Quantum motion driven charging pile thermal runaway active protection method and system

Abstract

This invention discloses a quantum-sensory driven active protection method and system for thermal runaway in charging piles, relating to the field of charging pile protection technology. It includes the following steps: S1, real-time acquisition of quantum thermal runaway data, data preprocessing, quantification of thermal risk sensitivity, and construction of a thermal runaway risk dataset; S2, construction of a thermal runaway spatiotemporal matrix, and based on this matrix, construction of a principal component extraction model for thermal runaway risk, combined with the spatiotemporal matrix and the principal component extraction model, to construct a risk trend prediction model; S3, comprehensive judgment of thermal runaway risk, achieving early detection and active protection; S4, determination of safety thresholds, implementation of graded protection responses for thermal runaway, and closed-loop optimization of protection capabilities. This solves the problem that existing traditional charging pile thermal runaway monitoring technologies struggle to detect and control potential thermal runaway hazards in a timely and accurate manner, thus affecting the operational safety of charging piles.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of charging pile protection, in particular to a quantum body-sensing driven charging pile thermal runaway active protection method and system. BACKGROUND

[0002] At present, with the continuous expansion of the new energy vehicle market and the rapid construction of charging infrastructure, electric vehicle charging piles, as a key link of intelligent transportation and green energy systems, have been widely used in urban, park, highway and other multi-scene scenarios. With the continuous improvement of high-power charging and high-energy efficiency demand, advanced temperature control solutions such as liquid cooling are gradually becoming a new trend of industry development, effectively supporting the safe, stable and efficient operation of charging equipment.

[0003] For example, the patent for invention with the publication number CN120180913A discloses a charging pile intelligent heat dissipation method, device and medium based on heat conduction line optimization, belonging to the technical field of charging pile heat dissipation. The technical problem to be solved by the application is how to avoid the defects of high thermal resistance, response lag and large energy consumption in the traditional charging pile heat dissipation scheme, and improve the heat dissipation efficiency and operation reliability of the charging pile. The technical solution adopted is to optimize the heat conduction structure, deploy thin film thermocouples and build a three-dimensional temperature field model through infrared thermal imaging, use a fuzzy PID and model predictive control hybrid algorithm to establish a thermal resistance-flow-power transfer function model, and based on LSTM neural network analysis of historical charging data, predict the power peak value 300ms in advance and pre-start the heat dissipation equipment.

[0004] For example, the patent for invention with the publication number CN118013427B discloses a liquid cooling heat dissipation automobile charging pile and a warning method. The method comprises: obtaining historical liquid leakage fault data of the liquid cooling heat dissipation automobile charging pile, and constructing a liquid leakage fault evolution chain of the liquid cooling heat dissipation automobile charging pile according to the historical liquid leakage fault data; monitoring the liquid cooling system state data of the liquid cooling heat dissipation automobile charging pile to obtain an initial liquid cooling system state data set, and preprocessing the initial liquid cooling system state data set through a subspace recognition algorithm to obtain a target liquid cooling system state data set; extracting fault features and constructing a feature matrix based on the liquid leakage fault evolution chain to obtain a liquid cooling system fault feature matrix; inputting the liquid cooling system fault feature matrix into a pre-set multi-classifier fusion decision model for liquid leakage fault diagnosis to obtain a liquid leakage fault diagnosis result. The application improves the liquid leakage warning accuracy of the liquid cooling heat dissipation automobile charging pile.

[0005] However, in the process of implementing the technical scheme of the application, the application found that the above-mentioned technology at least has the following technical problems:

[0006] The existing charging pile thermal runaway monitoring technology has the problems of lagging early warning, weak anti-interference ability, poor adaptability to different batteries, and low data utilization rate, which makes it difficult to discover and prevent thermal runaway hazards in time and accurately, and it is difficult to effectively ensure the safe operation of the charging pile.

[0007] Therefore, in view of the above problems, there is an urgent need for a quantum body sensing driven charging pile thermal runaway active protection method and system. SUMMARY

[0008] Technical problems to be solved

[0009] In view of the deficiencies of the prior art, the quantum body sensing driven charging pile thermal runaway active protection method and system are provided, which solves the problem that the existing traditional charging pile thermal runaway monitoring technology is difficult to discover and prevent thermal runaway hazards in time and accurately, thereby affecting the safe operation of the charging pile.

[0010] Technical scheme

[0011] To achieve the above purpose, the following technical scheme is used: a quantum body sensing driven charging pile thermal runaway active protection method, comprising the following steps: S1, real-time acquisition of quantum thermal runaway data, data preprocessing of the quantum thermal runaway data, quantification of thermal risk sensitivity based on the preprocessed quantum thermal runaway data, and construction of a thermal runaway risk data set; S2, using the thermal runaway risk data set to construct a thermal runaway space-time matrix, constructing a thermal runaway risk principal component extraction model based on the thermal runaway space-time matrix, combining the thermal runaway space-time matrix and the thermal runaway risk principal component extraction model, and constructing a risk trend prediction model; S3, fusion of the thermal runaway risk principal component extraction model and the risk trend prediction model, comprehensive judgment of the thermal runaway risk, early discovery and active protection of the thermal runaway risk according to the thermal runaway risk judgment result; S4, judging the safety threshold by comprehensively considering the thermal runaway data, the risk trend prediction model and the thermal runaway risk, and performing graded protection response of the charging pile thermal runaway according to the safety threshold judgment result, and realizing closed-loop optimization of the protection ability.

[0012] Further, the specific process of collecting quantum thermal runaway data in real time and pre-processing the quantum thermal runaway data is as follows: quantum sensors are deployed at key parts of the battery pack and charging pile to collect quantum thermal runaway data in real time. The quantum thermal runaway data includes average temperature of measuring points, thermal runaway gas concentration, and current data. The average value of the current data is calculated based on a sliding time window, and the difference between the current data and the average value of the current data in the sliding time window is obtained to get the current deviation. The original quantum thermal runaway data is aligned in time by a unified timestamp, and abnormal values are removed and short-time missing data is completed by a sliding window statistical method. Moving average and Kalman filtering algorithm are used for signal smoothing and noise suppression. The quantum thermal runaway data is dimensionless and normalized. The quantum thermal runaway monitoring database is constructed, the quantum thermal runaway data is attached with a timestamp, and is stored in the quantum thermal runaway monitoring database in groups according to a fixed sliding time window.

[0013] Further, the specific process of quantifying thermal risk sensitivity based on the pre-processed quantum thermal runaway data is as follows: based on the sliding time window, the average temperature of the measuring points is obtained, and the maximum temperature value is selected. The difference between the maximum temperature value in the current sliding time window and the maximum temperature value in the previous sliding time window is calculated to obtain the maximum temperature change rate. The standard deviation of the average temperature of all measuring points in the current sliding time window is calculated to obtain the temperature standard deviation, and the concentration of all thermal runaway gases is obtained to calculate the average concentration of thermal runaway gases. The maximum temperature change rate is divided by the sum of the temperature standard deviation and a very small positive number to obtain the temperature rise sensitivity value. The current thermal runaway gas concentration is divided by the average concentration of thermal runaway gases, and the natural logarithm of the result is added to a constant one to obtain the gas sensitive gain value. The temperature rise sensitivity value is multiplied by the gas sensitive gain value to obtain the thermal control sensitivity value.

[0014] Further, the specific process of constructing the thermal runaway risk data set is as follows: the thermal control sensitivity value is written into the quantum thermal runaway monitoring database, and the thermal control sensitivity value and the corresponding quantum thermal runaway data are combined to construct the thermal runaway risk data set. If the thermal control sensitivity value is higher than the thermal control sensitivity threshold for a continuous window exceeding the maximum allowable threshold, the sampling frequency is increased, and the verification of the thermal runaway data is strengthened, and the process of spatiotemporal feature fusion and trend prediction is entered.

[0015] Further, the thermal runaway risk dataset is used to construct a thermal runaway spatiotemporal matrix, and a thermal runaway risk principal component extraction model is constructed based on the thermal runaway spatiotemporal matrix. The specific process of constructing the risk trend prediction model in combination with the thermal runaway spatiotemporal matrix and the thermal runaway risk principal component extraction model is as follows: the intelligent algorithm controller receives the thermal runaway risk dataset, performs spatiotemporal feature fusion and trend prediction, constructs a thermal runaway spatiotemporal matrix by using the average temperature sequence of the measuring point, the thermal runaway gas concentration sequence, the current data sequence, the current deviation sequence and the corresponding thermal control sensitivity value sequence, uses the spatiotemporal attention network algorithm for training by taking the thermal runaway spatiotemporal matrix as the input, performs spatiotemporal alignment, feature extraction and principal component dimension reduction, constructs a thermal runaway risk principal component extraction model, and obtains thermal field main feature components, gas main trend components and implicit risk features; the thermal runaway spatiotemporal matrix, the thermal field main feature components, the gas main trend components and the implicit risk features are taken as the input, and the long short-term memory time series network algorithm is used for supervised training, the loss function is jointly optimized by using multi-objective regression, a risk trend prediction model is constructed, the evolution trend of the key risk indicators in the future time window is predicted, and the thermal field main feature component prediction value, the gas main trend component prediction value, the current deviation prediction value and the maximum measuring point average temperature prediction value are output.

[0016] Further, the thermal runaway risk principal component extraction model and the risk trend prediction model are fused, and the specific process of comprehensively judging the thermal runaway risk is as follows: the current thermal field main feature component and the thermal field main feature component prediction value are obtained, the natural logarithm of the ratio of the current thermal field main feature component to the thermal field main feature component prediction value is calculated, and the absolute value is taken to obtain a thermal field main feature mutation term; the current current deviation and the current deviation prediction value are obtained, the square root of the ratio of the absolute values of the current current deviation and the current deviation prediction value is taken to obtain a key current anomaly term; based on a sliding time window, the gas main trend component is obtained, and the standard deviation is calculated to obtain a gas main trend component standard deviation, the gas main trend component prediction value is obtained, the difference between the gas main trend component prediction value and the current gas main trend component is divided by the gas main trend component standard deviation, and the arctangent function is operated to obtain a gas concentration dramatic change term; the thermal field main feature mutation term, the key current anomaly term and the gas concentration dramatic change term are added to obtain a multi-modal main risk judgment value.

[0017] Furthermore, the specific process for early detection and proactive protection of thermal runaway risk based on the thermal runaway risk assessment results is as follows: The multimodal main risk assessment value is written into the quantum thermal runaway monitoring database in real time and compared with the risk threshold; when the multimodal main risk assessment value is less than the risk threshold, the multimodal main risk assessment value is continuously monitored, and the sampling frequency is periodically increased to capture potential anomalies; when the multimodal main risk assessment value is greater than or equal to the risk threshold, it is determined to be a warning condition. The multimodal main risk assessment value, thermal runaway data, thermal control sensitivity value, and the predicted average temperature of the maximum measurement point for the corresponding time window are packaged and marked as a warning. The data acquisition frequency is temporarily increased, and the circulation pump and electric valve are turned on to pre-cool the charging pile. At the same time, the fan is started in advance to discharge the thermal runaway gas to the safe zone. The gas emission channel is detected, and the safety threshold assessment process is initiated.

[0018] Furthermore, the specific process for determining the safety threshold by integrating thermal runaway data, risk trend prediction models, and thermal runaway risk is as follows: Receive the multimodal main risk judgment value, thermal runaway data, thermal control sensitivity value, and the predicted average temperature of the maximum measuring point within the warning operating condition time window, and conduct a safety threshold assessment: Calculate the second derivative of the average temperature of the measuring point with time, take the square root of the absolute value to obtain the thermal field fluctuation level value; Based on the warning operating condition time window, calculate the mean thermal runaway gas concentration and the standard deviation of the thermal runaway gas concentration, divide the difference between the current thermal runaway gas concentration and the mean thermal runaway gas concentration by the sum of the standard deviation of the thermal runaway gas concentration and the smallest positive value, and perform a square calculation; perform a natural logarithmic operation on the sum of the square calculation result and a constant to obtain the gas anomaly gain value; Based on the warning operating condition time window, integrate the thermal control sensitivity value within the time window and divide it by the window length to obtain the anomaly integral mean; Add the thermal field fluctuation level value, the gas anomaly gain value, the anomaly integral mean, and the corresponding predicted average temperature of the maximum measuring point to obtain the safety threshold value.

[0019] Furthermore, based on the safety threshold judgment results, the specific process of implementing graded protection response for thermal runaway of charging piles and achieving closed-loop optimization of protection capabilities is as follows: real-time comparison of safety threshold values. With multi-level security threshold and ;when < At that time, it outputs continuous monitoring commands, maintains the normal sampling frequency, adjusts the liquid cooling equipment and fan equipment, performs only low-power preheating, and maintains a slight positive pressure in the charging pile cabinet, while periodically pushing healthy samples to the quantum thermal runaway monitoring database; when ≥ and < When an abnormal condition occurs, output current limiting and warning commands, reduce the charging and discharging current to below the safe threshold, strengthen the operation of liquid cooling and gas extraction equipment, increase coolant flow and air velocity for abnormal points, and adjust the gas ventilation and filtration strategy; upload all abnormal operating condition data to the operation and maintenance terminal, and conduct remote verification through pop-up reminders on the operation and maintenance terminal; when ≥ When power outage and alarm commands are output, the main controller urgently cuts off relevant circuits, executes power outage protection, and the on-site buzzer and indicator lights strongly indicate the situation. The edge controller links gas extraction, cabinet positive pressure protection, and forced liquid cooling equipment to their highest levels. An abnormal work order is generated and a hazard alarm is pushed to the operation and maintenance end. Periodic statistical analysis is continuously performed on the quantum thermal runaway data, thermal control sensitivity value, multimodal main risk judgment value, safety threshold value, and corresponding protection effect throughout the entire process. Combined with healthy samples, the sliding adaptive threshold algorithm and Bayesian optimization algorithm are used to update and optimize the parameters of each algorithm, the thermal runaway risk principal component extraction model, the risk trend prediction model, the risk threshold, and the multi-level safety threshold. After each round of analysis and optimization, optimization suggestions and the latest updated strategies are pushed to the operation and maintenance end.

[0020] The second aspect of this invention provides a quantum haptic-driven active protection system for thermal runaway in charging piles, comprising: a quantum haptic-driven multimodal monitoring and processing module, used to collect quantum thermal runaway data in real time, preprocess the quantum thermal runaway data, quantify the thermal risk sensitivity based on the preprocessed quantum thermal runaway data, and construct a thermal runaway risk dataset; and a spatiotemporal feature fusion and trend prediction module, used to construct a thermal runaway spatiotemporal matrix using the thermal runaway risk dataset, construct a thermal runaway risk principal component extraction model based on the thermal runaway spatiotemporal matrix, and combine the thermal runaway spatiotemporal matrix with thermal... The system includes a principal component extraction model for thermal runaway risk, which is used to construct a risk trend prediction model. A multimodal risk discrimination and intervention module is used to integrate the principal component extraction model and the risk trend prediction model to comprehensively assess thermal runaway risk and achieve early detection and proactive protection based on the assessment results. A protection threshold judgment, execution, and optimization module is used to comprehensively analyze thermal runaway data, the risk trend prediction model, and the thermal runaway risk to determine safety thresholds. Based on the safety threshold judgment results, a graded protection response for charging pile thermal runaway is implemented, and closed-loop optimization of protection capabilities is achieved.

[0021] Beneficial effects

[0022] The present invention has the following beneficial effects:

[0023] (1) This invention achieves high-precision, real-time acquisition and intelligent processing of multi-source data such as temperature, gas concentration and current by deploying quantum-level multimodal sensors in key parts of charging piles and battery packs. It also constructs a thermal control sensitivity value, which can quantify the sensitivity to microscopic risks of thermal runaway, greatly improving the ability to actively perceive complex anomalies.

[0024] (2) This invention utilizes a spatiotemporal attention network and a long short-term memory network to achieve multimodal feature fusion, principal component extraction, and future risk trend prediction of the spatiotemporal matrix of thermal runaway. It not only accurately quantifies the current risk status but also intelligently extrapolates the future risk evolution trend, enabling advanced detection and precise graded early warning of thermal runaway risks, effectively reducing sudden failures and safety hazards.

[0025] (3) This invention achieves a graded active protection response mechanism by constructing multimodal main risk judgment values ​​and safety threshold values. It can implement various execution measures such as current limiting, forced liquid cooling, gas extraction, localized cooling, and power outage according to the risk level, ensuring that the equipment can obtain the optimal protection response under different risk conditions. Compared with traditional passive strategies, this significantly improves the safety and operational resilience of the charging pile system.

[0026] (4) This invention achieves dynamic self-learning and optimization of model parameters and criterion thresholds by collecting data and protection effects in a closed loop throughout the entire process, performing periodic cloud-based statistics and intelligent analysis, and combining sliding adaptive thresholds and Bayesian optimization algorithms. The update results are pushed to the operation and maintenance end in each round, continuously improving the accuracy of risk prediction and the level of intelligent protection decision-making, and realizing the full life cycle self-evolution and high reliability guarantee of the active protection capability against thermal runaway of charging piles.

[0027] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0028] Figure 1 Flowchart of an active protection method for thermal runaway of charging piles driven by quantum haptic feedback;

[0029] Figure 2 Structural diagram of a quantum-sensory-driven active protection system for thermal runaway in charging piles;

[0030] Figure 3 Schematic diagram of a quantum-sensory driven active protection system for thermal runaway in charging piles;

[0031] Figure 4 A flowchart illustrating the monitoring process of quantum sensors;

[0032] Figure 5 A flowchart illustrating the graded protection response process for thermal runaway in charging piles;

[0033] Figure 6 This is a synergistic distribution map of thermal runaway risk based on temperature rise sensitivity and gas gain.

[0034] In the diagram, 1 is the charging pile; 2 is the intelligent algorithm controller; 3 is the coolant reservoir; 4 is the condenser; 5 is the circulation pump; 6 is the electric valve; and 7 is the quantum sensor array. Detailed Implementation

[0035] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. As those skilled in the art will understand, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0036] Please see Figures 1-6 This invention provides a technical solution: a quantum-sensory driven active protection method and system for thermal runaway of charging piles, such as... Figure 1 As shown, the process includes the following steps: S1, real-time acquisition of quantum thermal runaway data, data preprocessing of the quantum thermal runaway data, quantification of thermal risk sensitivity based on the preprocessed quantum thermal runaway data, and construction of a thermal runaway risk dataset; S2, construction of a thermal runaway spatiotemporal matrix using the thermal runaway risk dataset, construction of a thermal runaway risk principal component extraction model based on the thermal runaway spatiotemporal matrix, and construction of a risk trend prediction model by combining the thermal runaway spatiotemporal matrix and the thermal runaway risk principal component extraction model; S3, fusion of the thermal runaway risk principal component extraction model and the risk trend prediction model, comprehensive judgment of thermal runaway risk, and early detection and proactive protection of thermal runaway risk based on the judgment results; S4, comprehensive assessment of thermal runaway data, risk trend prediction model, and thermal runaway risk to determine a safety threshold, and graded protection response for thermal runaway of charging pile 1 based on the safety threshold judgment results, and closed-loop optimization of protection capabilities.

[0037] Specifically, the real-time acquisition of quantum thermal runaway data and the data preprocessing process are as follows: A quantum sensor group 7, including a graphene quantum dot temperature measurement array, a quantum conductivity gas sensor, and a high-precision current sensor, is deployed at key locations in the battery pack and charging pile 1. This group offers superior detection accuracy and response speed compared to conventional sensors, enabling precise acquisition under high heat flux density and complex electromagnetic environments. Real-time acquisition of quantum thermal runaway data includes: average temperature at measurement points, thermal runaway gas concentration, and current data. The average temperature at each measurement point is the average value of the temperatures at all measurement points. The thermal runaway gas concentration is the real-time value of typical thermal runaway gases such as hydrogen and carbon monoxide. The current data includes the charging and discharging current of each channel and the current of local branches. Simultaneously, the mean current data is calculated based on a sliding time window, and the difference between the current data at each moment within the sliding time window and the mean current data is calculated to obtain the current deviation. The sliding time window refers to a data window that is continuously updated with a fixed duration, adaptively updated according to actual operating conditions, and used for real-time dynamic analysis of data trends and fluctuations. The raw quantum thermal runaway data is synchronized by aligning it with a unified timestamp, meaning that the multi-source data streams are corrected according to the same time base to ensure that the data collected by all sensors can be accurately matched to the same moment, improving the accuracy of subsequent data fusion and modeling. Outliers are removed and short-term missing values ​​are filled in using a sliding window statistical method, specifically using the Z-score method for outlier removal and interpolation for short-term missing value filling. Moving average and Kalman filtering algorithms are used for signal smoothing and noise suppression. The moving average algorithm is suitable for removing short-term high-frequency noise, while the Kalman filter is more effective in dynamic state estimation, and can adaptively filter noise and predict missing values. The quantum thermal runaway data is dimensionless normalized and standardized to eliminate data dimensions and achieve a unified feature distribution, which facilitates subsequent machine learning modeling. A quantum thermal runaway monitoring database is constructed, supporting high-frequency writing and multi-dimensional data querying, ensuring full-process data traceability and efficient retrieval. The quantum thermal runaway data is timestamped and stored in the quantum thermal runaway monitoring database in groups according to a fixed sliding time window, realizing segmented management and efficient retrieval of historical data.

[0038] like Figure 3The diagram shows the principle of a quantum-sensing driven active protection system for thermal runaway in charging piles. Charging pile 1 is the main component, integrating a quantum sensor group 7 to collect real-time multi-modal data on temperature, gas concentration, and current in key areas. The intelligent algorithm controller 2 preprocesses, intelligently analyzes, models, and calculates criteria for the collected quantum thermal runaway data, enabling quantitative risk sensitivity and early warning decisions. A high-efficiency liquid cooling circulation regulation system, consisting of a coolant reservoir 3, condenser 4, circulation pump 5, and electric valve 6, automatically adjusts the coolant flow and gas emission path through algorithmic linkage control when an anomaly is detected, actively cooling and protecting charging pile 1. The entire system, through deep integration of intelligent algorithms and physical cooling devices, achieves early detection, graded protection, and closed-loop optimization of thermal runaway in charging pile 1, ensuring the safe and highly reliable operation of the charging equipment under complex conditions.

[0039] like Figure 4 The diagram illustrates the workflow of the quantum sensor monitoring process. First, after charging preparation and physical connection detection are completed, the quantum sensor group monitoring module is activated to achieve highly sensitive quantum monitoring of the charging pile 1 device status. Through the linear response relationship between quantum dot fluorescence intensity and temperature, temperature can be acquired in real time, capturing the local temperature rise signal in the early stages of thermal runaway and further accurately locating the abnormal heat source. Simultaneously, the quantum sensor can also convert and detect key gas signals through molecular state changes induced by molecular adsorption, ultimately achieving simultaneous detection of characteristic gases of thermal runaway and acquiring the concentration of thermal runaway gases. This process realizes multi-dimensional monitoring from physical detection and quantum response to heat source location and synchronous gas sensing, providing a solid data foundation and technical support for early warning and efficient active protection against thermal runaway of charging pile 1.

[0040] In this implementation plan, quantum motion-sensing multimodal sensors are deployed at key locations in the battery pack and charging pile 1 to achieve high-precision real-time acquisition of average temperature, thermal runaway gas concentration, and current data at the measurement points. Combined with sliding time window statistics, time-series synchronization, outlier removal, moving average and Kalman filtering, normalization, and standardization preprocessing techniques, the completeness, accuracy, and engineering applicability of the data are greatly improved. Based on the quantum thermal runaway monitoring database, a solid data foundation is provided for subsequent spatiotemporal feature fusion, principal component extraction, and risk trend prediction models.

[0041] Specifically, the process of quantifying thermal risk sensitivity based on preprocessed quantum thermal runaway data is as follows: Based on a sliding time window, the average temperature of the measuring points is obtained and the maximum temperature is selected. The maximum temperature change rate is obtained by calculating the difference between the maximum temperature in the current sliding time window and the maximum temperature in the previous sliding time window. This difference calculation involves subtracting the maximum temperature of the previous window from the current maximum temperature and then dividing by the time interval, thus quantifying the rate of temperature change and sensitively capturing early abnormal signals of rapid local temperature rise. Simultaneously, the standard deviation of the average temperature of all measuring points within the current sliding time window is calculated to obtain the temperature standard deviation. The temperature standard deviation measures the distribution fluctuation of the average temperature of the measuring points within the current period. A smaller standard deviation indicates a more uniform thermal field distribution, while a larger standard deviation indicates the presence of significant local hotspots and imbalances. The average thermal runaway gas concentration is obtained by acquiring the concentration of all thermal runaway gases and calculating it as a benchmark for subsequent gas anomaly gain, which helps to dynamically adapt to environmental changes. The temperature rise sensitivity value is obtained by dividing the maximum temperature change rate by the sum of the temperature standard deviation and the smallest positive value. The smallest positive value is used to prevent the denominator from being zero, ensuring computational stability and the engineering usability of the results. The current thermal runaway gas concentration is divided by the average thermal runaway gas concentration, and then added to a constant to perform a natural logarithmic operation. The natural logarithm helps to expand the magnitude response range of gas concentration changes and improve the nonlinear sensitivity to gas anomalies. The result of the natural logarithmic operation is added to a constant to obtain the gas sensitivity gain value. The introduction of a constant ensures that meaningful results can be obtained even when the gas concentration is low or zero, avoiding algorithm anomalies and data distortion. The temperature rise sensitivity value is multiplied by the gas sensitivity gain value to obtain the thermal control sensitivity value, which comprehensively reflects the three factors of local thermal field heating rate, temperature fluctuation and gas concentration anomaly. It is the core criterion for measuring the thermal runaway risk sensitivity of charging pile 1.

[0042] The specific formula for the thermal control sensitivity value is as follows:

[0043] ;

[0044] In the formula, The value represents the thermal control sensitivity, which is used to quantitatively assess the sensitivity of charging pile 1 to the micro-risk of thermal runaway within the sliding time window in real time. It comprehensively considers the maximum rate of change of the temperature field, the fluctuation of temperature distribution and the nonlinear enhancement of gas concentration. It can sensitively reflect the early signs of thermal runaway such as local rapid heating and associated gas leakage, and is an important input for subsequent model construction. This indicates the maximum temperature within the current sliding time window; It represents the maximum temperature change rate, which is used to capture and quantify the rate of change of the point where the temperature rises the fastest in real time. It is a core physical quantity for early warning of local overheating and early anomalies of thermal runaway. It represents the standard deviation of temperature, reflecting the fluctuation of the overall temperature distribution; To represent extremely small positive values ​​and prevent the denominator from being zero, the value is [value to be filled in]. ; This represents the temperature rise sensitivity value, which reflects the normalized magnitude of the maximum temperature change rate within the sliding time window relative to the overall temperature distribution fluctuation. The larger the value, the more significant the local sudden temperature rise and the higher the uniformity of the system temperature field. This indicates the current concentration of gases in thermal runaway; The average concentration of thermal runaway gas is used as a baseline for discrimination, which helps to dynamically identify abnormal gas signals. This represents the gas sensitivity gain value, which is logarithmically amplified and dimensionlessly normalized to the concentration of thermal runaway gases, enhancing the sensitivity to gas leaks and accumulation anomalies. The more gas there is and the higher the concentration, the greater the gas sensitivity gain value, and the more obvious the risk enhancement effect.

[0045] In this embodiment, Table 1 is a data table of thermal control sensitivity values. The table details the maximum temperature change rate, temperature standard deviation, thermal runaway gas concentration, average thermal runaway gas concentration, and thermal control sensitivity value for five time windows. Specifically, time window 1 corresponds to a maximum temperature change rate of 2.5, a temperature standard deviation of 0.4, a thermal runaway gas concentration of 22, an average thermal runaway gas concentration of 21, and a thermal control sensitivity value of 10.73; time window 2 corresponds to a maximum temperature change rate of 3.0, a temperature standard deviation of 0.6, a thermal runaway gas concentration of 28, an average thermal runaway gas concentration of 23, and a thermal control sensitivity value of 8.98; time window 3… The corresponding maximum temperature change rate is 1.8, the temperature standard deviation is 0.5, the thermal runaway gas concentration is 18, the mean thermal runaway gas concentration is 19, and the thermal control sensitivity value is 6.00; the maximum temperature change rate corresponding to time window 4 is 4.2, the temperature standard deviation is 0.7, the thermal runaway gas concentration is 36, the mean thermal runaway gas concentration is 24, and the thermal control sensitivity value is 11.50; the maximum temperature change rate corresponding to time window 5 is 2.2, the temperature standard deviation is 0.5, the thermal runaway gas concentration is 20, the mean thermal runaway gas concentration is 22, and the thermal control sensitivity value is 7.25.

[0046] Table 1. Thermal control sensitivity value data table

[0047]

[0048] like Figure 6 As shown, this is a combined distribution diagram of thermal runaway risk based on temperature rise sensitivity and gas gain. It illustrates the dynamic changes of thermal runaway sensitivity value, maximum temperature change rate, temperature standard deviation, and thermal runaway gas concentration within different sliding time windows. The horizontal axis represents the sliding window number, and the vertical axis represents the numerical values. (Based on Table 1 and...) Figure 6It can be seen that the thermal control sensitivity value fluctuates significantly with the change of time window, with the highest value in window 4, indicating the greatest risk of thermal runaway. A significant increase in gas concentration is the main driving factor for the peak value of thermal control sensitivity, while changes in temperature change rate and temperature standard deviation jointly affect the value. When the maximum temperature change rate is large, the average temperature at the measuring point rises uniformly, and the gas concentration increases significantly, the thermal control sensitivity value is amplified, indicating a potentially high-risk state. Overall, the thermal control sensitivity value is the result of multiple factors working together, comprehensively reflecting the dynamic changes in local rapid heating, temperature fluctuations, and gas anomalies, providing effective data for dynamic early warning and graded response to the thermal runaway risk of charging pile 1.

[0049] In this implementation scheme, by performing sliding window analysis, calculating the maximum temperature change rate and temperature standard deviation, and normalizing and enhancing the nonlinearity of the preprocessed quantum thermal runaway data, the thermal control sensitivity value can be accurately quantified. This comprehensively integrates three key factors: the local thermal field heating rate, overall temperature fluctuation, and abnormal gas concentration. This effectively improves the sensitivity of identifying and providing graded early warning of early risks of thermal runaway in charging piles, providing reliable input for subsequent model construction, and offering a scientific and engineering basis for proactive protection decisions and intelligent safety responses.

[0050] Specifically, the process of constructing a thermal runaway risk dataset is as follows: Thermal control sensitivity values ​​are written into a quantum thermal runaway monitoring database. The thermal control sensitivity values ​​and corresponding quantum thermal runaway data are combined to construct the thermal runaway risk dataset. The quantum thermal runaway data includes synchronously collected average temperature at measurement points, thermal runaway gas concentration, current data, and multimodal spatiotemporal information on current deviation, each corresponding one-to-one with the thermal control sensitivity value. This facilitates subsequent feature fusion, risk analysis, and AI modeling. Furthermore, if a sustained window where the thermal control sensitivity value exceeds the thermal control sensitivity threshold exceeds the maximum allowable threshold, the sampling frequency is increased. By shortening the data acquisition interval and increasing the data refresh rate, the ability to capture abnormal evolution processes can be enhanced, improving the real-time performance and accuracy of early warnings. The specific range for increasing the sampling frequency can be set from 1.5 times to 5 times the original conventional sampling frequency, depending on the actual application scenario. For example, if the conventional sampling frequency is 1Hz, it can be increased to 1.5Hz, 2Hz, 3Hz, 4Hz, up to a maximum of 5Hz. To avoid data redundancy and system load, the frequency can be dynamically and adaptively adjusted according to equipment performance, with the maximum sampling frequency not exceeding 10Hz. The recommended increment is 0.5Hz, and the duration of the increment can be dynamically adjusted based on the duration of the abnormal thermal control sensitivity value. Once the abnormality is resolved, the sampling frequency should be promptly restored to the normal frequency. Furthermore, the verification of thermal runaway data should be strengthened, including automatic removal of abnormal data, intelligent imputation of missing data, and multi-channel cross-verification of key measurement points. This ensures the reliability, accuracy, and engineering usability of the data input into subsequent models and databases, before proceeding to the spatiotemporal feature fusion and trend prediction process.

[0051] In this implementation plan, by writing the thermal control sensitivity values ​​and their corresponding multimodal quantum thermal runaway data into the quantum thermal runaway monitoring database in real time, and dynamically constructing a thermal runaway risk dataset by window, the complete archiving and efficient management of key risk characteristics can be ensured. When the thermal control sensitivity is continuously detected to exceed the threshold, the data acquisition frequency is increased and data verification is strengthened, further improving the sensitivity of abnormal operating conditions and the accuracy of data. This provides a solid and reliable data foundation for subsequent spatiotemporal feature fusion, risk trend prediction, and proactive protection decision-making.

[0052] Specifically, a thermal runaway spatiotemporal matrix is ​​constructed using a thermal runaway risk dataset. A principal component extraction model for thermal runaway risk is then built based on this matrix. Finally, a risk trend prediction model is constructed by combining the thermal runaway spatiotemporal matrix and the principal component extraction model. The specific process is as follows: The intelligent algorithm controller 2 receives the thermal runaway risk dataset and performs spatiotemporal feature fusion and trend prediction. A thermal runaway spatiotemporal matrix is ​​constructed using the average temperature sequence of measurement points, the thermal runaway gas concentration sequence, the current data sequence, the current deviation sequence, and the corresponding thermal control sensitivity value sequence. The thermal runaway spatiotemporal matrix is ​​a multivariate time-space matrix aligned to a unified time base. An empty dataset, where each row corresponds to a specific moment and each column represents a different physical quantity, comprehensively reflects the multidimensional dynamic changes during thermal runaway. The thermal runaway spatiotemporal matrix is ​​used as input, and a spatiotemporal attention network algorithm is employed for training. This algorithm, a deep learning approach that integrates spatial and temporal dependencies, adaptively assigns attention weights to extract key features from critical measurement points and anomalous time periods. Spatiotemporal alignment, feature extraction, and principal component analysis (PCA) are then performed. Spatiotemporal alignment synchronizes the data order of different measurement points and physical quantities, ensuring the integrity of the model input. PCA is used for PCA dimensionality reduction. This method significantly reduces data redundancy, improves model generalization ability, and constructs a principal component extraction model for thermal runaway risk. It obtains the principal feature components of the thermal field, the principal trend components of the gas, and implicit risk features. The principal feature components of the thermal field represent the main trend of the global temperature distribution, the principal trend components of the gas capture the dynamic changes in the concentration of gas during thermal runaway, and the implicit risk features reflect the synergistic anomalies and potential risk evolution among different physical quantities. Using the thermal runaway spatiotemporal matrix, the principal feature components of the thermal field, the principal trend components of the gas, and the implicit risk features as inputs, supervised training is performed using a Long Short-Term Memory (LSTM) temporal network algorithm. LSTM is a type of recurrent neural network... The network effectively captures the evolution patterns of risk indicators within future time windows by modeling the nonlinear dependencies of long-term time series data. The loss function is jointly optimized using multi-objective regression, which incorporates multiple objective variables from the thermal runaway spatiotemporal matrix, the principal characteristic component of the thermal field, the principal trend component of the gas, and the implicit risk characteristics into a unified loss function for collaborative optimization. This improves the prediction accuracy of each key risk indicator, constructs a risk trend prediction model, predicts the evolution trend of key risk indicators within future time windows, and outputs predicted values ​​for the principal characteristic component of the thermal field, the principal trend component of the gas, the current deviation, and the average temperature of the maximum measurement point.

[0053] This implementation scheme constructs a multidimensional thermal runaway spatiotemporal matrix, employs a spatiotemporal attention network for key feature extraction and principal component dimensionality reduction, and combines it with multi-objective regression training using a long short-term memory temporal network. This allows for the comprehensive fusion of multimodal dynamic information on temperature, gas concentration, current, and sensitivity, enabling accurate extraction of the main features of thermal runaway risk in charging piles and intelligent prediction of future trends. This effectively improves the foresight and accuracy of risk assessment criteria, providing robust and intelligent data and model support for subsequent multimodal risk assessment and tiered proactive protection decisions. It significantly enhances the ability to provide early warning of thermal runaway hazards and improves the level of safety assurance.

[0054] Specifically, the process of comprehensively judging thermal runaway risk by integrating the principal component extraction model and the risk trend prediction model is as follows: First, obtain the current principal characteristic component of the thermal field and its predicted value. Then, calculate the natural logarithm of the ratio of the current principal characteristic component to the predicted value, and take the absolute value to obtain the thermal field principal characteristic mutation term. The natural logarithm enhances the sensitivity to abnormal mutations, and the absolute value calculation ensures that the risk measure is non-negative. The thermal field principal characteristic mutation term can keenly capture significant drifts and sudden turning points in the temperature field. Second, obtain the current current deviation and its predicted value. Then, take the square root of the ratio of the absolute values ​​of the current current deviation and its predicted value to obtain the key current anomaly term. The square root calculation can mitigate the impact of extreme anomalies and improve the stability and usability of the indicator in actual engineering. The key current anomaly term is used to quantify local short circuits, abnormal loads, and... The system identifies risk signals related to power imbalance. Based on a sliding time window, it acquires the main gas trend component and calculates its standard deviation. It then obtains the predicted value of the main gas trend component, divides the difference between the predicted value and the current main gas trend component by the standard deviation, and performs an arctangent function operation to obtain a gas concentration fluctuation term. The arctangent function (arctan) can compress extreme abnormal changes, ensuring the indicators are bounded and sensitive to large-scale gas leaks. The gas concentration fluctuation term can capture the risks of sudden abnormal gas releases and hazardous gas accumulation. Finally, it adds the thermal field main characteristic mutation term, the key current anomaly term, and the gas concentration fluctuation term to obtain a multimodal main risk judgment value. This comprehensively integrates the three key physical quantities of heat, gas, and electricity, as well as their future evolution trends, achieving multi-channel, multi-temporal-dimensional collaborative intelligent risk identification. This serves as the core quantitative basis for proactive early warning and risk response.

[0055] The specific formula for the multimodal main risk assessment value is as follows:

[0056] ;

[0057] In the formula, It represents the multimodal main risk judgment value, which is used to quantitatively assess the multimodal comprehensive risk level of charging pile 1 in real time. It integrates three core physical quantities, namely thermal field, key current and gas main trend and their future prediction information. It can capture potential thermal runaway, abnormal operating conditions and early risks in a multidimensional and proactive manner. It is an important quantitative criterion for active protection and risk classification. It represents the main characteristic component of the current thermal field, and is a comprehensive quantity that reflects the main trend of the current thermal distribution after fusion and dimensionality reduction of all temperature measurement points. It represents the predicted value of the main characteristic components of the thermal field, which is used to measure the degree of deviation between the current thermal field state and the future trend, and to detect the signs of thermal runaway in advance; It represents the current current deviation and is a direct quantitative indicator of electrical risk; This indicates that the current deviates from the predicted value, and is used to determine whether the current anomaly has a tendency to worsen further; It represents the predicted value of the main trend component of gas, reflecting the potential development of current gas anomalies in the future, and is a key indicator for early identification of gas emergencies. This represents the current main trend component of the gas, namely the global gas concentration anomaly trend obtained through dimensionality reduction and feature fusion, which is used to reflect the potential gas leakage and decomposition release hazards of charging pile 1. This represents the standard deviation of the main trend component of the gas, used to normalize abnormal gas variations. It represents the abrupt change of the main characteristic of the thermal field, measures the logarithmic change between the current main characteristic components of the thermal field and the future predicted values, and sensitively reflects the abrupt changes, drifts and overall pattern changes of the abnormal trend of the thermal field. It represents key current anomalies, measures the normalized comparison between current current deviation and future predicted current anomalies, and is used to capture risk signals of short circuits, failures, and local power anomalies. It represents the term of dramatic changes in gas concentration, measures the normalized difference between the predicted value and the current value of the main trend component of future gas, sensitively reflects the abnormal dynamics of gas leakage and decomposition release, and improves the identification and stability of extreme anomalies.

[0058] This implementation plan integrates a thermal runaway risk principal component extraction model with a risk trend prediction model, enabling comprehensive analysis of current observations and AI predictions across multiple modes of thermal field, current, and gas. It quantitatively extracts key thermal field abrupt changes, critical current anomalies, and dramatic gas concentration variations, and calculates multimodal principal risk assessment values. This significantly enhances the sensitivity and anomaly detection capabilities for complex thermal runaway risks in charging piles, achieving collaborative discrimination and early warning of risk information across multiple physical quantities and spatiotemporal dimensions. It provides comprehensive and scientific criteria for subsequent protective responses, effectively strengthening intelligent safety protection and overall risk controllability.

[0059] Specifically, the process of early detection and proactive protection against thermal runaway risk based on the assessment results is as follows: Multimodal main risk assessment values ​​are written into the quantum thermal runaway monitoring database in real time and compared with risk thresholds. When the multimodal main risk assessment value is less than the risk threshold, the multimodal main risk assessment value is continuously monitored, entering a routine inspection state to ensure smooth data flow throughout the process and real-time updates of the multimodal main risk assessment value. The sampling frequency is periodically increased to capture potential anomalies. Regularly increasing the sampling frequency helps to promptly detect short-term and sudden subtle anomalies, improving the accuracy of early warnings. When the multimodal main risk assessment value is greater than or equal to the risk threshold, it is determined to be a warning condition, indicating that the current thermal runaway risk has reached a critical safety level requiring intervention. The corresponding time window's multimodal main risk assessment value, thermal runaway data, thermal control sensitivity value, and maximum measurement point average temperature prediction are then used. The values ​​are packaged and marked as warnings to facilitate subsequent tracing, expert review, and fault analysis, ensuring the closed-loop integrity of the data link. The data collection frequency is temporarily increased, greatly shortening the monitoring cycle, capturing the dynamic details of the abnormal evolution process, and activating circulation pump 5 and electric valve 6. Circulation pump 5 is used to increase the flow rate of liquid cooling medium, while electric valve 6 precisely controls the flow direction and opening of the cooling circuit to achieve rapid local cooling. This pre-cools the charging pile 1, effectively suppressing the further development of thermal runaway and reducing the scope of the abnormal risk. At the same time, the fan is started in advance to discharge the thermal runaway gas to the safe area. Forced ventilation by the fan can quickly reduce the local gas concentration, ensuring the safety of the charging pile 1 cabinet and its surrounding environment. The gas emission channel is detected and its status is monitored in real time to prevent blockage and abnormal diffusion, ensuring the controllable and safe emission of dangerous gases, and entering the safety threshold assessment process.

[0060] This implementation plan establishes a closed-loop monitoring mechanism for the entire process data by recording and dynamically comparing multimodal main risk judgment values ​​in real time. When the risk does not reach the threshold, continuous inspection and periodic increases in sampling frequency improve the ability to detect short-term anomalies and early risks. When the risk exceeds the threshold and enters the warning condition, key risk data can be automatically packaged, the monitoring frequency can be temporarily increased, and the circulating pump 5, electric valve 6, and fan actuator can be intelligently linked to achieve precise control of liquid cooling medium flow and cooling circuit and rapid local cooling, while strengthening the detection of gas extraction and emission channel status. This effectively ensures the safety of the charging pile 1 cabinet and its surrounding environment, enhances the ability to actively identify and respond to thermal runaway risks, and comprehensively improves the level of early warning, anomaly handling, and operational safety.

[0061] Specifically, the process of determining the safety threshold by integrating thermal runaway data, risk trend prediction models, and thermal runaway risk is as follows: Receive the multimodal main risk judgment value, thermal runaway data, thermal control sensitivity value, and the predicted average temperature of the maximum measuring point within the warning operating condition time window, and conduct a safety threshold assessment: Calculate the second derivative of the average temperature of the measuring point over time, take the square root of the absolute value to obtain the thermal field fluctuation value. The second derivative reflects the acceleration of temperature change, and the square root process helps normalize the fluctuation magnitude of different measuring points. The final thermal field fluctuation value is used to quantify the drastic changes and dynamic instability of the local thermal field. Based on the warning operating condition time window, calculate the mean and standard deviation of the thermal runaway gas concentration. The mean is used to characterize the gas concentration baseline, while the standard deviation characterizes the dispersion of the gas distribution, helping to identify abnormal aggregation and drastic changes. Divide the difference between the current thermal runaway gas concentration and the mean thermal runaway gas concentration by the sum of the standard deviation of the thermal runaway gas concentration and the minimum positive value, and then square the result. The minimum positive value is taken as... To avoid division by zero and improve the stability and robustness of engineering calculations, the square calculation enhances the sensitivity to extreme anomalies. The gas anomaly gain value is obtained by performing a natural logarithmic operation on the sum of the square calculation result and a constant 1. The constant 1 ensures that the result is positive and comparable, and the natural logarithm can normalize extreme anomalies to a finite interval, improving interpretability. Based on the warning condition time window, the thermal control sensitivity value is integrated within the time window and divided by the window length to obtain the anomaly integral mean. The integral mean reflects the overall microscopic anomaly activity level within the warning condition time window, which helps to capture persistent and slow-changing risks. The safety threshold value is obtained by adding the thermal field fluctuation value, the gas anomaly gain value, the anomaly integral mean, and the corresponding maximum measurement point average temperature prediction value. By comprehensively integrating multi-source key indicators of thermal field dynamics, gas anomalies, and microscopic sensitivity, a quantitative and scientific decision-making basis is provided for the final protection action judgment and graded response.

[0062] The specific formula for the safety threshold value is as follows:

[0063] ;

[0064] In the formula, It represents the safety threshold value, which serves as the final trigger threshold for active protection actions. It is used to make the final decision after considering multiple risk signals, such as whether to cut off power and enforce safety measures, and provides quantitative and engineered criteria for the automatic triggering of subsequent active protection measures. It represents the predicted average temperature at the maximum measurement point, reflecting the future risk of thermal runaway and providing an early warning function; This indicates the average temperature at the measuring point; It represents the second derivative of the average temperature at the measuring point with time, reflecting the degree of drastic temperature change; It represents the degree of thermal field fluctuation, measures the intensity of fluctuation in the overall thermal field, and detects the risk of local abnormal temperature rise; This indicates the current concentration of gases in thermal runaway; This represents the average concentration of gases involved in thermal runaway. Indicates the standard deviation of the thermal runaway gas concentration; To represent extremely small positive values ​​and prevent the denominator from being zero, the value is [value to be filled in]. ; It represents the abnormal gain value of gas, which is highly sensitive to and normalized to abnormal and drastic changes in gas, and suppresses the influence of extreme values. This represents the thermal control sensitivity value, which is used to quantitatively assess the sensitivity of charging pile 1 to the microscopic risk of thermal runaway within a sliding time window in real time. It represents the average of the anomaly integral, reflecting the overall level of micro-level anomaly activity in the recent period and capturing potential early multimodal hazards.

[0065] In this implementation plan, by integrating and analyzing multimodal main risk judgment values, thermal runaway data, thermal control sensitivity values, and predicted average temperatures at maximum measurement points, multiple quantitative indicators are constructed, including thermal field fluctuation degree values, gas anomaly gain values, and the average integral value of anomaly degree. This scientifically establishes a comprehensive safety threshold value. This multidimensional and dynamic approach reflects the overall risk status of charging pile 1 under early warning conditions, providing a solid quantitative foundation for triggering subsequent graded protection measures and making decisions on active protection strategies. It improves the accuracy of thermal runaway risk assessment and the scientific nature of the response, effectively ensuring the safety and high reliability of the entire operation process.

[0066] Specifically, based on the safety threshold judgment results, the process of implementing graded protection response for thermal runaway of charging pile 1 and achieving closed-loop optimization of protection capabilities is as follows: real-time comparison of safety threshold values. With multi-level security threshold and ;when < At the same time, continuous monitoring commands are output to maintain the normal sampling frequency, adjust the liquid cooling equipment and fan equipment, perform only low-power preheating, and maintain a slight positive pressure in charging pile 1 cabinet. The slight positive pressure in charging pile 1 cabinet is achieved through the air supply system to prevent external dust and harmful gases from seeping into the cabinet, thereby improving the safety of the equipment operating environment. At the same time, healthy samples are periodically pushed to the quantum thermal runaway monitoring database to facilitate subsequent model self-learning and dynamic adjustment of the early warning judgment baseline. ≥ and < When an abnormal condition occurs, the system outputs current limiting and warning commands, lowering the charging and discharging current below the safe threshold. Through precise control of the output current via BM, it reduces load pressure, enhances the operation of liquid cooling and gas extraction equipment, automatically increases coolant flow in the liquid cooling system, and accelerates the emission of hazardous gases in the gas extraction system, effectively suppressing the spread of the abnormality. For abnormal points, it increases coolant flow and air velocity, adjusts the gas ventilation and filtration strategy, and dynamically switches multi-stage filtration modules to improve the purification capacity for harmful gases and particulate matter. All abnormal operating condition data is uploaded to the maintenance terminal, and remote verification is performed via pop-up reminders on the maintenance terminal, ensuring that maintenance engineers can intervene remotely immediately to verify risk conditions and formulate emergency measures. ≥ In case of emergency, the system outputs power failure and alarm commands, the main controller immediately cuts off relevant circuits, executes power failure protection, and completely blocks abnormal energy flow by forcibly disconnecting the main circuit breaker. On-site buzzers and indicator lights provide strong alerts and high-priority alarm signals. The edge controller, in conjunction with gas extraction, cabinet positive pressure protection, and forced liquid cooling equipment, all enter their highest settings, achieving full-system active protection under extreme conditions and minimizing accident risks. An abnormal work order is generated and linked to a hazard alarm, which is pushed to the maintenance end. The abnormal work order includes safety threshold values ​​and corresponding real-time thermal runaway data, facilitating traceability and rapid response by the maintenance end. The system continuously monitors the entire process's quantum thermal runaway data, thermal control sensitivity values, multimodal main risk judgment values, safety threshold values, and corresponding... The protective effect is periodically statistically analyzed, including closed-loop data tracking of the results and recovery process of each protective action to improve the model's self-learning ability. Combined with healthy samples, the sliding adaptive threshold algorithm and Bayesian optimization algorithm are used. The sliding adaptive threshold algorithm is used to dynamically adjust various thresholds, while the Bayesian optimization algorithm is used for parameter tuning and automatic model iteration and upgrading. The algorithm parameters, thermal runaway risk principal component extraction model, risk trend prediction model, risk threshold, and multi-level safety threshold are updated and optimized. After each round of analysis and optimization, optimization suggestions and the latest updated strategies are pushed to the operation and maintenance end to realize end-to-cloud collaboration and intelligent optimization throughout the entire lifecycle, continuously improving risk perception, early warning, and protection capabilities.

[0067] like Figure 5The diagram shows the workflow for graded protection and response to thermal runaway in charging piles. Starting with the extraction and fusion of feature data, multiple features are integrated, including multimodal main risk assessment values, thermal runaway data, thermal control sensitivity values, and predicted average temperatures at maximum measurement points. This process calculates safety thresholds and completes the core safety assessment. Subsequently, decisions are made at the edge based on real-time data. Based on the comparison between the safety thresholds and multi-level safety thresholds, graded protection responses are implemented: under normal conditions, continuous monitoring commands are output to ensure stable operation; during a Level 1 warning, current limiting and warning commands are output to reduce charging and discharging current, provide localized cooling, and incorporate key historical data into a deep learning model for self-evolution; during a Level 2 warning, power-off and alarm commands are triggered, immediately cutting off power and notifying relevant maintenance personnel to prevent the accident from escalating. Simultaneously, all historical data and processing results are uploaded to the cloud. The cloud periodically performs deep learning and model optimization, driving adaptive updates to the entire process's judgment and response strategies, achieving closed-loop safety management through edge intelligent judgment and cloud-based collaborative optimization.

[0068] In this implementation plan, a graded protection response and intelligent closed-loop optimization for thermal runaway of charging piles are achieved through dynamic comparison of safety threshold values ​​with multi-level safety threshold values. Under different risk levels, the system can adaptively adjust liquid cooling and fan equipment, dynamically limit current, enhance gas extraction, and switch to power failure protection and high-priority alarms according to the level of abnormality. Key data and protection effects throughout the entire process are continuously collected, analyzed, and archived. Combined with sliding adaptive thresholds and Bayesian optimization algorithms, the algorithm parameters and risk criteria are periodically self-learned and updated. This effectively ensures the proactive early warning and efficient protection capabilities of charging pile 1 under various complex operating conditions.

[0069] Reference Figure 2As shown, the second aspect of the present invention provides a quantum haptic-driven active protection system for thermal runaway in charging piles, applied to the aforementioned quantum haptic-driven active protection method for thermal runaway in charging piles. The system includes: a quantum haptic-driven multimodal monitoring and processing module, used to collect quantum thermal runaway data in real time, preprocess the quantum thermal runaway data, quantify the thermal risk sensitivity based on the preprocessed quantum thermal runaway data, and construct a thermal runaway risk dataset; and a spatiotemporal feature fusion and trend prediction module, used to construct a thermal runaway spatiotemporal matrix using the thermal runaway risk dataset, and construct a thermal runaway risk principal component extraction module based on the thermal runaway spatiotemporal matrix. The system employs a multimodal risk discrimination and intervention module, which integrates the thermal runaway risk principal component extraction model with the thermal runaway risk principal component extraction model to comprehensively assess thermal runaway risk and achieve early detection and proactive protection based on the assessment results. A protection threshold judgment, execution, and optimization module is used to comprehensively assess thermal runaway data, the risk trend prediction model, and the thermal runaway risk to determine safety thresholds. Based on the safety threshold assessment results, a graded protection response for thermal runaway of charging pile 1 is implemented, achieving closed-loop optimization of protection capabilities.

[0070] In this implementation plan, the quantum-sensing multimodal monitoring and processing module achieves high-precision, multi-channel real-time data acquisition and risk sensitivity quantification, providing a solid data foundation for subsequent analysis. Combined with the spatiotemporal feature fusion and trend prediction module, the main component of thermal runaway risk can be effectively extracted, and intelligent prediction of future trends can be achieved, significantly improving the foresight of risk assessment. The multimodal risk discrimination and intervention module realizes comprehensive discrimination and early proactive protection against thermal runaway risk, greatly enhancing anomaly identification and response capabilities. The protection threshold judgment, execution, and optimization module realizes hierarchical protection and closed-loop self-learning optimization, ensuring intelligent protection and continuous evolution throughout the entire process under various complex operating conditions. Overall, this significantly improves the proactive perception, dynamic discrimination, and safety assurance level of charging pile thermal runaway risk, effectively promoting high reliability and intelligence in operation.

[0071] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0072] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. As those skilled in the art will understand, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A quantum-sensory-driven active protection method for thermal runaway in charging piles, characterized in that, Includes the following steps: S1: Real-time acquisition of quantum thermal runaway data, data preprocessing of quantum thermal runaway data, quantification of thermal risk sensitivity based on preprocessed quantum thermal runaway data, and construction of thermal runaway risk dataset; The specific process for quantifying thermal risk sensitivity based on preprocessed quantum thermal runaway data is as follows: Based on the sliding time window, the average temperature of the measuring points is obtained and the maximum temperature is selected. The difference between the maximum temperature in the current sliding time window and the maximum temperature in the previous sliding time window is calculated to obtain the maximum temperature change rate. At the same time, the standard deviation of the average temperature of all measuring points in the current sliding time window is calculated to obtain the temperature standard deviation. The average thermal runaway gas concentration is obtained and the average thermal runaway gas concentration is calculated. The temperature rise sensitivity value is obtained by dividing the maximum temperature change rate by the sum of the temperature standard deviation and the smallest positive value; the gas sensitivity gain value is obtained by dividing the current thermal runaway gas concentration by the mean thermal runaway gas concentration and adding it to a constant. The thermal control sensitivity value is obtained by multiplying the temperature rise sensitivity value by the gas sensitivity gain value. S2. A thermal runaway spatiotemporal matrix is ​​constructed using a thermal runaway risk dataset. A thermal runaway risk principal component extraction model is constructed based on the thermal runaway spatiotemporal matrix. A risk trend prediction model is constructed by combining the thermal runaway spatiotemporal matrix and the thermal runaway risk principal component extraction model. The specific process of constructing a thermal runaway spatiotemporal matrix using a thermal runaway risk dataset, building a thermal runaway risk principal component extraction model based on the thermal runaway spatiotemporal matrix, and combining the thermal runaway spatiotemporal matrix and the thermal runaway risk principal component extraction model to construct a risk trend prediction model is as follows: The intelligent algorithm controller (2) receives the thermal runaway risk dataset and performs spatiotemporal feature fusion and trend prediction: it constructs a thermal runaway spatiotemporal matrix using the average temperature sequence of the measuring point, the thermal runaway gas concentration sequence, the current data sequence, the current deviation sequence and the corresponding thermal control sensitivity value sequence, and uses the thermal runaway spatiotemporal matrix as input. It then trains the spatiotemporal attention network algorithm to perform spatiotemporal alignment, feature extraction and principal component dimensionality reduction, and constructs a thermal runaway risk principal component extraction model to obtain the thermal field principal feature component, the gas principal trend component and the implicit risk feature. The thermal runaway spatiotemporal matrix, the main feature component of the thermal field, the main trend component of the gas, and the implicit risk features are used as inputs. The long short-term memory time-series network algorithm is used for supervised training. The loss function is jointly optimized by multi-objective regression. A risk trend prediction model is constructed to predict the evolution trend of key risk indicators within the future time window. The predicted values ​​of the main feature component of the thermal field, the main trend component of the gas, the current deviation, and the average temperature of the maximum measurement point are output. S3 integrates the principal component extraction model of thermal runaway risk and the risk trend prediction model to comprehensively judge the risk of thermal runaway and realize early detection and active protection of thermal runaway risk based on the judgment results. The specific process by which the principal component extraction model for thermal runaway risk and the risk trend prediction model are combined to comprehensively determine the risk of thermal runaway is as follows: Obtain the current principal characteristic component of the thermal field and the predicted value of the principal characteristic component of the thermal field. Calculate the natural logarithm of the ratio of the current principal characteristic component of the thermal field to the predicted value of the principal characteristic component of the thermal field, and take the absolute value to obtain the thermal field principal characteristic mutation term. Obtain the current current deviation and the predicted current deviation. Take the square root of the ratio of the absolute values ​​of the current current deviation and the predicted current deviation to obtain the key current anomaly. Based on the sliding time window, the main trend component of the gas is obtained, and the standard deviation is calculated to obtain the standard deviation of the main trend component of the gas. The predicted value of the main trend component of the gas is obtained. The difference between the predicted value of the main trend component of the gas and the current main trend component of the gas is divided by the standard deviation of the main trend component of the gas, and the arctangent function is performed to obtain the gas concentration drastic change term. The multimodal main risk assessment value is obtained by adding the main characteristic mutation term of the thermal field, the key current anomaly term, and the gas concentration drastic change term. The specific process for early detection and proactive protection against thermal runaway risk based on the assessment results is as follows: The multimodal main risk assessment value is written into the quantum thermal runaway monitoring database in real time and compared with the risk threshold. When the multimodal main risk assessment value is less than the risk threshold, the multimodal main risk assessment value is continuously monitored and the sampling frequency is periodically increased to capture potential anomalies. When the multimodal main risk judgment value is greater than or equal to the risk threshold, it is judged as an early warning condition. The multimodal main risk judgment value, thermal runaway data, thermal control sensitivity value and the predicted value of the average temperature of the maximum measuring point in the corresponding time window are packaged and marked as an early warning. The data acquisition frequency is temporarily increased, and the circulating pump (5) and electric valve (6) are turned on to cool down the charging pile (1) in advance. At the same time, the fan is started in advance to discharge the thermal runaway gas to the safe area. The gas emission channel is detected and the safety threshold assessment process is entered. S4, integrate thermal runaway data, risk trend prediction model and thermal runaway risk to determine safety threshold, and based on the safety threshold judgment result, carry out graded protection response for thermal runaway of charging pile (1) and realize closed-loop optimization of protection capability.

2. The active protection method for thermal runaway of charging piles driven by quantum haptic feedback according to claim 1, characterized in that, The specific process of real-time acquisition of quantum thermal runaway data and data preprocessing of the quantum thermal runaway data is as follows: Quantum sensor groups (7) are deployed in key parts of the battery pack and charging pile (1) to collect quantum thermal runaway data in real time. The quantum thermal runaway data includes: average temperature of the measuring point, thermal runaway gas concentration and current data. At the same time, the average value of current data is calculated based on the sliding time window, and the difference between the current data and the average value of current data at each moment in the sliding time window is calculated to obtain the current deviation. The original quantum thermal runaway data was synchronized by aligning with a unified timestamp. Outliers were removed and short-term missing data were filled in using a sliding window statistical method. Moving average and Kalman filtering algorithms were used for signal smoothing and noise suppression. The quantum thermal runaway data was then normalized and standardized using dimensionless methods. A quantum thermal runaway monitoring database was constructed by attaching timestamps to the quantum thermal runaway data and storing it in the database in groups according to a fixed sliding time window.

3. The active protection method for thermal runaway of charging piles driven by quantum haptic feedback according to claim 1, characterized in that, The specific process for constructing the thermal runaway risk dataset is as follows: The thermal control sensitivity value is written into the quantum thermal runaway monitoring database. The thermal control sensitivity value and the corresponding quantum thermal runaway data are combined to construct a thermal runaway risk dataset. At the same time, if the thermal control sensitivity value is found to be higher than the thermal control sensitivity threshold for a continuous window exceeding the maximum allowable threshold, the sampling frequency is increased and the verification of thermal runaway data is strengthened, and the process of spatiotemporal feature fusion and trend prediction is initiated.

4. The active protection method for thermal runaway of charging piles driven by quantum haptic feedback according to claim 1, characterized in that, The specific process for determining the safety threshold based on the integrated thermal runaway data, risk trend prediction model, and thermal runaway risk is as follows: Receive multimodal primary risk assessment values, thermal runaway data, thermal control sensitivity values, and predicted average temperature at the maximum measurement point within the warning operating condition time window, and perform safety threshold assessment: Calculate the second derivative of the average temperature at the measuring point with time, and take the square root of the absolute value to obtain the degree of thermal field fluctuation. Based on the warning operating condition time window, the mean concentration of thermal runaway gas and the standard deviation of thermal runaway gas concentration are calculated. The difference between the current thermal runaway gas concentration and the mean concentration of thermal runaway gas is divided by the sum of the standard deviation of thermal runaway gas concentration and the smallest positive value, and then squared. The gas anomaly gain value is obtained by performing a natural logarithmic operation on the sum of the squared result and the constant one. Based on the warning operating condition time window, the thermal control sensitivity value is integrated within the time window and divided by the window length to obtain the average value of the anomaly integral; The safety threshold is obtained by adding the thermal field fluctuation value, the gas anomaly gain value, the average value of the anomaly integral, and the predicted value of the average temperature of the corresponding maximum measuring point.

5. The active protection method for thermal runaway of charging piles driven by quantum haptic feedback according to claim 4, characterized in that, The specific process of performing graded protection response for thermal runaway of charging piles based on the safety threshold judgment result, and realizing closed-loop optimization of protection capabilities is as follows: Real-time comparison of security threshold values With multi-level security threshold and ;when At the same time, output continuous monitoring instructions, maintain the normal sampling frequency, adjust the liquid cooling equipment and fan equipment, perform low power preheating, and maintain the charging pile (1) cabinet with slight positive pressure, while regularly pushing health samples to the quantum thermal runaway monitoring database; when When the current is limited and warning commands are issued, the charging and discharging current is reduced to below the safe threshold, the operation of liquid cooling and gas extraction equipment is strengthened, the coolant flow rate and wind speed are increased for abnormal points, and the gas ventilation and filtration strategy is adjusted; all abnormal operating conditions data are uploaded to the operation and maintenance terminal, and remote verification is carried out through the operation and maintenance terminal pop-up reminder. when When power failure and alarm commands are output, the main controller will urgently cut off the relevant circuits and execute power failure protection. The on-site buzzer and indicator light will strongly indicate the power failure. The edge controller will link the gas extraction, cabinet positive pressure protection and forced liquid cooling equipment to the highest level. Generate an exception work order and push a linked danger alert to the operations and maintenance terminal; We continuously perform periodic statistical analysis on the quantum thermal runaway data, thermal control sensitivity values, multimodal principal risk judgment values, safety threshold values, and corresponding protection effects throughout the entire process. Combining healthy samples, we use sliding adaptive threshold algorithm and Bayesian optimization algorithm to update and optimize the parameters of each algorithm, the principal component extraction model of thermal runaway risk, the risk trend prediction model, the risk threshold, and the multi-level safety threshold. After each round of analysis and optimization, we simultaneously push optimization suggestions and the latest updated strategies to the operation and maintenance end.

6. A quantum-sensory-driven active protection system for thermal runaway in charging piles, employing the quantum-sensory-driven active protection method for thermal runaway in charging piles as described in any one of claims 1-5, characterized in that, include: The quantum somatosensory multimodal monitoring and processing module is used to collect quantum thermal runaway data in real time, preprocess the quantum thermal runaway data, quantify the thermal risk sensitivity based on the preprocessed quantum thermal runaway data, and construct a thermal runaway risk dataset. The spatiotemporal feature fusion and trend prediction module is used to construct a thermal runaway spatiotemporal matrix using a thermal runaway risk dataset, build a thermal runaway risk principal component extraction model based on the thermal runaway spatiotemporal matrix, and combine the thermal runaway spatiotemporal matrix and the thermal runaway risk principal component extraction model to construct a risk trend prediction model. The multimodal risk identification and intervention module is used to integrate the thermal runaway risk principal component extraction model and the risk trend prediction model to comprehensively judge the thermal runaway risk and realize early detection and active protection of thermal runaway risk based on the thermal runaway risk judgment results. The protection threshold judgment and optimization module is used to comprehensively analyze thermal runaway data, risk trend prediction model and thermal runaway risk to judge the safety threshold. Based on the safety threshold judgment result, it performs graded protection response for thermal runaway of charging pile (1) and realizes closed-loop optimization of protection capability.

Images