A defense-in-depth protection method for charging stations

Abstract

The present invention discloses a method for in-depth defense of charging stations, which relates to the field of fire warning and safety protection technology, and aims to solve the problems of high false alarm rate of fire warning, delayed fire isolation and insufficient credibility of accident tracing data in the prior art. The method constructs a full-cycle protection system of "monitoring-disposal-tracing": before the disaster, temperature field, electrical parameters and environmental data are collected through multi-source sensor fusion, adaptive safety thresholds are generated based on dynamic game models, and risk paths are predicted in combination with digital twin deduction; a three-level response mechanism is adopted during the disaster; after the disaster, operation logs are stored in shards through blockchain space-time anchoring technology, and the accident evolution chain is reversely reconstructed using digital twins to generate tamper-resistant judicial evidence. The present invention improves the warning accuracy through dynamic threshold optimization, shortens the response time through multi-level linkage control, and ensures the credibility of traceability through quantum evidence storage technology, forming a closed-loop optimization system for fire protection of charging stations.

Description

Technical Field

[0001] The present invention relates to the field of fire warning and safety protection technology, and in particular to a defense-in-depth protection method for a charging station. Background Art

[0002] Current charging station fire protection systems generally employ a single-dimensional monitoring mechanism based on fixed thresholds. Fire warnings are provided through basic sensors such as smoke and temperature sensors, and fire suppression strategies often rely on pre-set linkage programs. Existing technologies typically analyze fire risk factors using linear regression models, predict periodic risks using time series methods, and implement targeted fire extinguishing agent injection using solenoid valve control. While these solutions can provide basic risk warnings, they still have significant limitations in adaptability, response accuracy, and system self-optimization capabilities under complex operating conditions.

[0003] Chinese invention patent CN118491019B proposes a fire extinguishing system and optimization method for charging stations. However, the invention has the following technical defects: (1) Risk modeling relies on linear regression analysis and cannot capture the nonlinear correlation of multiple factors such as battery thermal runaway and arc ignition, resulting in a sharp decrease in model prediction accuracy as the complexity of the working conditions increases; (2) The fire extinguishing strategy is fixed on the opening and closing control of the solenoid valve, lacking a precise spraying mechanism based on dynamic positioning of the fire source, which is prone to blind spots in the coverage of fire extinguishing agents in scenarios of smoke obscuration or multiple fire points; (3) Data closed-loop feedback only realizes parameter adjustment through simple historical data analysis, and does not build a tamper-proof evidence storage system with judicial effectiveness, resulting in a lack of credible data support for accident tracing; (4) Time series analysis is limited to periodic trend identification, and does not introduce a real-time environmental parameter correction mechanism, making it difficult to deal with sudden risks such as charging load mutation and cooling failure. The above defects restrict the improvement of the safety protection capabilities of charging facilities. Summary of the Invention

[0004] The present invention aims to solve the three major technical problems of insufficient fire warning accuracy, poor timeliness of fire isolation, and low credibility of accident tracing data in the existing charging station fire protection system. Specifically, the present invention includes: (1) improving the accuracy of fire warning under complex working conditions through multi-dimensional environmental parameter coupling analysis and dynamic threshold optimization model, avoiding missed alarms and false alarms that lead to delayed disposal; (2) constructing a multi-spectral feature verification and hierarchical response mechanism to achieve centimeter-level positioning of fire sources and millisecond-level deployment of fire barriers, blocking the chain spread of thermal runaway of lithium batteries; (3) establishing a tamper-proof data storage and digital twin reverse deduction system, solidifying the full-cycle evidence chain of accidents based on blockchain space-time anchoring technology, providing high-confidence traceability support for responsibility determination, and ultimately forming a closed-loop optimization protection system of warning-disposal-traceability.

[0005] The present invention proposes a defense-in-depth protection method for charging stations, which includes: using a thermal imager to monitor the temperature field distribution of the target area in real time, while collecting multi-dimensional environmental parameters to generate a dynamic threshold judgment benchmark, and triggering an environmental adaptive algorithm to optimize the threshold when it is determined that the threshold needs to be updated; when the monitoring data exceeds the optimized threshold, a multi-spectral verification module is activated to confirm the fire characteristics, and the fire source is located based on the verification characteristics and the direction of disaster spread is predicted, and a three-level fire extinguishing strategy is dynamically adopted according to the spread direction; after the fire extinguishing is completed, the disposal process data is fragmented and stored in an anti-tampering database, and the cause chain of the accident is reversely traced based on the stored data to form a closed-loop feedback, driving the continuous iterative optimization of the monitoring threshold and response strategy.

[0006] Preferably, this method constructs a dynamic optimization function based on a game theory model by integrating the heat dissipation efficiency factor and the fluctuation characteristics of environmental parameters, generates an adaptive threshold range according to the real-time cooling system operating conditions and the chemical characteristics of the battery pack, and triggers a multi-level early warning mechanism that matches the battery type when the battery pack temperature rise rate is detected to exceed the preset chemical system sensitivity threshold.

[0007] Preferably, the method loads the corresponding temperature rise history model according to the chemical system type of the battery pack, calculates the threshold correction coefficient through the spatial distribution deviation between the real-time monitoring data and the historical model, and automatically lowers the temperature anomaly judgment threshold and increases the monitoring frequency when it is detected that the charge state deviation of the lithium iron phosphate battery pack reaches a preset safety margin boundary.

[0008] Preferably, this method generates a dynamic temperature field distribution based on the thermal radiation signal collected by the multi-spectral sensor array, and synchronously analyzes the evolution trend of thermal anomalies through a time series prediction model of preset temperature gradient threshold and temperature rise rate. When the radiation spectrum characteristics of the hot spot area are detected to match the preset lithium battery thermal runaway characteristic pattern, the ultra-wideband positioning data and visual positioning coordinates are integrated to construct a three-dimensional model of the fire source, and the environmental fluid parameters are loaded through the digital twin engine to simulate the disaster diffusion path in real time.

[0009] Preferably, the three-level fire extinguishing strategy includes: first-level response, when it is detected that the local temperature mutation exceeds the preset threshold, the high-precision positioning system is activated to control the sprinkler device to perform spatial directional fire extinguishing, and the spray angle and flow are adjusted in real time to match the dynamic position of the fire source; second-level response, when the spectral characteristics of the deflagration are identified, the high-reflectivity fireproof cover is synchronously activated to cover the fire source area and an electromagnetic isolation barrier is generated to block the thermal radiation propagation path; third-level response, when the monitoring system determines the risk of thermal runaway, the linkage protection mechanism is triggered, the energy input is cut off according to the emergency stop instruction of the charging pile, the sound and light alarm device is linked to send a warning signal, the asphyxiation fire extinguishing system is activated according to the location of the fire source to cover the target area, and the gradient cooling process is executed synchronously.

[0010] Preferably, the high-precision positioning system controls the spray device including: constructing a dynamic flow regulation system based on multi-sensor fusion, realizing real-time calibration of the spray angle through coordinated control of the solenoid valve array and the millimeter-wave radar, and dynamically matching the two-phase mixing ratio of fine water mist and gas fire extinguishing agent based on the energy density of the fire source.

[0011] Preferably, the high-reflectivity fireproof cover comprises a multi-layer composite protective structure, including an outer layer of high thermal conductivity composite coating, a middle layer of adaptive deformation skeleton and an inner layer of energy absorption layer. The centimeter-level positioning system drives the robotic arm to perform rapid packaging of the fireproof cover, while triggering the magnetic sealing array to construct an oxygen-deficient suppression environment.

[0012] Preferably, the linkage protection mechanism triggers the fire blanket deployment mechanism according to the emergency stop command, so that the flame retardant isolation layer completely covers the fire source area; starts the cooling process according to the fire source coverage status, and executes preliminary water mist injection, corrosion inhibitor cooling and pulse maintenance cooling in sequence; constructs a closed-loop control signal based on real-time fire data, and dynamically coordinates the action timing of the fire blanket deployment and the cooling system; activates the emergency power supply unit according to the energy input interruption status to maintain the continuous operation of key protection components.

[0013] Preferably, this method constructs a multimodal data sharding storage architecture, binds sensor data and operation records with three-dimensional features through spatiotemporal anchoring technology, generates sharding keys using a quantum-resistant encryption algorithm, and writes verification hash values ​​into blockchain nodes based on a consensus mechanism, forming an unalterable chain of evidence with judicial effect.

[0014] Preferably, this method reconstructs a multi-dimensional evolution model of the accident scenario through a digital twin engine, identifies threshold failure nodes and response strategy defects based on a causal reasoning algorithm, converts the analysis results into training samples to inject into the defense model, and generates an upgrade instruction package containing an optimized parameter set and a disposal plan and pushes it to associated devices.

[0015] The present invention has the following beneficial effects:

[0016] 1. Through dynamic threshold optimization models and multi-dimensional parameter fusion analysis, the risk of false alarms and missed alarms in complex scenarios is significantly reduced, achieving early and accurate identification and rapid response to fire hazards, effectively curbing the spread of fire.

[0017] 2. Based on multi-spectral feature verification and dynamic fire source positioning technology, rapid deployment of fire barriers and intelligent matching of fire extinguishing strategies are achieved, significantly improving the blocking efficiency of lithium battery thermal runaway spread and the accuracy of fire extinguishing agent coverage.

[0018] 3. Through tamper-proof data storage and reverse deduction using digital twins, a complete and verifiable chain of evidence for the accident is formed, providing judicial-level credible data support for liability determination and significantly simplifying the dispute resolution process.

[0019] 4. Based on real-time data feedback and causal reasoning mechanisms, dynamically adjust monitoring thresholds and response strategies to continuously improve the system's adaptability and defense effectiveness to new risk patterns.

[0020] 5. Integrated multiple protection mechanisms and redundant designs ensure stable operation of critical equipment under extreme operating conditions. At the same time, privacy protection technology is used to achieve data security compliance and meet high-level functional safety requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a schematic flow chart of the method of the present invention;

[0022] Figure 2 It is a schematic diagram of the fire graded response process of the present invention. DETAILED DESCRIPTION

[0023] Example 1

[0024] according to Figure 1 As shown, the present invention proposes a defense-in-depth protection method for charging stations, which specifically includes three parts: pre-disaster warning, firefighting during disasters, and post-disaster evidence storage.

[0025] The pre-disaster prediction system of this invention constructs a multidimensional monitoring network. Through a distributed sensor array deployed in charging areas, it provides continuous status awareness of charging equipment, vehicle batteries, and environmental parameters. The core monitoring module includes a thermal imaging scanner, a multispectral sensor unit, and a high-precision electrical parameter acquisition system, forming a comprehensive sensing capability covering temperature field distribution, energy transmission characteristics, and gas composition changes. The sensor network utilizes a redundant topology design to ensure comprehensive data collection in key areas, providing a complete raw data foundation for subsequent analysis.

[0026] The system introduces an adaptive threshold generation mechanism based on deep learning, overcoming the limitations of traditional fixed thresholds. By building a multi-dimensional parameter correlation model, it analyzes the dynamic coupling relationship between temperature change rate, voltage fluctuation pattern, heat dissipation efficiency, and environmental factors in real time. The threshold generation engine continuously receives input parameters such as the cooling system's operating status, battery chemistry, and historical failure modes. Combined with digital twin simulation results, it dynamically adjusts the warning ranges for each monitoring indicator, establishing a flexible safety boundary that matches environmental conditions and device status.

[0027] The collected raw data enters a multi-level processing pipeline, first undergoing noise suppression and feature enhancement to eliminate signal distortion caused by factors such as electromagnetic interference and mechanical vibration. A spatiotemporal alignment algorithm then fuses the discrete sensor data into a multidimensional state matrix in a unified spatiotemporal coordinate system, accurately characterizing the charging device's three-dimensional thermal field distribution, the dynamic characteristics of its electrical parameters, and the surrounding environment. The data fusion module specifically enhances the ability to track the formation of local hotspots, capturing the gradual temperature rise signals caused by micro-shorts within the battery pack.

[0028] The system constructs a high-fidelity digital twin model, accurately mapping the charging equipment, vehicle battery, and environmental factors in virtual space. The twin engine combines physical modeling with machine learning to simulate thermal runaway propagation paths, electrical fault evolution trends, and multi-device chain reaction scenarios in real time. During the simulation, real-time monitoring data is dynamically loaded as boundary conditions to generate a disaster development prediction model that incorporates parameters such as fault propagation speed, energy release intensity, and environmental impact range.

[0029] Based on a multi-agent reinforcement learning framework, the system establishes an attack-defense game model to simulate potential risk scenarios. A protection strategy generator and a virtual attack agent conduct adversarial training in a digital twin environment, continuously optimizing protection decisions through millions of game iterations. This game process focuses on addressing complex threat scenarios such as the exploitation of monitoring blind spots, threshold evasion attacks, and sensor deception, dynamically generating a multi-level strategy library encompassing preventative control, early intervention, and emergency response.

[0030] When monitoring data reaches a dynamic safety threshold, the system activates a tiered early warning mechanism. The primary warning stage implements preventive control measures through flexible measures such as adjusting charging power and enhancing heat dissipation. The intermediate warning stage triggers in-depth device status diagnosis, locating potential fault sources based on digital twin simulation results. The advanced warning stage initiates emergency protocol preloading, laying the foundation for subsequent rapid response. Warning information is pushed to the operation and maintenance terminal in real time via a multi-channel redundant transmission system, simultaneously activating the standby status of protective equipment.

[0031] The system establishes a continuous evolutionary mechanism, continuously absorbing new monitoring data and case studies through online learning modules. Regular incremental training of the early warning model optimizes the balance between sensitivity and specificity of the feature extraction network. The model's adaptability to new battery technologies and unique environmental conditions is particularly strengthened. Transfer learning techniques are used to rapidly adapt existing knowledge to new scenarios, ensuring the forward-looking and universal nature of the protection system.

[0032] To ensure monitoring reliability, the system integrates multiple self-check modules. It periodically performs self-checks, including sensor calibration, communication link verification, and computing resource health diagnosis. Any anomalies detected automatically switch to a backup unit. A consensus algorithm is used to cross-validate key monitoring data, identifying and resolving false alarms caused by single points of failure. A data credibility assessment system is established, and manual review is initiated for conflicting monitoring information, minimizing the risk of misjudgment.

[0033] During the fire confirmation phase, the system uses a heterogeneous sensor network to integrate thermal imaging, gas composition analysis, and acoustic signature recognition to establish a three-dimensional fire source location model. A multispectral sensor array captures the radiation characteristics of flames in specific frequency bands. Combined with a deep learning algorithm, it eliminates interference from ambient reflections to accurately determine the coordinates of the fire's origin and the direction of its spread. This location information is synchronized in real time to the dynamic protection decision engine, triggering the parallel loading of multimodal emergency response protocols.

[0034] The protection system activates a graded response mechanism, dynamically adjusting the fire response strategy based on the fire's stage of development. During the primary response phase, targeted suppression measures are deployed. High-pressure water mist creates a localized cooling zone to block heat conduction paths, while pre-protection mechanisms for adjacent equipment are activated. During the intermediate response phase, composite physical isolation is implemented. Driven by a servo mechanism, the intelligent fire shield rapidly closes along a pre-set track. The sealed structure is filled with inert gas to alter the combustion environment, and electromagnetic field restraint devices simultaneously form an energy shield. During the advanced response phase, a coordinated protection mechanism is implemented. First, intelligent monitoring of charging stations is used to determine the risk in real time, immediately disconnecting the charging circuit and activating audible and visual alarms. Simultaneously, the asphyxiation fire extinguishing system is activated, driving a flame-retardant coating to precisely seal the fire source. A graded cooling process is deployed to control temperature, creating a coordinated response model combining physical isolation and chemical suppression.

[0035] The dynamic fire extinguishing system utilizes a multi-parameter feedback control mechanism. The fire extinguishing agent injection device dynamically adjusts its range and intensity based on real-time fire scene scanning data. The flow control unit intelligently matches the water mist particle size distribution with the chemical inhibitor concentration gradient, combining thermal radiation intensity and smoke diffusion rate. The injection angle automatically adjusts to the movement of the fire source, ensuring that the extinguishing agent consistently covers the core combustion area while preventing secondary damage to unaffected equipment.

[0036] The physical isolation system creates a multi-layered protective structure. The nanocomposite reflective layer redirects thermal radiation back to the fire scene, the buffer layer absorbs the blast shockwave through controlled deformation, and the shape-shifting memory alloy skeleton, triggered by high temperatures, enhances sealing performance. An integrated environmental reconstruction module within the isolation barrier rapidly reduces oxygen concentration to below the combustion threshold through gas displacement, while a negative pressure maintenance system prevents the escape of toxic fumes.

[0037] The energy management and control system implements an intelligent power coordination strategy, simultaneously shutting off the main power supply circuit and switching to backup energy modules to maintain the operation of critical protection components. The power distribution unit analyzes the operating requirements of the fire extinguishing system and cooling equipment in real time, dynamically optimizing power supply paths and power allocation ratios to ensure a continuous and stable power supply to core disposal units. The safety protection module simultaneously reconfigures the grounding network topology, directing the discharge of fault energy along pre-defined safe paths and establishing a safe operating boundary for personnel.

[0038] Multi-dimensional operation logs are synchronously recorded throughout the fire response process. Data such as the protective equipment's action sequence, environmental parameter change curves, and decision-making logic deduction paths are encrypted and written to tamper-resistant memory. A quantum communication module uploads information about key response nodes to a distributed evidence storage network in real time, providing verifiable evidence of the operational chain of events for subsequent accountability. After suppressing a fire, the system automatically transitions to post-fire monitoring mode, continuously scanning for re-ignition risks until environmental parameters return to safe thresholds.

[0039] After the fire is completely extinguished and environmental parameters stabilize, the system initiates the multimodal data consolidation process. A complete record of the fire disposal process is transmitted back through the distributed sensor network, including thermal imaging sequences, extinguishing agent injection trajectories, equipment operation logs, and environmental parameter change curves. The data preprocessing module performs spatiotemporal alignment of heterogeneous information, eliminating timing deviations caused by transmission delays and constructing a multidimensional event timeline encompassing physical features, operational instructions, and environmental conditions. All data is feature extracted to generate standardized evidence units, each embedded with a unique spatiotemporal identifier to ensure the integrity and traceability of the data chain.

[0040] The system establishes a three-dimensional data anchoring mechanism, deeply binding collected raw data to geographic location and precise timestamps. The spatial anchoring module obtains the device's absolute coordinates through a satellite positioning system and combines them with UWB positioning data to establish a local coordinate system mapping relationship. Temporal anchoring utilizes a high-precision clock synchronization protocol, accurately tracking operation nodes down to the millisecond level. The data anchoring engine iteratively calculates the characteristic hash values ​​of key events to generate a unique Merkle tree root value. After anchoring, the data packets generate sharding keys using a quantum-resistant encryption algorithm, enabling distributed storage across multiple media.

[0041] The blockchain evidence storage network initiates the smart contract execution process, conducting multi-node consensus verification of the solidified data. This verification process includes data integrity verification, operational compliance review, and physical law compliance analysis. Through an improved authoritative proof mechanism, verified evidence units are written to legally binding consortium chain nodes, forming an unalterable evidence record. This evidence information is simultaneously pushed to regulatory agencies' nodes, supporting zero-knowledge proof technology to verify data authenticity while protecting privacy.

[0042] The digital twin engine loads the complete accident dataset and reconstructs a full-cycle evolutionary model of disaster occurrence and response. This reconstruction utilizes multi-scale fusion technology to couple macro-environmental parameters with micro-device states for analysis, aligning the time series data of each subsystem using a dynamic time warping algorithm. The reverse engineering module, based on an improved causal inference model, locates the initial node of threshold failure, identifies deviations in protection strategy execution, and marks performance degradation characteristics of key equipment.

[0043] The responsibility determination system initiates a multi-source evidence cross-validation process, integrating offline analysis results from physical and chemical testing equipment. A mobile detection unit collects combustion residue samples and identifies the composition of the igniting material through spectral feature comparison. The electrical safety diagnostic module analyzes device communication logs to reconstruct the control command sequence preceding the failure. The environmental simulation unit replicates the boundary conditions of the accident, such as temperature, humidity, and electromagnetic interference, to verify the rationality of the protection strategy. All verification results are correlated and analyzed using a knowledge graph engine, generating a multi-dimensional determination report that includes the allocation of responsibility weights.

[0044] The system's self-optimization mechanism initiates the federated learning model update process, breaking down accident cases into training sample sets. The optimization module compares the deviation between expected protection effectiveness and actual response outcomes to identify model flaws and calculate parameter corrections. The incremental training process utilizes a curriculum learning strategy, prioritizing high-confidence typical failure modes and gradually expanding to edge cases. The updated policy library is pushed to similar devices via a secure channel, simultaneously refreshing the digital twin model's deduction rule library.

[0045] The Judicial Evidence Packaging module structures data throughout the entire lifecycle, generating electronic files that meet international standards. These files include copies of original data, evidence verification reports, liability determinations, and system self-optimization records, all ensured by digital signatures. The evidence package utilizes a hierarchical access control system, enabling on-demand access to verification data by different roles, including regulators, operators, and manufacturers. Homomorphic encryption protects commercially sensitive information.

[0046] The system's final verification module performs a full-link replay test to verify the legal compliance of stored evidence data and the integrity of device status recovery. Simulated attack tests verify the blockchain network's tamper resistance, and fault injection techniques are used to verify the effectiveness of the system's self-optimization mechanisms. Upon passing final verification, the system resets all protective equipment to standby status, updates the dynamic threshold parameter library, and generates an operations and maintenance guidance document containing improvement recommendations, completing a fully closed-loop post-disaster response process.

[0047] Example 2

[0048] according to Figure 2As shown, within the intelligent charging station fire response system, this invention establishes a three-level progressive response mechanism, using a dynamic game-based protection (DGAP) model to precisely match the disaster development stage with the response strategy. Each level of response includes independent triggering conditions, response processes, and effectiveness verification, forming a closed-loop control chain of "monitoring-decision-execution-feedback," ensuring targeted control of fires throughout their inception, development, and outbreak stages.

[0049] The first-level response is initiated when the system detects that the local temperature difference exceeds the dynamic threshold. The positioning subsystem constructs a three-dimensional spatial coordinate system through the ultra-wideband base station array, combines thermal imaging data with the gas concentration gradient to lock the core area of ​​the fire source, and drives the servo mechanism to adjust the azimuth and pitch angles of the sprinkler. The sprinkler control module dynamically adjusts the coverage density and range of action of the fine water mist based on the real-time temperature field distribution model to form a continuous cooling zone in the target area. At the same time, the adjacent charging unit receives the pre-alarm signal, starts the cooling system enhancement mode and reduces the output power to form a preventive control zone with a radius of 3 meters. The system collects thermal radiation flux change data every 200 milliseconds after the disposal is started. If the temperature rise rate does not drop below the safety threshold within 5 seconds, the strategy upgrade evaluation is triggered.

[0050] The coordinated control mechanism for the first-level response is implemented through multi-bus data fusion. The spray flow controller receives spatial coordinate deviation compensation values ​​from the positioning subsystem and dynamically adjusts the nozzle array's opening combination. The environmental monitoring unit simultaneously analyzes the impact of air humidity on the evaporation rate of the water mist and optimizes the spray duration using a PID algorithm. When the system detects a localized electrical arc flash risk, it automatically superimposes a high-frequency pulsed fire extinguishing agent delivery mode, mixing nano-sized perfluorohexanone particles into the water mist for a dual suppression effect.

[0051] The secondary response is automatically activated when fire characteristics meet the deflagration precursor model. The physical isolation system operates in two stages: first, a high-precision servo mechanism drives the fire shield along the slide rails. The shape-memory alloy skeleton deforms upon contact with the high-temperature area, creating a curved seal between the edge of the shield and the ground. The electromagnetic confinement device then activates the superconducting coils, generating a circular magnetic field within a 0.5-meter radius of the fire source to suppress the movement of charged particles. An inert gas injection system deployed within the isolation space displaces oxygen at a specific flow gradient, reducing the oxygen concentration from 21% to below 12% within four seconds. Simultaneously, a negative pressure maintenance device is activated to prevent toxic smoke leakage.

[0052] The effectiveness verification phase of the secondary response includes dynamic parameter calibration. The gas analysis module monitors changes in gas composition within the isolation zone every 100 milliseconds. If a sudden increase in methane concentration or an abnormal rate of oxygen concentration recovery is detected, the system automatically applies active cooling to the aerogel insulation layer. Distributed strain sensors on the exterior of the fire shield monitor structural deformation in real time. If the impact pressure exceeds the designed threshold, the controlled collapse mechanism of the buffer layer's honeycomb structure is immediately triggered, absorbing over 80% of the explosion energy through material deformation.

[0053] The third-level response activates a linkage protection mechanism for thermal runaway chain reactions. The charging pile intelligent monitoring system continuously collects battery temperature data at a 100ms cycle. When an abnormal temperature rise rate is identified, a remote emergency stop command is triggered within 200ms via industrial Ethernet and the sound and light alarm device is activated. The asphyxiation fire extinguishing system is simultaneously activated. A glass fiber fire blanket driven by compressed air unfolds within 15 seconds to form a 6m×9m flame-retardant insulation layer with an oxygen barrier rate of over 95%. The chassis cooling system implements a three-level gradient cooling: first, room temperature water mist is sprayed through a 7cm branch pipe for 5 seconds of rapid cooling, then switched to a water mist containing corrosion inhibitor for targeted cooling, and finally a pulse spray with a 0.5s interval is used to maintain thermal stability.

[0054] The system dynamically adjusts the water mist flow rate using a pressure sensor with ±0.1 bar accuracy, ensuring full-load operation for at least 30 minutes. The emergency power supply unit immediately takes over powering critical equipment after an emergency stop, prioritizing the continued operation of positioning sensors and communication modules. The safety protection module simultaneously reconfigures the grounding network topology, guiding the safe release of fault energy along a pre-set path. Upon completion, it automatically generates an effectiveness evaluation report containing parameters such as fire blanket coverage completeness and coolant consumption, providing data support for system optimization.

[0055] The multi-level response system dynamically coordinates via a data bus, collecting key parameters such as the rate of change of the temperature field and gas concentration gradients in real time, and assessing the effectiveness of the response every 500 milliseconds. If the previous level of response fails to effectively contain the fire, the system automatically escalates the response level and optimizes the strategy combination. For example, if the oxygen concentration in the isolated area is detected to rise during the second-level response, the spray intensity is increased and the gas replacement time is extended. During the third-level response, if insufficient fire blanket coverage or abnormal cooling water pressure is detected, the system immediately activates a dynamic compensation mechanism: additional auxiliary spray points are added based on infrared positioning data, or a backup pump unit is switched to maintain water mist pressure. All operational commands are transmitted in real time via industrial Ethernet with a latency of less than 1ms, ensuring precise synchronization of audible and visual alarms, fire blanket deployment, and cooling spray, forming a complete "monitoring-isolation-cooling" response chain. During the response process, oxygen barrier efficiency and temperature reduction rate indicators are continuously verified. If the threshold is not reached, a secondary response sequence is automatically triggered until a stable and controlled state is achieved.

Claims

1. A defense-in-depth protection method for a charging station, characterized in that: The method comprises: The temperature field distribution of the target area is monitored in real time using a thermal imager. Multi-dimensional environmental parameters are collected simultaneously to generate a dynamic threshold judgment benchmark. When it is determined that the threshold needs to be updated, an environmental adaptive algorithm is triggered to optimize the threshold. A dynamic optimization function is constructed based on a game theory model that integrates the heat dissipation efficiency factor and the fluctuation characteristics of environmental parameters. An adaptive threshold interval is generated based on the real-time cooling system operating conditions and the differences in the chemical characteristics of the battery pack. When the battery pack temperature rise rate is detected to exceed the preset chemical system sensitivity threshold, a multi-level early warning mechanism matching the battery type is triggered. When the monitoring data exceeds the optimization threshold, the multispectral verification module is activated to confirm the fire characteristics. Based on the verification characteristics, the fire source is located and the direction of disaster spread is predicted. A three-level fire extinguishing strategy is dynamically adopted according to the spread direction. After the fire is extinguished, the disposal process data will be sharded and stored in a tamper-proof database. Based on the stored data, the cause chain of the accident will be traced back to form a closed-loop feedback, driving the continuous iterative optimization of monitoring thresholds and response strategies.

2. A charging station defense-in-depth protection method according to claim 1, characterized in that: The multi-level warning mechanism includes: in the primary warning stage, preventive control is implemented by adjusting the charging power and enhancing the heat dissipation intensity; the intermediate warning triggers in-depth diagnosis of the equipment status and locates the potential source of the fault in combination with the results of digital twin deduction; the advanced warning stage starts the preloading of the emergency protocol; the warning information is pushed to the operation and maintenance terminal in real time through the multi-channel redundant transmission system, and the standby status of the protective equipment is activated simultaneously.

3. A charging station defense-in-depth protection method according to claim 1 or 2, characterized in that: The method loads a corresponding temperature rise history model based on the chemical system type of the battery pack, calculates a threshold correction coefficient based on the spatial distribution deviation between real-time monitoring data and the historical model, and automatically lowers the temperature anomaly judgment threshold and increases the monitoring frequency when it is detected that the charge state deviation of the lithium iron phosphate battery pack reaches a preset safety margin boundary.

4. A charging station defense-in-depth protection method according to claim 1, characterized in that: The method generates a dynamic temperature field distribution based on the thermal radiation signals collected by a multi-spectral sensor array, and synchronously analyzes the evolution trend of thermal anomalies through a time-series prediction model of preset temperature gradient thresholds and temperature rise rates. When the radiation spectrum characteristics of the detected hot spot area match the preset lithium battery thermal runaway characteristic pattern, the ultra-wideband positioning data and visual positioning coordinates are integrated to construct a three-dimensional model of the fire source, and the environmental fluid parameters are loaded through the digital twin engine to simulate the disaster diffusion path in real time.

5. The method for defense-in-depth protection of a charging station according to claim 1, characterized in that: The three-level fire extinguishing strategy includes: first-level response, when it is detected that the local temperature mutation exceeds the preset threshold, the high-precision positioning system is activated to control the sprinkler device to perform spatial directional fire extinguishing, and the spray angle and flow are adjusted in real time to match the dynamic position of the fire source; second-level response, when the spectral characteristics of the deflagration are identified, the high-reflectivity fireproof cover is synchronously activated to cover the fire source area and an electromagnetic isolation barrier is generated to block the thermal radiation propagation path; third-level response, when the monitoring system determines the risk of thermal runaway, the linkage protection mechanism is triggered, the energy input is cut off according to the emergency stop command of the charging pile, the sound and light alarm device is linked to send a warning signal, the asphyxiation fire extinguishing system is activated according to the location of the fire source to cover the target area, and the gradient cooling process is synchronously executed.

6. A charging station defense-in-depth protection method according to claim 5, characterized in that: The high-precision positioning system controls the sprinkler device, including: building a dynamic flow regulation system based on multi-sensor fusion, achieving real-time calibration of the spray angle through the coordinated control of the solenoid valve array and the millimeter-wave radar, and dynamically matching the two-phase mixing ratio of fine water mist and gas fire extinguishing agent based on the energy density of the fire source.

7. A charging station defense-in-depth protection method according to claim 5, characterized in that: The high-reflectivity fireproof cover comprises a multi-layer composite protective structure, including an outer layer of high thermal conductivity composite coating, a middle layer of adaptive deformation skeleton and an inner layer of energy absorption layer. The centimeter-level positioning system drives the robotic arm to perform rapid packaging of the fireproof cover, while triggering the magnetic sealing array to create an oxygen-deficient suppression environment.

8. The method for defense-in-depth protection of a charging station according to claim 5, characterized in that: The linkage protection mechanism triggers the fire blanket deployment mechanism according to the emergency stop command, so that the flame-retardant isolation layer completely covers the fire source area; starts the cooling process according to the fire source coverage status, and executes preliminary water mist injection, corrosion inhibitor cooling and pulse maintenance cooling in sequence; constructs a closed-loop control signal based on real-time fire data, and dynamically coordinates the action timing of the fire blanket deployment and the cooling system; and activates the emergency power supply unit according to the energy input interruption status to maintain the continuous operation of key protection components.

9. The method for defense-in-depth protection of a charging station according to claim 1, characterized in that: The method constructs a multimodal data sharding storage architecture, binds sensor data and operation records with three-dimensional features through spatiotemporal anchoring technology, generates sharding keys using a quantum-resistant encryption algorithm, and writes verification hash values ​​into blockchain nodes based on a consensus mechanism, forming an unalterable chain of evidence with judicial effect.

10. A charging station defense-in-depth protection method according to claim 1 or 9, characterized in that: The method reconstructs a multi-dimensional evolution model of the accident scenario through a digital twin engine, identifies threshold failure nodes and response strategy defects based on a causal reasoning algorithm, converts the analysis results into training samples and injects them into the defense model, and generates an upgrade instruction package containing an optimized parameter set and a disposal plan, which is pushed to associated devices.

Images