temperature distribution of the battery module. By analyzing the surface temperature distribution image and combining it with the gas data from the gas sensor, the final anomaly type determination result is obtained. Based on the final anomaly type determination result, output the anomaly type and handling suggestions.

[0023] Through this technical solution, this application can conduct in-depth analysis of potential early abnormal signals using multiple methods such as multi-parameter cross-comparison, time-series analysis, and infrared thermal imaging scanning. This enables accurate identification of abnormality types and provides specific processing suggestions, significantly improving the intelligence and practicality of the early warning system.

[0024] Secondly, this application also discloses an early warning system for energy storage cabinet fires based on multi-sensor fusion, used to perform early warning of energy storage cabinet fires based on multi-sensor fusion, specifically including: The operation status acquisition module is used to acquire temperature data from the temperature sensor, gas data from the gas sensor, and operation status information of the cooling fan inside the energy storage cabinet. The baseline response establishment module is used to monitor temperature and gas data in response to changes in the operating state of the cooling fan during the ideal working period of the energy storage cabinet, by controlling the changes in the operating state of the cooling fan, and to record the response patterns of the temperature and gas sensors to changes in the operating state of the cooling fan, so as to establish the baseline response patterns of the temperature and gas sensors to changes in the operating state. The real-time response acquisition module is used to acquire the real-time response mode of the temperature sensor and gas sensor to the changes in the operating status of the cooling fan in the actual working environment during the daily operation of the energy storage cabinet. The response mode comparison module is used to compare the real-time response mode of the temperature sensor with the reference response mode to obtain the temperature response mode comparison result. When the temperature response mode comparison result indicates that there is a preset deviation, it is determined that the temperature sensor has an abnormal heat transfer path. The abnormal response determination module is used to compare the real-time response mode of the gas sensor with the reference response mode to obtain the gas response mode comparison result. When the gas response mode comparison result indicates that there is a preset deviation, it is determined that the gas sensor has an abnormal response to the fan disturbance. The early warning output module is used to output an early fire warning when both the temperature sensor and the gas sensor are found to have abnormal heat transfer paths and abnormal responses to fan disturbances.

[0025] Through this technical solution, this application can effectively implement the early warning method for energy storage cabinet fires through modular design. The various modules work together to ensure the functional integrity and operational efficiency of the early warning system, thereby providing reliable hardware and software support for the safe operation of energy storage cabinets. Beneficial effects

[0026] This application provides an early fire warning method for energy storage cabinets based on multi-sensor fusion. By comprehensively analyzing temperature sensor data, gas sensor data, and the operating status information of the cooling fan, and establishing a dynamic baseline response pattern, it effectively solves the predicament of early fire warning failure caused by sensor response lag, gas dilution, and misjudgment in system fault elimination logic in existing technologies. Specifically, this application establishes a baseline response pattern of sensors to changes in fan operating status during ideal operating conditions, and acquires the actual response pattern in real time during daily operation, then compares the real-time response pattern with the baseline response pattern. When both the temperature sensor and the gas sensor show a preset deviation from the baseline pattern, it is determined that there is an abnormal heat transfer path and an abnormal response to fan disturbance, thereby outputting an early fire warning. This method can overcome the limitations of single sensor failure or environmental interference. For example, even if the temperature sensor has a delayed response due to aging of the heat-conducting medium, or the gas sensor has a low concentration due to dilution by fan airflow, this application can still identify anomalies by analyzing changes in their response patterns to fan disturbances, avoiding misjudging real early abnormal signals as invalid data. Therefore, this application significantly improves the accuracy, sensitivity, and reliability of early warning of fires in energy storage cabinets, providing a more solid guarantee for the safe operation of energy storage cabinets. Attached Figure Description

[0027] Figure 1 This is a flowchart of an early warning method for energy storage cabinet fire based on multi-sensor fusion in one embodiment of the present invention. Figure 2This is one of the flowcharts of an early warning method for energy storage cabinet fire based on multi-sensor fusion in another embodiment of the present invention; Figure 3 This is a second flowchart of a method for early warning of fire in an energy storage cabinet based on multi-sensor fusion, as described in another embodiment of the present invention. Figure 4 This is a system block diagram of an early warning system for energy storage cabinet fires based on multi-sensor fusion, according to another embodiment of the present invention. Explanation of reference numerals in the attached figures: 1. Early warning system for energy storage cabinet fire based on multi-sensor fusion; 11. Operation status acquisition module; 12. Baseline response establishment module; 13. Real-time response acquisition module; 14. Response mode comparison module; 15. Abnormal response judgment module; 16. Early warning output module. Detailed Implementation

[0028] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0029] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0030] Traditional early warning systems for energy storage cabinet fires face numerous challenges in practical operation. For example, temperature sensors may become sluggish due to the aging of the heat-conducting medium, failing to accurately reflect the true thermal state of the battery modules; gas sensors may experience a significant reduction in the concentration of early-generated combustible gases due to the dilution effect of the cooling fan, causing their data to remain within the normal range for an extended period; furthermore, the fault-detection logic of the warning system itself may incorrectly mask out sluggish early warning signals as invalid data. This complex situation, with its multiple intertwined factors, makes it difficult for traditional early warning methods relying on a single information source or simple logical judgments to accurately identify and issue high-precision early fire warnings when early thermal runaway occurs in the battery modules inside the energy storage cabinet.

[0031] To address this, this application proposes an early warning method for energy storage cabinet fires based on multi-sensor fusion, combining... Figure 1 As shown, it includes: S1, acquire temperature data from the temperature sensor inside the energy storage cabinet, gas data from the gas sensor, and operating status information of the cooling fan; S2, during the ideal working period of the energy storage cabinet, by controlling the change of the operating state of the cooling fan, monitoring the temperature data and gas data in response to the change of the operating state, and recording the response patterns of the temperature sensor and gas sensor to the change of the operating state of the cooling fan, so as to establish the reference response patterns of the temperature sensor and gas sensor to the change of the operating state. S3 is a real-time response mode that acquires the temperature and gas sensors in response to changes in the operating status of the cooling fan under the actual working environment during the daily operation of the energy storage cabinet. S4. Compare the real-time response mode corresponding to the temperature sensor with the reference response mode to obtain the temperature response mode comparison result. When the temperature response mode comparison result indicates that there is a preset deviation, it is determined that the temperature sensor has an abnormal heat transfer path. S5. Compare the real-time response mode corresponding to the gas sensor with the reference response mode to obtain the gas response mode comparison result. When the gas response mode comparison result indicates that there is a preset deviation, it is determined that the gas sensor has an abnormal response to fan disturbance. S6, when it is determined that the temperature sensor has an abnormal heat transfer path and the gas sensor has an abnormal response to fan disturbance, an early fire warning is output.

[0032] The term "energy storage cabinet" as used in this application refers to a device used to store electrical energy, typically comprising multiple battery modules, a battery management system, a cooling system, and various sensors. Its main function is to charge the battery during off-peak hours and discharge it during peak hours, thereby achieving peak shaving and valley filling, and improving grid stability. "Temperature sensors" are used to monitor the temperature of the internal environment of the energy storage cabinet or the surface of the battery modules in real time, typically employing thermistors, thermocouples, or infrared sensors. "Gas sensors" are used to detect the concentration of specific gases in the air inside the energy storage cabinet, such as carbon monoxide and hydrogen; these gases are early indicators of battery thermal runaway. "Cooling fans" are an important component of the energy storage cabinet's cooling system, dissipating heat generated inside the cabinet through forced convection to maintain the battery modules within a suitable temperature range. "Operating status information" refers to the current operating mode of the cooling fans, such as start / stop status and speed. "Response mode" refers to the specific pattern or curve of the temperature and gas sensor readings changing over time when the cooling fan's operating status changes.

[0033] In early warning methods for energy storage cabinet fires, the first step is to acquire temperature data from temperature sensors, gas data from gas sensors, and the operating status information of cooling fans within the cabinet. Temperature data can be collected in real-time by temperature sensors installed on the surface of the battery modules or inside the cabinet. For example, temperature signals can be converted into electrical signals using thermocouples or thermistors and transmitted to the data acquisition unit. Gas data can be acquired in real-time by gas sensors installed inside the cabinet. For example, the concentration of specific gases can be detected using electrochemical or semiconductor sensors and converted into electrical signals. The operating status information of the cooling fans can be directly provided by the fan controller, for example, by reading fan start / stop signals or speed feedback signals. This data and information are periodically collected and transmitted to the data processing unit for subsequent analysis.

[0034] Next, during the ideal operating period of the energy storage cabinet, by controlling the changes in the operating state of the cooling fan, temperature and gas data in response to these changes are monitored. The response patterns of the temperature and gas sensors to these changes are recorded to establish baseline response patterns for the temperature and gas sensors. For example, when the energy storage cabinet is in a stable operating state, the cooling fan can be manually started and stopped, or its speed can be adjusted from low to high and then back to low. During this process, the data changes from the temperature and gas sensors are continuously monitored. When the fan starts, the temperature typically decreases, and the gas concentration may be diluted; when the fan stops, the temperature typically rises, and the gas concentration may accumulate. By recording the temperature and gas concentration curves during these changes and extracting their characteristic parameters (such as rate of change, peak value, trough value, recovery time, etc.), typical response patterns of the sensors to fan disturbances under normal operating conditions can be established. These patterns are stored as baseline response patterns for subsequent comparative analysis.

[0035] During the daily operation of the energy storage cabinet, the system acquires real-time response patterns from temperature and gas sensors in response to changes in the operating status of the cooling fan under actual working conditions. For example, the system continuously monitors the operating status of the cooling fan, triggering a real-time response pattern acquisition whenever the fan starts, stops, or changes its speed. For a period before and after the fan status change, the system records data from the temperature and gas sensors at a high sampling frequency, generating real-time temperature and gas concentration change curves. Subsequently, feature extraction is performed on these curves to obtain the real-time response patterns. These real-time response patterns reflect the actual response to the current internal environment and sensor status of the energy storage cabinet.

[0036] Then, the real-time response mode of the temperature sensor is compared with the reference response mode to obtain the temperature response mode comparison result. When the temperature response mode comparison result indicates a preset deviation, it is determined that the temperature sensor has an abnormal heat transfer path. For example, the difference between the real-time response mode and the reference response mode can be quantified by calculating the Euclidean distance, correlation coefficient, or dynamic time warping (DTW) distance. If the temperature drop rate of the real-time response mode is significantly slower than that of the reference mode, or the temperature recovery time is significantly prolonged, and this deviation exceeds a preset threshold, it indicates that the path of heat transfer from the battery module to the temperature sensor may be abnormal, such as aging of the thermal conductive medium or poor contact, leading to increased thermal resistance.

[0037] Simultaneously, the real-time response mode of the gas sensor is compared with the baseline response mode to obtain the gas response mode comparison result. When the comparison result indicates a preset deviation, it is determined that the gas sensor has an abnormal response to fan disturbances. For example, if in the real-time response mode, the gas concentration decreases less after the fan starts than in the baseline mode, or the gas concentration increases faster after the fan stops than in the baseline mode, and this deviation exceeds a preset threshold, it indicates that the gas sensor's response to fan disturbances may be abnormal. This may mean that trace amounts of combustible gas have been generated inside the energy storage cabinet, but due to the dilution effect of the fan, the gas concentration has not reached the traditional threshold, but its response characteristics to fan disturbances have changed.

[0038] Finally, when both the temperature sensor and the gas sensor are simultaneously detected to have an abnormal heat transfer path and an abnormal response to fan disturbances, an early fire warning is issued. This dual-judgment mechanism effectively avoids false alarms caused by a single sensor malfunction or environmental interference. For example, if only the temperature sensor responds slowly, it may be a sensor failure rather than a fire; if only the gas sensor responds abnormally, it may be an ambient gas fluctuation rather than a fire. A warning is only issued when both are abnormal, and the abnormal pattern matches the characteristics of early thermal runaway, thus greatly improving the accuracy and reliability of the warning.

[0039] Optional, combined Figure 2 As shown, during the ideal operating period of the energy storage cabinet, S2 monitors temperature and gas data in response to changes in the operating state of the cooling fan by controlling the changes in the operating state of the cooling fan, and records the response patterns of the temperature and gas sensors to changes in the operating state of the cooling fan, in order to establish a reference response pattern of the temperature and gas sensors to changes in the operating state. The steps include: S21, Obtain the external environmental parameters and internal battery load parameters of the energy storage cabinet; S22, during the ideal working period of the energy storage cabinet, by controlling the change of the operating state of the cooling fan, monitoring the temperature data and gas data in response to the change of the operating state, recording the response mode of the temperature sensor and gas sensor to the change of the operating state of the cooling fan, and characterizing the influence characteristics of the external environmental parameters and internal battery load parameters on the response mode based on the external environmental parameters and internal battery load parameters. S23, based on external environmental parameters, internal battery load parameters, and influencing characteristics, calculate the dynamic expected response modes of the temperature sensor and gas sensor to changes in operating status. S24, compare the real-time response mode corresponding to the temperature sensor with the dynamic expected response mode. When there is a preset deviation between the real-time response mode and the dynamic expected response mode, it is determined that the temperature sensor has an abnormal heat transfer path. S25, compare the real-time response mode of the gas sensor with the dynamic expected response mode. When there is a preset deviation between the real-time response mode and the dynamic expected response mode, it is determined that the gas sensor has an abnormal response to the fan disturbance. S26, when it is simultaneously determined that the temperature sensor has an abnormal heat transfer path and the gas sensor has an abnormal response to fan disturbance, an early fire warning is output.

[0040] Specifically, external environmental parameters can be understood as various physical quantities of the external environment in which the energy storage cabinet is located, such as ambient temperature, humidity, and atmospheric pressure. These parameters affect the heat dissipation efficiency and gas diffusion rate inside the energy storage cabinet. Internal battery load parameters refer to parameters related to the operating status of the battery pack inside the energy storage cabinet, such as the battery's charging and discharging current, voltage, power, state of charge (SOC), and state of health (SOH). These parameters directly affect the heat generation and gas release of the battery. The purpose of obtaining these parameters is to gain a more comprehensive understanding of the energy storage cabinet's operating environment and to provide contextual information for subsequent response mode analysis.

[0041] Characterizing the impact of external environmental parameters and internal battery load parameters on response modes involves quantifying, through data analysis and modeling, the specific effects of changes in the cooling fan's operating state on the response modes of temperature and gas sensors under different environmental and load conditions. For example, in high-temperature environments, the temperature drop caused by fan startup may not be as significant as in low-temperature environments; during high-load operation, the battery itself may release trace amounts of gas, causing a shift in the baseline response of the gas sensor to fan disturbances. This step aims to establish the mapping relationship between these parameters and sensor response modes, providing a basis for subsequent calculations of the expected dynamic response modes.

[0042] In practical applications, calculating the dynamic expected response mode of temperature and gas sensors to changes in operating status refers to predicting the normal response mode that the temperature and gas sensors should exhibit when the cooling fan's operating status changes under a specific operating condition, based on currently acquired external environmental parameters and internal battery load parameters, combined with the characterized influence characteristics. This dynamic expected response mode can adaptively adjust according to actual operating conditions, rather than using a single fixed baseline mode, aiming to improve the accuracy and robustness of early warnings.

[0043] In some preferred embodiments, it is assumed that the energy storage cabinet operates in a high-temperature environment during summer, with an external ambient temperature of 35°C and the internal battery pack in a high-power discharge state. At this time, the temperature drop caused by the start-up or speed change of the cooling fan will be significantly different from the temperature drop in a low-temperature environment (e.g., 5°C) or under low-power operation. The solution of this application first obtains the ambient temperature of 35°C and the battery load parameters under high-power discharge. Then, based on a pre-established influence characteristic model, the system identifies the expected response pattern of the cooling fan to the temperature and gas sensors under the specific operating scenario of "high temperature and high load." For example, under high temperature and high load, the rate and magnitude of temperature drop after the fan starts may be smaller, and gas concentration fluctuations may be more drastic. The system calculates a dynamic expected response pattern for the current operating condition based on these parameters. Subsequently, the real-time sensor response pattern is compared with this dynamic expected response pattern. If there is a preset deviation between the real-time response pattern and this dynamic expected response pattern, such as a temperature drop much smaller than expected, or an abnormally high gas concentration, the system will determine that there is an abnormal heat transfer path or an abnormal response to fan disturbances, and output an early fire warning. This dynamic adjustment mechanism ensures the accuracy of early warnings and avoids false alarms under normal operating conditions such as high temperature and high load, where the response modes differ.

[0044] Optionally, during the ideal operating period of the energy storage cabinet, the steps of monitoring temperature and gas data in response to changes in the operating state of the cooling fan by controlling the changes in the operating state of the cooling fan, recording the response patterns of the temperature and gas sensors to changes in the operating state of the cooling fan, and characterizing the influence characteristics of external environmental parameters and internal battery load parameters on the response patterns based on external environmental parameters and internal battery load parameters include: During the ideal operating period of the energy storage cabinet, the operating conditions of the energy storage cabinet are identified based on external environmental parameters and internal battery load parameters. For each operating scenario, maintain a feature set of expected response patterns of temperature and gas sensors to changes in operating status under that operating scenario; The feature set of the real-time response mode is compared with the expected response mode under the corresponding runtime scenario to obtain the expected comparison result; Based on the expected comparison results, the feature set of the expected response mode under the operating scenario is adjusted.

[0045] Specifically, identifying the operating scenarios of an energy storage cabinet refers to classifying its operating state into several discrete or continuous scenarios based on a combination of external environmental parameters (such as ambient temperature and humidity) and internal battery load parameters (such as charging and discharging current, battery pack voltage, battery state of health (SOH), and state of charge (SOC)). For example, operating scenarios can be classified into "high temperature and high load," "low temperature and low load," and "medium temperature and medium load" based on factors such as ambient temperature and load current. The purpose is to refine the behavioral patterns of the energy storage cabinet under different operating conditions, so as to build a more accurate expected response model later.

[0046] Specifically, for each operating scenario, a feature set is maintained to represent the expected response patterns of the temperature and gas sensors to changes in operating conditions. This can be understood as establishing and continuously updating a set of parameters describing the typical response behavior of the temperature and gas sensors to changes in the cooling fan's operating state (such as start / stop and speed changes) for each identified operating scenario. These feature parameters may include, but are not limited to, the magnitude of temperature or gas concentration changes, response delay time, rise / fall rate, and settling time. The purpose is to provide a precise benchmark for "normal" behavior for each specific operating scenario.

[0047] In practical applications, the real-time response pattern is compared with the feature set of the expected response pattern under the current operating conditions to obtain the expected comparison results. Specifically, this involves comparing the real-time response patterns of the temperature and gas sensors to changes in the cooling fan's operating status, as currently monitored, with the feature set of the expected response pattern maintained under the current operating conditions, across multiple dimensions. For example, the Euclidean distance, correlation coefficient, or deviation value of specific characteristic parameters between the real-time response pattern and the expected feature set can be calculated. The purpose is to quantify the degree of difference between the real-time response and the expected baseline.

[0048] Furthermore, adjusting the feature set of the expected response mode corresponding to the operating scenario based on the expected comparison results means that when the expected comparison results show a certain degree of deviation between the real-time response mode and the expected feature set, and this deviation is judged to be non-abnormal (e.g., due to slight drift caused by long-term system operation or minor changes in environmental parameters), the feature set of the expected response mode maintained under that operating scenario is updated or corrected. For example, algorithms such as weighted averaging and exponential smoothing can be used to integrate new real-time response data into the feature set, enabling it to adaptively learn and adapt to the long-term operating characteristics of the energy storage cabinet. The purpose is to ensure that the feature set of the expected response mode can dynamically reflect the actual operating status of the energy storage cabinet, improving the accuracy and robustness of the benchmark model.

[0049] In some preferred embodiments, this application is implemented as follows. Assume that the operating conditions of the energy storage cabinet are divided into "Situation A: Ambient temperature 20-30℃, load rate 50-70%" and "Situation B: Ambient temperature 35-45℃, load rate 80-100%".

[0050] When the energy storage cabinet is in scenario A, the system maintains a set of characteristics for the expected response patterns of the temperature and gas sensors in scenario A. For example, this set of characteristics may include: when the cooling fan changes from stopped to full speed, the temperature sensor temperature drops by 2-3°C within 30 seconds and stabilizes within 60 seconds; the gas sensor concentration drops by 5-10 ppm within 10 seconds and stabilizes within 30 seconds.

[0051] During normal operation, if the system detects that the energy storage cabinet is currently in scenario A and the cooling fan's operating status changes, the system will acquire the response patterns of the temperature sensor and the gas sensor in real time. For example, if the temperature sensor detects a 2.5°C drop within 30 seconds and the gas sensor detects a 6 ppm drop in concentration within 10 seconds, the characteristic parameters of these real-time response patterns will be extracted.

[0052] Subsequently, the characteristic parameters of these real-time response modes are compared with the feature set of the expected response modes maintained under scenario A. If the comparison results show a slight but persistent deviation between the real-time response and the expected feature set (e.g., the temperature drop remains stable at 2.8°C for a long period, slightly higher than the expected range), the system will fine-tune the feature set of the expected response modes for scenario A according to a preset adjustment strategy (e.g., using a progressive weighted average) to better reflect the actual normal behavior of the energy storage cabinet under the current scenario A. In this way, the feature set of the expected response modes can be continuously optimized to ensure that it always remains consistent with the actual operating state of the energy storage cabinet, thereby improving the accuracy of anomaly detection.

[0053] Optionally, the steps for adjusting the feature set of the expected response pattern under the operating scenario based on the expected comparison results include: The average deviation value is obtained by applying a moving average to the real-time response pattern. When the average deviation value remains stable above the preset adjustment threshold after a preset number of consecutive cycles, the feature set of the expected response mode corresponding to the operating scenario is adjusted.

[0054] Specifically, applying a moving average to the real-time response pattern involves continuously averaging the deviation between the real-time response pattern and the expected response pattern within a certain time window. This eliminates the influence of instantaneous fluctuations and noise, resulting in a smoother and more representative average deviation value. This average deviation value reflects the long-term trend and stability of the deviation. The moving average process can be implemented by setting a sliding window size; for example, the deviation values ​​over the most recent N sampling periods can be averaged.

[0055] Furthermore, only when the average deviation value remains stable above a preset adjustment threshold after a preset number of consecutive cycles will an adjustment to the feature set of the expected response mode corresponding to the operating scenario be triggered. Here, the "preset number of cycles" refers to a time length to ensure the persistence of the deviation; the "preset adjustment threshold" is a quantitative standard used to determine whether the deviation is significant enough to trigger adjustment. Only when the average deviation value not only persists but also exceeds the set threshold is the deviation considered systematic and requires adjustment of the baseline mode.

[0056] In some preferred embodiments, a specific example is given below. Suppose that under a specific operating condition of the energy storage cabinet, a deviation occurs between the real-time response patterns of the temperature and gas sensors to changes in the cooling fan's operating status and the feature set of the expected response pattern maintained under that condition. To avoid immediately adjusting the feature set of the expected response pattern due to instantaneous fluctuations, the system first performs a moving average on the deviation between the real-time response pattern and the expected response pattern. For example, a moving window containing 5 sampling periods can be set, and the average deviation value within these 5 periods can be calculated. If this average deviation value remains stably above a preset adjustment threshold (e.g., 0.5 degrees Celsius or 5 ppm) for three consecutive periods, it indicates that this deviation is not accidental but a persistent, systematic change. Only then will the system trigger an adjustment to the feature set of the expected response pattern corresponding to that operating condition. For example, a progressive weighted average method can be used to incorporate the current average deviation value into the feature set of the expected response pattern, slowly and steadily updating the baseline pattern to better adapt it to the actual operating conditions of the energy storage cabinet. This mechanism ensures that adjustments to the baseline model are made prudently and based on sufficient evidence, thereby maintaining the long-term stability and accuracy of the early warning system.

[0057] Optional, combined Figure 3 As shown, the steps for acquiring the real-time response mode of the temperature sensor and gas sensor to changes in the operating status of the cooling fan under actual working conditions during the daily operation of the energy storage cabinet include: A1 continuously monitors the start / stop and speed of the cooling fan, and generates a disturbance event when the speed change exceeds a preset threshold; A2, set a monitoring time window before and after the disturbance event, and record the temperature change curve of the temperature sensor and the concentration change curve of the gas sensor at a preset sampling frequency within the monitoring time window; A3, extract characteristic parameters from temperature change curves and concentration change curves; A4 aligns the characteristic parameters with the changes in operating status over time to form a temperature response mode and a gas response mode, which serve as a real-time response mode. A5 stores the real-time response mode.

[0058] "Continuous monitoring of the cooling fan's start / stop and speed" refers to the system continuously acquiring the cooling fan's operating data, such as its current status through the fan controller or dedicated sensors. A significant change in fan speed, such as from standstill to operation, or a large fluctuation in speed within a short period exceeding a preset threshold, is identified as a "disturbance event." This preset threshold can be set based on the fan model, the energy storage cabinet's size, and actual operating experience, aiming to capture fan behavior sufficient to cause changes in the cabinet's temperature and gas distribution.

[0059] "Setting a monitoring time window before and after the disturbance event" refers to establishing a specific data acquisition interval centered on the time of the disturbance event, extending forward and backward by a certain period. For example, a monitoring time window could be set from 5 seconds before the disturbance event to 30 seconds after it. Within this monitoring time window, "the temperature change curve from the temperature sensor and the concentration change curve from the gas sensor are recorded at a preset sampling frequency." The preset sampling frequency can be determined based on the sensor's response speed and the required data accuracy, such as once or multiple times per second, to ensure that subtle changes in temperature and gas concentration caused by fan disturbances are captured. The temperature change curve and the concentration change curve respectively reflect the dynamic response process of the temperature sensor and the gas sensor under the influence of fan disturbances.

[0060] "Extracting characteristic parameters from temperature and concentration change curves" refers to extracting key values ​​that characterize the response properties of the acquired curves. For example, peak values, trough values, rates of change, settling time, and integral area can be extracted. These characteristic parameters can quantify the intensity, speed, and duration of the sensor's response to fan disturbances.

[0061] "Aligning characteristic parameters with changes in operating status over time to form temperature and gas response modes, serving as real-time response modes" refers to associating extracted characteristic parameters with corresponding changes in the operating status of the cooling fan (e.g., fan start-up, acceleration, deceleration, and stop) on a time axis. This time alignment clearly establishes the causal relationship between changes in fan operating status and sensor response, thus forming a complete "real-time response mode" that reflects the sensor's response characteristics under the current actual working environment.

[0062] "Storing real-time response patterns" refers to saving the generated temperature and gas response patterns to a data storage medium for subsequent comparative analysis with a baseline response pattern. The stored data may include timestamps, fan operating status change types, extracted feature parameters, and other information.

[0063] Optionally, when the average deviation value remains stable above a preset adjustment threshold after a preset number of consecutive cycles, the step of triggering the adjustment of the feature set of the expected response mode corresponding to the operating scenario includes: Determine whether the average deviation value remains stable above the preset adjustment threshold after a preset number of consecutive cycles, and obtain the average deviation judgment result; Determine whether the overall operating status of the energy storage cabinet is stable and there are no early fire indicators, and obtain the operating status judgment result; When both the average deviation judgment result and the operation status judgment result are yes, the feature set of the expected response mode corresponding to the operation scenario is adjusted. A progressive weighted averaging method is used to incorporate the average deviation value into the feature set of the expected response pattern; Limit the adjustment range to a preset maximum value.

[0064] The process of determining whether the average deviation value remains stable above a preset adjustment threshold after a predetermined number of consecutive cycles aims to confirm that the observed deviation is not an instantaneous fluctuation or noise, but a systematic shift with a certain degree of persistence. This average deviation determination result is one of the necessary conditions for triggering adjustment. Furthermore, determining whether the overall operating status of the energy storage cabinet is stable and free of early fire indications is to introduce more comprehensive safety considerations. Stable overall operating status means that all key operating parameters of the energy storage cabinet (such as voltage, current, temperature in other areas, auxiliary gas concentration, etc.) are normal and without drastic fluctuations; the absence of early fire indications means that, apart from the sensor response mode currently being analyzed, no other independent sensor or system indications suggest a potential fire risk. Only when both the average deviation determination result and the operating status determination result are positive will the feature set of the expected response mode corresponding to the operating scenario be adjusted, ensuring that the adjustment is carried out under the premise of overall safety and stability of the energy storage cabinet.

[0065] Furthermore, this application employs a progressive weighted averaging method, integrating the average deviation value into the feature set of the expected response model. The progressive weighted averaging method means that when updating the feature set of the expected response model, the newly observed average deviation value does not immediately and completely replace the old feature values, but rather gradually influences the update of the feature set with a preset weight ratio. This makes the adjustment process smoother and less susceptible to the influence of short-term outliers. Simultaneously, limiting the magnitude of each adjustment to no more than a preset maximum value is to prevent excessively large and unreasonable jumps in the feature set due to a large average deviation value in certain extreme cases, further enhancing the robustness and stability of the system adjustment.

[0066] In some preferred embodiments, assuming that during normal operation of the energy storage cabinet, the average deviation of the response pattern of a certain temperature sensor to the start / stop of the cooling fan is consistently higher than a preset adjustment threshold for five consecutive cycles (e.g., each cycle is one day), the system will first determine the average deviation as "yes". Next, the system will further determine the overall operating status of the energy storage cabinet. For example, by monitoring the voltage and current of all battery modules, the temperature of other areas, and the concentration data of auxiliary gas sensors (such as CO, H2, etc.), it is found that these parameters are all within the normal fluctuation range, and there are no early fire indicators (e.g., CO concentration has not reached the warning threshold, and the battery internal resistance has not increased abnormally). At this time, the operating status determination result is also "yes". Since both determination results are "yes", the system will trigger an adjustment to the feature set of the expected response pattern for the temperature sensor under the corresponding operating scenario. During the adjustment process, the system will use a progressive weighted averaging method to incorporate the current continuous average deviation value into the original feature set with a small weight (e.g., the new deviation value accounts for 10% and the old feature set accounts for 90%), and ensure that the magnitude of this adjustment does not exceed the preset maximum value (e.g., not exceeding 0.5℃), so that the expected response mode can smoothly adapt to the new environment or sensor characteristics without introducing new uncertainties due to a one-time large adjustment.

[0067] Optionally, the steps to determine whether the overall operating status of the energy storage cabinet is stable and without early fire indicators, and to obtain the operating status determination result, include: Acquire real-time data of various auxiliary indicator parameters within the energy storage cabinet; Based on real-time data, calculate the fluctuation range of each auxiliary indicator parameter within a preset time window; The fluctuation range is compared with a preset range threshold to obtain the fluctuation range comparison result; When the fluctuation range comparison result indicates that the fluctuation range exceeds the preset range threshold, the real-time data of the corresponding auxiliary indicator parameter will be removed from the overall operation status judgment. When all auxiliary indicator parameters are in a stable state after the removal operation is completed, and there are no auxiliary indicators of early fire, determine whether the overall operating status of the energy storage cabinet is stable and there are no auxiliary indicators of early fire, and obtain the operating status judgment result.

[0068] Auxiliary indicator parameters can be understood as other key parameters used to monitor the operating status of the energy storage cabinet, besides temperature and gas sensors. These include battery module voltage, current, internal resistance, smoke sensor data, and humidity sensor data. Real-time data for these parameters is continuously acquired to provide a comprehensive view of the overall health of the energy storage cabinet. The preset time window refers to a period of time used to calculate the parameter fluctuation range. Its length can be set according to the actual application scenario and parameter characteristics; for example, it can be several seconds, several minutes, or longer. The fluctuation range refers to the difference between the maximum and minimum values ​​of a certain auxiliary indicator parameter within the preset time window, or its standard deviation or other statistical quantities, used to measure the stability of the parameter. The preset range threshold is a benchmark used to determine whether the parameter fluctuation is abnormal. When the fluctuation range exceeds this threshold, it indicates that the parameter may be abnormal or unstable.

[0069] Specifically, when the fluctuation range of a certain auxiliary indicator parameter exceeds a preset threshold, the real-time data of that parameter will be removed from the overall operating status assessment. This removal operation aims to prevent single or a few abnormally fluctuating auxiliary indicator parameters from interfering with or misleading the assessment of the overall operating status of the energy storage cabinet. For example, the voltage of a battery module may fluctuate briefly due to a local fault, but the energy storage cabinet as a whole may still be in a stable state. By removing abnormally fluctuating parameters, the overall stability of the energy storage cabinet can be assessed more accurately. After the removal operation is completed, if all remaining auxiliary indicator parameters are stable and there are no other early fire indicators (e.g., smoke sensors do not detect smoke, specific gas concentrations do not exceed limits), then the overall operating status of the energy storage cabinet can be determined to be stable and without early fire signs.

[0070] Optionally, when the fluctuation range comparison result indicates that the fluctuation range exceeds a preset range threshold, the step of removing the real-time data of the corresponding auxiliary indicator parameter from the overall operating status judgment includes: The average deviation value is compared with the preset group deviation threshold. When the average deviation value exceeds the group deviation threshold, it is determined that the corresponding auxiliary indicator sensor has a group offset. When a collective offset is detected in the corresponding auxiliary indicator sensor, the real-time data of the auxiliary indicator parameter corresponding to the auxiliary indicator sensor is retained as a potential early abnormal signal, and further anomaly analysis is triggered.

[0071] Specifically, the average deviation value refers to the deviation value obtained after applying a moving average to the real-time response pattern. It reflects the degree of persistent deviation of sensor data from the baseline or expected pattern. The group deviation threshold is a preset critical value used to determine whether multiple sensors simultaneously exhibit systematic shifts. When the average deviation value of a single auxiliary indicator parameter exceeds this group deviation threshold, the system determines that the corresponding auxiliary indicator sensor has a group shift. This group shift differs from random noise or occasional failures of a single sensor; it suggests that there may be some general, but not yet severe, anomaly within the energy storage cabinet. Once a group shift is determined, the real-time data of the corresponding auxiliary indicator parameter will not be simply discarded but will be retained as a potential early anomaly signal, further triggering a more in-depth anomaly analysis process.

[0072] In some preferred embodiments, it is assumed that multiple auxiliary temperature sensors are distributed near different battery modules within the energy storage cabinet. During normal operation, the real-time data fluctuation range of a certain auxiliary temperature sensor A1 slightly exceeds a preset range threshold, but its average deviation value, after moving average processing, does not exceed a preset group deviation threshold. In this case, the data may be considered as normal fluctuation or local interference and excluded from the overall operating status assessment. However, if the average deviation values ​​of multiple sensors, such as auxiliary temperature sensors A1, A2, and A3, consistently and stably exceed the preset group deviation threshold over a period of time, this solution can identify such group deviation even if the fluctuation amplitude of a single sensor is small. At this time, the system will determine that these auxiliary indicator sensors have a group deviation and retain their corresponding real-time data as potential early anomaly signals. Subsequently, the system will trigger further anomaly analysis, such as initiating infrared thermal imaging scans of these battery modules, or obtaining more detailed data such as voltage parameters, current parameters, and internal resistance parameters for cross-comparison and time-series analysis to more comprehensively assess potential fire risks. This approach avoids missing early warning signals due to simple exclusion, thus improving the accuracy and timeliness of early warnings.

[0073] Optionally, when a collective shift is detected in the corresponding auxiliary indicator sensor, the real-time data of the auxiliary indicator parameter corresponding to the auxiliary indicator sensor is retained as a potential early anomaly signal, and the steps to trigger further anomaly analysis include: Obtain real-time data on the voltage, current, temperature, and internal resistance parameters of the battery modules involved in the group offset; By cross-comparing and time-series analyzing voltage, current, temperature, and internal resistance parameters, preliminary results of anomaly type judgment are obtained. Based on the preliminary anomaly type assessment, an infrared thermal imaging scan of the battery module was initiated to obtain an image of the surface temperature distribution of the battery module. By analyzing the surface temperature distribution image and combining it with the gas data from the gas sensor, the final anomaly type determination result is obtained. Based on the final anomaly type determination result, output the anomaly type and handling suggestions.

[0074] Specifically, upon detecting a collective shift in the auxiliary indicator sensors, the system first acquires detailed operational data of the battery modules directly related to this shift. This data includes, but is not limited to, real-time data on voltage, current, temperature, and internal resistance parameters. Voltage and current parameters reflect the battery's charge / discharge state and energy flow; temperature parameters directly reflect the battery's thermal state and are crucial for early thermal runaway detection; and internal resistance parameters reflect the health of the battery's internal chemical reactions and structural changes. Acquiring these parameters aims to provide multi-dimensional and refined data support for subsequent anomaly analysis.

[0075] Furthermore, the acquired voltage, current, temperature, and internal resistance parameters are cross-referenced and subjected to time-series analysis. Cross-reference involves comparing different parameters horizontally; for example, observing whether a temperature increase in a battery module is accompanied by abnormal changes in current or internal resistance. Time-series analysis analyzes the changing trends of individual parameters over time to identify abnormal fluctuations, drifts, or abrupt changes. Through these two analytical methods, a preliminary assessment of the anomaly type can be made, such as overcharging, over-discharging, internal short circuits, or localized overheating.

[0076] As a preferred implementation, based on the preliminary anomaly type assessment, the system initiates an infrared thermal imaging scan targeting the specific battery module. Infrared thermal imaging technology can non-contactly acquire images of the surface temperature distribution of the battery module, visually displaying whether there are localized hot spots or abnormal temperature areas. This is crucial for accurately locating the initiation point of thermal runaway and assessing its propagation risk.

[0077] The acquired surface temperature distribution images will be further analyzed and combined with gas data from gas sensors. Gas data, especially changes in the concentration of combustible gases or specific decomposition gases, are important chemical indicators of impending battery thermal runaway or fire. By fusing the physical thermal distribution information provided by infrared thermal imaging with the chemical gas information provided by gas sensors, the nature and severity of the anomaly can be determined more comprehensively and accurately, thus obtaining the final anomaly type assessment result.

[0078] Finally, based on the final anomaly type determination, the system will output the specific anomaly type and corresponding handling suggestions. For example, if it is determined that a certain battery module has localized overheating accompanied by an increase in flammable gas concentration, the system may output the handling suggestion: "Battery module XX has a risk of thermal runaway; it is recommended to immediately isolate it and initiate cooling measures." This output not only informs the user of the anomaly's existence but also provides actionable guidance, helping to take timely intervention measures to prevent the occurrence or spread of fire.

[0079] In some preferred embodiments, a specific example is given below. Suppose that during the daily operation of the energy storage cabinet, the system, through monitoring auxiliary indicator parameters, discovers that multiple temperature sensors and gas sensors in a certain area have consistently shown average deviation values ​​exceeding preset thresholds. Furthermore, after assessing the operational status, it is confirmed that the overall operation of the energy storage cabinet is stable and there are no other early fire auxiliary indicators. At this point, it is determined that there is a cluster deviation.

[0080] According to the scheme of this application, the system will immediately initiate further anomaly analysis. First, the system will acquire the real-time voltage, current, temperature, and internal resistance parameters of the battery modules involved in the collective shift. For example, it may find that the temperature parameter of a certain battery module is continuously rising, while its internal resistance parameter also shows an abnormally increasing trend, and the voltage and current parameters show slight fluctuations. By cross-comparing and time-series analyzing these parameters, the system initially judges that the module may have early signs of internal local short circuit or thermal runaway.

[0081] Based on this initial assessment, the system initiates an infrared thermal imaging scan of the specific battery module. The infrared thermal image shows that the surface temperature of a particular battery cell within the module is significantly higher than the surrounding area, forming a localized hotspot. Simultaneously, data from the gas sensor indicates a slight increase in the concentration of combustible gases (such as CO and H2) within the energy storage cabinet. The system fuses and analyzes the surface temperature distribution image from the infrared thermal image with the gas data from the gas sensor, ultimately determining that cell number XX in the battery module is experiencing thermal runaway, and is in its early stages.

[0082] Ultimately, the system will output an anomaly type as "Early thermal runaway of cell XX in battery module XX" and provide handling recommendations: "Isolate battery module XX immediately, activate the local cooling system, and arrange for professional personnel to conduct on-site inspection and replacement." Through this series of detailed analyses and recommendations, energy storage cabinet managers can quickly take targeted measures to effectively control risks and prevent fire accidents.

[0083] This application also discloses a multi-sensor fusion-based early warning system for energy storage cabinet fires, used to perform early warning of energy storage cabinet fires based on multi-sensor fusion, combined with... Figure 4As shown, the early warning system for energy storage cabinet fires based on multi-sensor fusion 1 includes: The operating status acquisition module 11 is used to acquire temperature data from the temperature sensor, gas data from the gas sensor, and operating status information from the cooling fan inside the energy storage cabinet. The reference response establishment module 12 is used to monitor temperature and gas data in response to changes in the operating state of the cooling fan by controlling the changes in the operating state of the cooling fan during the ideal working period of normal operation of the energy storage cabinet, and to record the response patterns of the temperature sensor and the gas sensor to changes in the operating state of the cooling fan, so as to establish a reference response pattern of the temperature sensor and the gas sensor to changes in the operating state. The real-time response acquisition module 13 is used to acquire the real-time response mode of the temperature sensor and gas sensor to the changes in the operating status of the cooling fan under the actual working environment during the daily operation of the energy storage cabinet. The response mode comparison module 14 is used to compare the real-time response mode corresponding to the temperature sensor with the reference response mode to obtain the temperature response mode comparison result. When the temperature response mode comparison result indicates that there is a preset deviation, it is determined that the temperature sensor has an abnormal heat transfer path. The abnormal response determination module 15 is used to compare the real-time response mode corresponding to the gas sensor with the reference response mode to obtain the gas response mode comparison result. When the gas response mode comparison result indicates that there is a preset deviation, it is determined that the gas sensor has an abnormal response to the fan disturbance. The early warning output module 16 is used to output an early fire warning when it is simultaneously determined that the temperature sensor has an abnormal heat transfer path and the gas sensor has an abnormal response to fan disturbance.

[0084] Specifically, the operational status acquisition module can be a data acquisition unit configured to connect via wired or wireless means to temperature sensors, gas sensors, and the controller of the cooling fan within the energy storage cabinet, periodically acquiring the required temperature data, gas data, and fan operational status information. This module can employ various communication protocols, such as Modbus, CAN bus, or Ethernet, to transmit the acquired raw data to the central processing unit.

[0085] The benchmark response establishment module can be a data processing and model training unit. It is configured to trigger changes in the operating state of the cooling fan via control signals under specific ideal operating conditions, and simultaneously receive sensor data provided by the operating state acquisition module. This module constructs and stores a benchmark response pattern by performing time-series analysis and feature extraction on this data, such as calculating the time response curve, rate of change, and settling time of temperature or gas concentration as a function of fan start / stop or speed changes. The benchmark response pattern establishment process has already been described in the above embodiments and will not be repeated here.

[0086] The real-time response acquisition module can be a real-time data monitoring and feature extraction unit. It is configured to continuously monitor the operating status of the cooling fan and trigger real-time data acquisition from the temperature and gas sensors when the fan status changes, such as starting / stopping or the speed change exceeding a preset threshold. This module performs feature extraction on the acquired real-time data similar to that of the baseline response establishment module to generate a real-time response pattern. The process of acquiring the real-time response pattern has already been described in the above embodiments and will not be repeated here.

[0087] The response pattern comparison module can be a pattern matching and deviation calculation unit, configured to receive real-time response patterns provided by the real-time response acquisition module and read the corresponding baseline response patterns from the storage unit. This module compares the two patterns using an algorithm (e.g., Euclidean distance, correlation coefficient, or dynamic time warping algorithm), calculates the temperature response pattern comparison result, and determines whether a preset deviation exists. The specific method for temperature response pattern comparison has been described in the above embodiments and will not be repeated here.

[0088] The abnormal response determination module can be a logical judgment unit configured to work in parallel with the response mode comparison module. It receives the real-time response mode and the reference response mode corresponding to the gas sensor and compares them. This module also calculates the gas response mode comparison result using an algorithm and determines whether there is a preset deviation, thereby determining that the gas sensor has an abnormal response to fan disturbances. The specific method for gas response mode comparison has been described in the above embodiments and will not be repeated here.

[0089] The early warning output module can be an alarm triggering and information dissemination unit, configured to receive the judgment results from the response mode comparison module and the abnormal response judgment module. When both modules simultaneously output judgment results indicating an anomaly, this module is triggered, generating an early fire warning signal and issuing warning information to relevant personnel through various means such as audible and visual alarms, SMS, email, or online platforms.

[0090] The above are merely embodiments of this application and are not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for early warning of fires in energy storage cabinets based on multi-sensor fusion, characterized in that, include: Acquire temperature data from the temperature sensor inside the energy storage cabinet, gas data from the gas sensor, and operating status information from the cooling fan; During the ideal working period of the energy storage cabinet, by controlling the changes in the operating state of the cooling fan, temperature data and gas data in response to the changes in the operating state are monitored, and the response patterns of the temperature sensor and gas sensor to the changes in the operating state of the cooling fan are recorded, so as to establish the reference response patterns of the temperature sensor and gas sensor to the changes in the operating state. During the daily operation of the energy storage cabinet, the real-time response mode of the temperature sensor and gas sensor to the changes in the operating status of the cooling fan in the actual working environment is acquired. The real-time response mode corresponding to the temperature sensor is compared with the reference response mode to obtain the temperature response mode comparison result. When the temperature response mode comparison result indicates that there is a preset deviation, it is determined that the temperature sensor has an abnormal heat transfer path. The real-time response mode corresponding to the gas sensor is compared with the reference response mode to obtain the gas response mode comparison result. When the gas response mode comparison result indicates that there is a preset deviation, it is determined that the gas sensor has an abnormal response to fan disturbance. When both the temperature sensor and the gas sensor are simultaneously detected to have an abnormal heat transfer path, an early fire warning is output.

2. The method for early warning of fire in energy storage cabinets based on multi-sensor fusion according to claim 1, characterized in that, The steps of establishing a reference response pattern for the temperature and gas sensors to the changes in the operating state of the cooling fan during the ideal operating period of the energy storage cabinet include: Obtain the external environmental parameters and internal battery load parameters of the energy storage cabinet; During the ideal working period of the energy storage cabinet, by controlling the change in the operating state of the cooling fan, temperature data and gas data in response to the change in the operating state are monitored, the response patterns of the temperature sensor and gas sensor to the change in the operating state of the cooling fan are recorded, and the influence characteristics of the external environmental parameters and the internal battery load parameters on the response patterns are characterized according to the external environmental parameters and the internal battery load parameters. Based on the external environmental parameters, the internal battery load parameters, and the influencing characteristics, calculate the dynamic expected response patterns of the temperature sensor and the gas sensor to changes in operating status. The real-time response mode corresponding to the temperature sensor is compared with the dynamic expected response mode. When there is a preset deviation between the real-time response mode and the dynamic expected response mode, it is determined that the temperature sensor has an abnormal heat transfer path. The real-time response mode corresponding to the gas sensor is compared with the dynamic expected response mode. When there is a preset deviation between the real-time response mode and the dynamic expected response mode, it is determined that the gas sensor has an abnormal response to fan disturbance. When both the temperature sensor and the gas sensor are found to have abnormal heat transfer paths, an early fire warning is output.

3. The method for early warning of fire in energy storage cabinets based on multi-sensor fusion according to claim 2, characterized in that, The steps of controlling the operating state changes of the cooling fan, monitoring temperature and gas data in response to these changes, recording the response patterns of the temperature and gas sensors to the cooling fan's operating state changes, and characterizing the influence of the external environmental parameters and internal battery load parameters on the response patterns based on the external environmental parameters and internal battery load parameters during the ideal operating period of the energy storage cabinet include: During the ideal operating period of the energy storage cabinet, the operating conditions of the energy storage cabinet are identified based on the external environmental parameters and the internal battery load parameters. For each of the aforementioned operating scenarios, maintain a feature set of the expected response patterns of the temperature sensor and gas sensor to changes in operating state under that operating scenario; The feature set of the real-time response mode is compared with the feature set of the expected response mode under the corresponding operating scenario to obtain the expected comparison result; Based on the expected comparison results, the feature set of the expected response mode corresponding to the operating scenario is adjusted.

4. The method for early warning of fire in energy storage cabinets based on multi-sensor fusion according to claim 3, characterized in that, The step of adjusting the feature set of the expected response mode corresponding to the running scenario based on the expected comparison results includes: The real-time response mode is subjected to a moving average process to obtain the average deviation value; When the average deviation value remains stable above the preset adjustment threshold after a preset number of consecutive cycles, the feature set of the expected response mode corresponding to the operating scenario is adjusted.

5. The method for early warning of fire in energy storage cabinets based on multi-sensor fusion according to claim 1, characterized in that, The steps for acquiring the real-time response mode of the temperature sensor and gas sensor to changes in the operating status of the cooling fan under the actual working environment during the daily operation of the energy storage cabinet include: Continuously monitor the start / stop and speed of the cooling fan, and generate a disturbance event when the speed change exceeds a preset threshold; Set a monitoring time window before and after the disturbance event, and record the temperature change curve of the temperature sensor and the concentration change curve of the gas sensor at a preset sampling frequency within the monitoring time window; Extract characteristic parameters from temperature change curves and concentration change curves; The characteristic parameters are time-aligned with changes in operating status to form a temperature response mode and a gas response mode, which serve as a real-time response mode. The real-time response mode is stored.

6. The method for early warning of fire in energy storage cabinets based on multi-sensor fusion according to claim 4, characterized in that, The step of triggering the adjustment of the feature set of the expected response mode corresponding to the operating scenario when the average deviation value remains stable above the preset adjustment threshold after a preset number of consecutive cycles includes: Determine whether the average deviation value remains stable above a preset adjustment threshold after a preset number of consecutive cycles to obtain the average deviation determination result; Determine whether the overall operating status of the energy storage cabinet is stable and there are no early fire indicators, and obtain the operating status judgment result; When both the average deviation judgment result and the running status judgment result are yes, the feature set of the expected response mode corresponding to the running situation is adjusted. A progressive weighted averaging method is used to incorporate the average deviation value into the feature set of the expected response pattern; Limit the magnitude of each adjustment to no more than the preset maximum value.

7. The method for early warning of fire in energy storage cabinets based on multi-sensor fusion according to claim 6, characterized in that, The steps for determining whether the overall operating status of the energy storage cabinet is stable and whether there are any early fire indicators, and obtaining the operating status determination result, include: Acquire real-time data of various auxiliary indicator parameters within the energy storage cabinet; Based on the real-time data, the fluctuation range of each auxiliary indicator parameter within a preset time window is calculated; The fluctuation range is compared with a preset range threshold to obtain the fluctuation range comparison result; When the fluctuation range comparison result indicates that the fluctuation range exceeds the preset range threshold, the real-time data of the corresponding auxiliary indication parameter will be removed from the overall operation status judgment. When all the auxiliary indicator parameters are in a stable state after the removal operation is completed, and there are no auxiliary indicators of early fire, it is determined whether the overall operating status of the energy storage cabinet is stable and there are no auxiliary indicators of early fire, and the operating status judgment result is obtained.

8. The method for early warning of fire in energy storage cabinets based on multi-sensor fusion according to claim 7, characterized in that, The step of removing the real-time data of the corresponding auxiliary indicator parameter from the overall operating status judgment when the fluctuation range comparison result indicates that the fluctuation range exceeds a preset range threshold includes: The average deviation value is compared with a preset group deviation threshold. When the average deviation value exceeds the group deviation threshold, it is determined that the corresponding auxiliary indicator sensor has a group offset. When a collective offset is detected in the corresponding auxiliary indicator sensor, the real-time data of the auxiliary indicator parameter corresponding to the auxiliary indicator sensor is retained as a potential early abnormal signal, and further anomaly analysis is triggered.

9. The method for early warning of fire in energy storage cabinets based on multi-sensor fusion according to claim 8, characterized in that, The step of retaining the real-time data of the auxiliary indication parameters corresponding to the auxiliary indication sensor as a potential early anomaly signal and triggering further anomaly analysis when it is determined that there is a collective offset in the corresponding auxiliary indication sensor includes: Real-time data of voltage, current, temperature, and internal resistance parameters of the battery modules involved in the collective offset are obtained; By cross-comparing and time-series analyzing voltage, current, temperature, and internal resistance parameters, preliminary results of anomaly type judgment are obtained. Based on the preliminary anomaly type determination result, an infrared thermal imaging scan of the battery module is initiated to obtain an image of the surface temperature distribution of the battery module. By analyzing the surface temperature distribution image and combining it with the gas data from the gas sensor, the final anomaly type determination result is obtained. Based on the final anomaly type determination result, output the anomaly type and handling suggestions.

10. A multi-sensor fusion-based early warning system for energy storage cabinet fires, used to perform early warning of energy storage cabinet fires based on multi-sensor fusion, characterized in that, include: The operation status acquisition module is used to acquire temperature data from the temperature sensor, gas data from the gas sensor, and operation status information of the cooling fan inside the energy storage cabinet. The reference response establishment module is used to monitor temperature and gas data in response to changes in the operating state of the cooling fan by controlling the changes in the operating state of the cooling fan during the ideal working period of the energy storage cabinet, and to record the response patterns of the temperature sensor and the gas sensor to the changes in the operating state of the cooling fan, so as to establish a reference response pattern of the temperature sensor and the gas sensor to the changes in the operating state. The real-time response acquisition module is used to acquire the real-time response mode of the temperature sensor and gas sensor to the changes in the operating status of the cooling fan in the actual working environment during the daily operation of the energy storage cabinet. The response mode comparison module is used to compare the real-time response mode corresponding to the temperature sensor with the reference response mode to obtain the temperature response mode comparison result. When the temperature response mode comparison result indicates that there is a preset deviation, it is determined that the temperature sensor has an abnormal heat transfer path. An abnormal response determination module is used to compare the real-time response mode corresponding to the gas sensor with the reference response mode to obtain a gas response mode comparison result. When the gas response mode comparison result indicates that there is a preset deviation, it is determined that the gas sensor has an abnormal response to fan disturbance. The early warning output module is used to output an early fire warning when both the temperature sensor and the gas sensor are found to have abnormal heat transfer paths and abnormal responses to fan disturbances.

Images