Optimized layout method and emergency response method for sensors in passenger compartment of hydrogen energy vehicle

Abstract

The invention relates to an optimized layout method and an emergency response method for sensors in a passenger compartment of a hydrogen energy vehicle, and the method comprises the steps: identifying a hydrogen leakage risk, and setting a leakage working condition; performing leakage simulation based on the leakage working condition; based on a leakage simulation result, comprehensive validity of three dimensions of space, time and frequency is integrated, and an optimal sensor arrangement position is determined; arranging a sensor based on the optimal sensor arrangement position, collecting a concentration signal in real time, and executing an emergency response strategy. According to the invention, automatic linkage control from hydrogen leakage detection, alarm prompt, ventilation, exhaust and hydrogen source cut-off is realized; the defects that hydrogen leakage detection of an existing hydrogen energy vehicle mainly focuses on the outer portion of the vehicle or the equipment level, and systematic detection and emergency disposal schemes for the interior of a passenger compartment are lacked are overcome, and efficient detection and rapid and safe response to hydrogen leakage in the passenger compartment are achieved.

Description

Technical Field

[0001] This invention relates to the fields of hydrogen safety and hydrogen fuel cell vehicle technology, and in particular to a method for optimizing the layout of sensors in the passenger compartment of a hydrogen fuel cell vehicle and an emergency response method. Background Technology

[0002] Hydrogen energy, as a clean and efficient energy source, is rapidly expanding its application in the transportation sector, and hydrogen fuel cell vehicles have become an important development direction for future transportation energy transformation. Due to its small molecular weight, high diffusion coefficient, and low density, hydrogen easily diffuses in the air and forms a flammable mixture after leakage, especially in enclosed or semi-enclosed spaces, where it is prone to the accumulation of localized flammable clouds. With the rapid promotion of hydrogen fuel cell vehicles, the safety risks of onboard hydrogen system leaks are becoming increasingly prominent. If hydrogen leaks into the relatively enclosed space of the passenger compartment, and is not detected and dealt with in a timely manner, it could potentially lead to serious safety accidents and catastrophic consequences. Therefore, research on hydrogen leak detection technologies and the design of emergency response strategies for hydrogen leaks in the passenger compartment of hydrogen-powered vehicles are crucial to ensuring passenger safety.

[0003] Current hydrogen leak detection technologies primarily focus on external areas or adjacent to equipment, such as hydrogen storage systems, pipeline interfaces, engine compartments, and vehicle chassis. Numerous patents have been published (e.g., CN202310980433.7, CN202410740221.6) addressing sensor placement, coordinated hydrogen source cutoff, and external alarm mechanisms for these areas. However, these solutions mainly focus on external vehicle protection, with relatively insufficient research on specialized detection and active ventilation strategies for the passenger compartment, resulting in a lack of systematic solutions. Furthermore, patents such as CN202310974117.9 and CN202111518585.2 primarily address the optimization of gas detector layouts in large, confined spaces such as parking lots and hydrogen production plants. CN202111431885.7 relies solely on time-scale optimization of sensor placement, neglecting spatial and other physical factors, which can lead to deviations from optimal layout results in complex scenarios. Furthermore, CN202411366302.0 relies solely on concentration thresholds for leak detection, which is prone to false alarms or missed alarms and fails to meet the high-reliability industrial monitoring requirements. None of the aforementioned studies addressed the optimization of sensor placement and emergency response control mechanisms in the confined and complex airflow environment of the passenger compartment of hydrogen-powered vehicles.

[0004] Therefore, there is an urgent need to establish an efficient hydrogen leak sensor optimization layout method and supporting emergency response strategy for the passenger compartment of hydrogen fuel cell vehicles, so as to achieve rapid detection and safe handling of hydrogen leaks in the passenger compartment. Summary of the Invention

[0005] The purpose of this invention is to propose an optimized sensor layout method and emergency response method for the passenger compartment of hydrogen fuel cell vehicles, addressing the problems existing in the prior art. This method first identifies potential leakage sources in hydrogen fuel cell vehicles, particularly the fuel cell stack, which could lead to hydrogen entering the passenger compartment after a leak. Simulation calculation conditions are set based on the structural parameters and operating conditions of the leaking component. High-precision transient numerical simulations are conducted for different leakage diffusion conditions to obtain the hydrogen concentration distribution characteristics within the passenger compartment. Based on the simulation results, the comprehensive effectiveness of the sensors in the time, space, and frequency dimensions is calculated, thereby determining the optimal sensor placement. Regarding hydrogen leak detection, this invention constructs a multi-layered fusion-based leak detection model. This model effectively reduces the false alarm rate and missed alarm rate through joint detection by a rule layer, a confidence fusion layer, and an intelligent correction layer, ensuring accurate identification of hydrogen leak conditions. In terms of emergency response strategies, this invention designs a multi-level emergency response system, realizing automated linkage control from hydrogen leak detection and alarm prompts to ventilation, exhaust, and hydrogen source cutoff. This invention overcomes the shortcomings of existing hydrogen leak detection methods for hydrogen-powered vehicles, which mainly focus on the exterior of the vehicle or the equipment level and lack a systematic detection and emergency response plan for the passenger compartment. It achieves efficient detection and rapid safety response to hydrogen leaks in the passenger compartment.

[0006] To achieve the above objectives, the present invention provides the following solution:

[0007] A method for optimizing the layout of sensors in the passenger compartment of a hydrogen fuel cell vehicle includes:

[0008] Identify hydrogen leak risks and set leak conditions;

[0009] Based on the aforementioned leakage conditions, a leakage simulation was performed;

[0010] Based on the leakage simulation results, the optimal sensor placement location is determined by comprehensively considering the effectiveness across three dimensions: space, time, and frequency.

[0011] Optionally, the leakage conditions can be set as follows:

[0012] Based on the specific system parameters of the hydrogen fuel cell vehicle, potential leaking components are identified; wherein, the potential leaking components include: fuel cell stack, valves and connectors;

[0013] Based on the technical parameters at the leaking component, the potential leakage flow rate is calculated using real gas thermodynamics; wherein the technical parameters include: pressure and pipe diameter;

[0014] The maximum leakage size within the highest order of magnitude of leakage frequency is selected as the high-frequency leakage size, and it is determined together with the maximum possible leakage flow rate as the simulation calculation condition.

[0015] Optionally, leakage simulation includes:

[0016] The equivalent outlet diameter is calculated using a virtual nozzle model.

[0017] A simplified three-dimensional geometric model is established based on the actual vehicle structure. Local mesh refinement is performed on key areas, and a boundary layer mesh is established. The key areas include, for example, the area near the leakage hole.

[0018] Monitoring points are set up in the main areas of the crew compartment to monitor the distribution and trend of hydrogen concentration in the compartment in real time; the main areas include: the area near the leak, the upper part of the compartment and the crew breathing area;

[0019] Independence verification was carried out by selecting different grid densities, and the simulation results were compared with experimental or literature data for verification.

[0020] Transient simulation calculations were performed using Fluent software.

[0021] Optionally, determining the optimal sensor placement location includes:

[0022] Post-processing of simulation results;

[0023] Based on the post-processing results, the coverage effectiveness of each monitoring point in the spatial, temporal and frequency dimensions is calculated respectively.

[0024] The optimal sensor placement location is determined by comprehensively considering the effectiveness of space, time, and frequency.

[0025] Optionally, post-processing of the simulation results includes:

[0026] Based on the simulation results, hydrogen mole fraction cloud maps and concentration change curves at monitoring points were generated at different times, and the volume and mass of combustible gas clouds in the crew cabin at different times were statistically analyzed.

[0027] Optionally, the coverage effectiveness of each monitoring point in the spatial, temporal, and frequency dimensions is calculated separately, including:

[0028] For different leakage conditions, for each monitoring point, the proportion covered by different mole fraction threshold contours at different times is calculated, and the spatial coverage effectiveness of each monitoring point is calculated by weighted summation.

[0029] Based on the concentration change curves of each monitoring point under different working conditions, the proportion of the time period during which each threshold can be monitored to the total leakage time is calculated, and the time coverage effectiveness of each monitoring point is calculated by weighted summation.

[0030] Based on the concentration change curves of monitoring points under different operating conditions, the number of times the concentration exceeds a specific threshold is counted, and the frequency coverage effectiveness of each monitoring point is calculated by weighted summation.

[0031] This invention also proposes an emergency response method for the passenger compartment of a hydrogen fuel cell vehicle, comprising:

[0032] Based on the optimal sensor layout method for optimizing the sensor layout in the passenger compartment of the hydrogen fuel cell vehicle, the sensors are arranged to collect concentration signals in real time.

[0033] The concentration signal acquired in real time is processed by moving average filtering;

[0034] Based on the filtered signal, a multi-level leakage determination strategy is adopted. Leakage is determined by identifying the nonlinear relationship between the temporal characteristics of sensor signals and environmental conditions under different operating conditions.

[0035] Based on the judgment results, a hydrogen leak alarm decision is made;

[0036] Based on hydrogen leak alarm decision-making, it outputs multi-level emergency alarm responses and performs adaptive cancellation and continuous monitoring.

[0037] Optionally, the multi-level leakage determination strategy includes:

[0038] Level 1 rapid leakage detection: rapid initial judgment is made based on the concentration thresholds and change rates of multiple sensors and the consistency of multiple points;

[0039] Secondary confidence fusion judgment: Based on the fusion of multi-point detection results and spatial arrangement differences, a weighted Bayesian confidence model is established to calculate the local leakage probability for each sensor point and output a confidence leakage signal;

[0040] Three-level intelligent correction: Based on the first-level rapid leakage judgment and the second-level confidence fusion judgment, a long short-term memory neural network model is further introduced to dynamically correct and optimize the leakage confidence.

[0041] Optionally, introducing a long short-term memory neural network model includes:

[0042] Input variable design: The model input variables include concentration signal type, dynamic feature type, environmental state type, and driving state type;

[0043] Training sample construction: The training samples include numerical simulation data and experimental data; among them, the numerical simulation data is based on the leakage diffusion transient simulation results, and the hydrogen concentration time series of each monitoring point under different states is extracted; the experimental data is obtained based on a typical cabin leakage test bench.

[0044] Model training and deployment: The model adopts a two-layer LSTM structure, with hidden nodes, optimization algorithms, and adaptive learning rates set for each layer; the loss function uses cross-entropy or mean squared error; after the model is trained in an offline environment, the mapping function is obtained and exported as a lightweight model for deployment.

[0045] Optionally, the output of multi-level emergency alarm responses includes:

[0046] Level 1 Alarm Response: Upon receiving a signal indicating a leak has been detected, the vehicle alarm system is immediately activated, including the instrument panel warning lights and buzzer, and a notification message is displayed on the vehicle display screen or central control interface; at the same time, the sensor data frames for the most recent preset duration are frozen for subsequent fault tracing and data reporting.

[0047] Level 2 ventilation response: After the alarm is issued, the vehicle's operating status is determined, and preset ventilation actions are automatically executed according to the vehicle's operating status;

[0048] Three-level linkage control stage: When the air conditioner is in the air supply or recirculation mode, immediately close the air supply vents and recirculation valve; if the external circulation mode is already on, temporarily maintain this state; package and upload the leakage time, sensor number, and maximum concentration value information to the vehicle safety control center or cloud monitoring platform.

[0049] The beneficial effects of this invention are as follows:

[0050] 1. This invention proposes a sensor optimization layout method and emergency response system for the passenger compartment of hydrogen fuel cell vehicles. It can make targeted deployment based on the vehicle structure and potential leak sources, thereby significantly improving the detection and response speed of hydrogen leaks in the compartment, extending the escape time for occupants and reducing the risk of accidents.

[0051] 2. Regarding the selection of leakage simulation conditions, this invention is based on leakage frequency data released by authoritative institutions, and prioritizes the selection of high-frequency representative leakage size and maximum leakage size to ensure the representativeness of the simulation conditions and avoid the waste of computing resources caused by indiscriminately expanding the range of conditions, thereby significantly improving the efficiency of simulation and evaluation.

[0052] 3. This invention proposes a sensor layout evaluation and optimization method based on the three-dimensional coverage effectiveness of time-space-frequency. This method comprehensively evaluates the effectiveness of each monitoring point under different operating conditions through weighted comprehensive assessment, ensuring stable and reliable detection results with a small number of sensors, thereby reducing system costs and improving engineering feasibility.

[0053] 4. The multi-layer fusion leakage detection algorithm constructed in this invention (including a rule layer, a Bayesian confidence fusion layer, and a lightweight machine learning correction layer) balances response speed and robustness: the rule layer enables rapid early warning, the confidence fusion layer reduces false alarms and missed alarms, and the LSTM layer intelligently corrects complex time-series signals and environmental interference, thereby comprehensively improving the accuracy of detection and the reliability of the system.

[0054] 5. This invention proposes a multi-level emergency response strategy (including a first-level display alarm system; a second-level ventilation response and a third-level linkage control), which can quickly alert occupants after a leak is confirmed, and form effective ventilation to suppress the accumulation of flammable clouds in the cabin, ensuring the safety of the vehicle and personnel; at the same time, combined with the leak event locking and cloud reporting functions, it is conducive to post-event tracing and safety improvement. Attached Figure Description

[0055] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0056] Figure 1 This is a flowchart of a sensor optimization layout method and emergency response strategy for the passenger compartment of a hydrogen fuel cell vehicle, according to an embodiment of the present invention.

[0057] Figure 2 This is a schematic diagram of the simulation model and monitoring point layout for hydrogen leakage in the passenger compartment of a hydrogen-powered vehicle, as described in an embodiment of the present invention.

[0058] Figure 3 This is a schematic diagram showing the distribution of combustible gas clouds and gas velocity streamlines in the passenger compartment at different times according to an embodiment of the present invention.

[0059] Figure 4 The figures show the volume and mass changes of the combustible gas cloud inside the passenger compartment under different emergency measures according to embodiments of the present invention.

[0060] Figure 5 This is an overall structural diagram of the emergency response system for the passenger compartment of a hydrogen fuel cell vehicle according to an embodiment of the present invention;

[0061] Figure 6 This is a flowchart illustrating the logic of the multi-layer fusion leakage detection algorithm according to an embodiment of the present invention.

[0062] Figure 7 This is a timing diagram illustrating the execution of a multi-level emergency response according to an embodiment of the present invention. Detailed Implementation

[0063] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0064] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0065] like Figure 1 As shown in the figure, this embodiment proposes a method for optimizing the layout of sensors in the passenger compartment of a hydrogen fuel cell vehicle, including:

[0066] S1. Identify hydrogen leak risks and set leak conditions;

[0067] S2. Based on the aforementioned leakage conditions, perform leakage simulation;

[0068] S3. Based on the leakage simulation results, the optimal sensor placement location is determined by comprehensively considering the effectiveness across the three dimensions of space, time, and frequency.

[0069] Specifically, in this embodiment, S1 hydrogen leakage risk identification and leakage condition setting:

[0070] Based on the specific system parameters of the hydrogen vehicle, S11 identifies potential leaking components such as fuel cell stacks and hydrogen supply pipeline valves and joints that could lead to hydrogen entering the passenger compartment.

[0071] S12 calculates the maximum possible leakage flow rate based on the technical parameters (pressure, pipe diameter, etc.) at the leaking component and according to the real gas thermodynamic relationship. The calculation formula is as follows:

[0072] (1);

[0073] In the formula: Represents leakage flow. The flow coefficient is represented by A, and the leakage area is represented by A. Represents the density of the gas leak. Represents the enthalpy of the upstream stagnant gas. This represents the enthalpy of the gas at the leak point. This represents the pressure at which the leak source stops. The entropy of the leakage source is stagnant. This represents the pressure at the leak point.

[0074] Based on the type of leaking component, S13 refers to the leakage frequency database of hydrogen-related components published by Sandia National Laboratories in the United States or authoritative domestic institutions, selects the maximum leakage size within the highest order of magnitude of leakage frequency as the high-frequency leakage size, and determines it together with the maximum leakage size as the simulation calculation condition.

[0075] Figure 2 This embodiment demonstrates a simulation model of hydrogen leakage in the passenger compartment of a hydrogen-powered vehicle and the layout of monitoring points.

[0076] Specifically, in this embodiment, S2 numerical modeling and leakage simulation are performed:

[0077] To improve computational efficiency and avoid computational instability caused by supersonic jet shock waves at the actual leak outlet, S21 employs an equivalent leak outlet technique. The equivalent outlet diameter can be calculated using a well-established "virtual nozzle" model, such as the Birch 1984 model, whose calculation formula is as follows:

[0078] (2);

[0079] In the formula: Represents the virtual nozzle diameter. Represents the actual leak diameter. The specific heat ratio representing the leaked gas. Represents the stagnation temperature. Represents environmental pressure, Represents ambient temperature.

[0080] S22 establishes a simplified three-dimensional geometric model based on the actual vehicle structure. To balance computational accuracy and efficiency, localized mesh refinement is applied to key areas (such as near leak holes), and boundary layer meshes are established to accurately capture shear flow characteristics.

[0081] S23 sets up monitoring points in key areas within the crew compartment (near the leak point, the upper part of the compartment, and the crew breathing area, etc.) to monitor the hydrogen concentration distribution and trends in the compartment in real time. The specific setup of the monitoring points is shown in Figure 2.

[0082] To eliminate the influence of mesh scale on the calculation results, S24 selected different mesh densities (coarse, medium, and fine) to conduct independence verification. Subsequently, the simulation results were compared with experimental or literature data to ensure the reliability and accuracy of the model.

[0083] S25 utilizes Fluent software for transient simulation calculations. An implicit time integration scheme is employed to ensure computational stability, with the time step determined based on flow field characteristics and the Courant number criterion to accurately capture the transient process of high-speed gas leakage. To balance data resolution and storage efficiency, a reasonable data retention interval is set to ensure the continuity and accuracy of post-processing analysis.

[0084] Specifically, in this embodiment, the layout of sensor S3 is optimized:

[0085] S31 performs post-processing on the simulation results, exporting the calculation data to common file formats such as CSV files. It generates hydrogen mole fraction cloud maps at different times, concentration change curves at monitoring points, and statistically analyzes the volume and mass of combustible gas clouds inside the crew cabin at different times. For example... Figure 3 (Charts of combustible cloud and velocity streamlines inside the vehicle cabin at 0.5 s and 10 s) and Figure 4 (Curves of flammable gas cloud changes under different emergency ventilation methods) By closing the air conditioning vents to block the entry of flammable gas and opening the vehicle windows to enhance the escape of flammable gas, the flammable cloud in the passenger compartment can be effectively reduced by 20% in a short period of time.

[0086] The S32 sensor effectiveness calculation, based on the post-processing results, calculates the coverage effectiveness of each monitoring point in the spatial, temporal, and frequency dimensions.

[0087] (1) For different leakage conditions, for each monitoring point, calculate the proportion covered by different mole fraction thresholds (such as 1% vol., 2% vol., 4% vol., etc.) at different times, and calculate the spatial coverage effectiveness of each monitoring point by weighted summation. The calculation formula is as follows:

[0088] (3);

[0089] In the formula: N represents the number of leakage conditions, and M represents the number of concentration thresholds. These are the weighting coefficients corresponding to each threshold. The coverage indicator function takes a value of 1 when monitoring point i is covered by the kth threshold concentration region in operating condition j, and 0 otherwise.

[0090] (2) Based on the concentration change curves of each monitoring point under different operating conditions, calculate the proportion of the time period during which each threshold can be monitored to the total leakage time, and calculate the time coverage effectiveness of each monitoring point by weighted summation. The calculation formula is as follows:

[0091] (4);

[0092] In the formula: Let j be the duration of leakage. The threshold function is set to 1 when the hydrogen concentration at monitoring point i at time t is higher than the k-th threshold, and 0 otherwise.

[0093] (3) Similarly, based on the concentration change curves of the monitoring points under different operating conditions, the number of times they exceeded a specific threshold was counted, and the frequency coverage effectiveness of each monitoring point was calculated by weighted summation, as shown in the following formula:

[0094] (5);

[0095] In the formula: Let be the frequency at which the concentration at monitoring point i exceeds the k-th threshold under operating condition j.

[0096] S33 sensor optimized layout:

[0097] The optimal sensor placement location is determined by comprehensively considering the effectiveness across three dimensions: space, time, and frequency. The formula for calculating the overall effectiveness is as follows:

[0098] (6);

[0099] In the formula: These are the weighting coefficients.

[0100] like Figure 5 As shown, this embodiment also proposes an emergency response method for the passenger compartment of a hydrogen fuel cell vehicle, including:

[0101] Based on the optimal sensor layout method determined by the sensor optimization layout method in the passenger compartment of hydrogen fuel cell vehicles, the sensors are arranged to collect concentration signals in real time.

[0102] The concentration signal acquired in real time is processed by moving average filtering;

[0103] Based on the filtered signal, a multi-level leakage determination strategy is adopted. Leakage is determined by identifying the nonlinear relationship between the temporal characteristics of sensor signals and environmental conditions under different operating conditions.

[0104] Based on the judgment results, a hydrogen leak alarm decision is made;

[0105] Based on hydrogen leak alarm decision-making, it outputs multi-level emergency alarm responses and performs adaptive cancellation and continuous monitoring.

[0106] Specifically, in this embodiment, the S4 emergency response includes:

[0107] Following the optimized sensor layout, hydrogen sensors were installed in the crew compartment of the S41 to collect concentration signals in real time. It is transmitted to the vehicle control unit (ECU) via the Controller Area Network (CAN) or Local Interconnect Network Bus (LIN).

[0108] The S42 vehicle control unit performs a moving average filter on the acquired signals to suppress transient fluctuations;

[0109] (7);

[0110] In the formula: This is the predicted concentration value. The concentration measurement value is used, with a smoothing coefficient α preferably between 0.3 and 0.7, and a time step. The preferred time is 0.1 to 0.5 s.

[0111] S43 as Figure 6As shown, a multi-level leakage detection strategy is adopted to reduce the probability of false alarms and missed alarms from sensors. This strategy improves the accuracy and robustness of leakage detection by identifying the nonlinear relationship between the temporal characteristics of sensor signals and environmental conditions under different operating conditions.

[0112] Level 1 rapid leakage detection (rule layer): rapid initial judgment based on the concentration thresholds and change rates of multiple sensors;

[0113] (1) Concentration threshold determination: When the concentration of any sensor exceeds the set threshold (preferably 1% vol.) and the duration is... When the time is ≥0.5 s, output a leak warning flag S1=1.

[0114] (2) Determination of concentration change rate, the time gradient is calculated as follows:

[0115] (8);

[0116] Set gradient threshold ,when Output the dynamic rising flag S2=1.

[0117] (3) Multi-point consistency determination: When the following conditions are met, the preliminary leakage probability signal is output. :

[0118] (9);

[0119] In the formula: Indicates monitoring point Warning signal S1, Indicates monitoring point The warning signal S2.

[0120] Second-level confidence fusion decision (Bayesian layer): Based on the fusion of multi-point detection results and spatial arrangement differences, a weighted Bayesian confidence model is established to enhance the reliability of the decision. For each sensor point i, its local leakage probability is calculated:

[0121] (10);

[0122] In the formula: The steepness coefficient is preferred (10-20). The confidence threshold (preferably 0.1% volume ratio)

[0123] Comprehensive probability calculation formula:

[0124] (11);

[0125] like If so, the output confidence leakage signal S3=1.

[0126] Level 3 Intelligent Correction (Machine Learning Layer): Building upon the rule-based and Bayesian layers, a Long Short-Term Memory (LSTM) neural network model is further introduced to dynamically correct and optimize the leakage confidence. The model design includes the following:

[0127] (1) Input variable design: The model input variables include concentration signal type (sensor concentration value), dynamic feature type (concentration change rate, gradient direction), environmental state type (in-vehicle temperature, humidity, air conditioning mode (internal circulation / external circulation), driving state type (vehicle speed, window status), etc. Under different vehicle models, the number of input variables can be flexibly adjusted according to the sensor layout and in-vehicle control parameters, but no less than 6 core variables should be maintained to ensure the generalization performance and robustness of the model.

[0128] (2) Training sample construction: The training samples are generated by fusing numerical simulation data and experimental data. The numerical simulation samples are based on the transient simulation results of leakage diffusion obtained in steps S1 to S3, and the hydrogen concentration time series of each monitoring point under different states are extracted; the experimental verification samples can build a typical cabin leakage test bench, and collect the real concentration response curve under safe conditions (low-pressure hydrogen or inert gas simulation) for model calibration and transfer learning.

[0129] (3) Model Training and Deployment: The model adopts a two-layer LSTM structure, with 32-64 hidden nodes in each layer; the optimization algorithm used is Adam, with an adaptive learning rate of 0.001; the loss function is either cross-entropy or mean squared error. After the model is trained offline, the mapping function is obtained and exported as a lightweight model for deployment to the vehicle ECU.

[0130] (12);

[0131] The input variables include cabin temperature Tin, vehicle speed V, and air conditioning mode. Wait, the model outputs the leakage confidence level. .

[0132] Based on the comprehensive confidence level determination, S44 outputs a hydrogen leak alarm decision.

[0133] Overall confidence level:

[0134] (13);

[0135] In the formula: Weighted values. Final leak confidence level. With threshold Compare:

[0136] (14);

[0137] When a hydrogen leak is confirmed, the system automatically triggers the emergency response module to execute corresponding actions. See [link / reference]. Figure 7 The present invention provides a timing diagram of the execution of a multi-level emergency response, including a three-level logic control process.

[0138] Upon receiving a leak detection signal, the S45 control unit immediately activates the multi-level emergency alarm system.

[0139] Level 1 Alarm Response: Upon receiving a leak detection signal, the control unit immediately activates the vehicle alarm system, including turning on the dashboard warning lights and buzzer (preferably with a flashing frequency of 2–4 Hz and a buzzer sound level of 75–85 dB), and displaying a "Hydrogen Leak Warning" message on the vehicle display screen or central control interface to guide occupants to remain calm and refrain from operating open flames. Simultaneously, the system automatically freezes sensor data frames from the last 2 seconds for subsequent fault tracing and data reporting.

[0140] Level 2 Ventilation Response: After the alarm is issued, the vehicle control unit determines the vehicle's operating status and automatically executes preset ventilation actions based on that status. When the sunroof is closed, the system automatically enters the "rear tilt ventilation mode," with an opening angle preferably between 10° and 15°. If the sunroof is a sliding type, the "rear slide fully open mode" is prioritized. If the vehicle is detected to be in motion, the vehicle control unit will prioritize opening the rear negative pressure side windows to utilize the airflow during vehicle movement to create negative pressure, accelerating the dilution and expulsion of gases from the cabin.

[0141] The three-level linkage control stage: When the air conditioner is in ventilation or recirculation mode, the system immediately closes the air conditioner vents and recirculation valve to prevent hydrogen from continuously entering the passenger compartment and causing the concentration to rise; if the external circulation mode is already on, it will be temporarily maintained to use external fresh air to dilute the hydrogen concentration in the cabin; finally, the system will package and upload information such as leakage time, sensor number, and maximum concentration value to the vehicle safety control center or cloud monitoring platform for subsequent safety diagnosis and statistical analysis.

[0142] S46 Adaptive Deactivation and Continuous Monitoring:

[0143] When the maximum hydrogen concentration inside the cabin is below 0.1%, the system issues a warning signal, allowing the sunroof and windows to be closed and the alarm to be deactivated. At the same time, it enters continuous monitoring mode to prevent recurrence of leaks.

[0144] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for optimizing sensor placement in a hydrogen vehicle passenger cabin, comprising: The method comprises the following steps: identifying hydrogen leakage risk and setting leakage working condition; based on the leakage working condition, carrying out leakage simulation; based on the leakage simulation result, comprehensively considering the comprehensive effective degree of space, time and frequency three dimensions, determining the optimal sensor arrangement position.

2. The method of claim 1, wherein, Setting leakage working condition includes: According to the specific system parameters of the hydrogen energy vehicle, the potential leakage components are identified; wherein, the potential leakage components include: fuel cell stack, valve and joint; According to the technical parameters at the leakage component, the possible leakage flow is calculated by the real gas thermodynamic relation; wherein, the technical parameters include: pressure, pipe diameter; Select the maximum leakage size in the highest number of leakage frequency as the high frequency leakage size, and determine it together with the maximum possible leakage flow as the simulation calculation working condition.

3. The method of claim 1, wherein, Leakage simulation includes: Using virtual nozzle model calculation method to calculate equivalent outlet diameter; Based on the actual vehicle structure, a simplified three-dimensional geometric model is established, the key areas are locally encrypted grid division, and the boundary layer grid is established; wherein, the key areas include: such as near the leakage hole; Setting monitoring points in the main area of the passenger cabin to monitor the hydrogen concentration distribution and change trend in the cabin in real time; wherein, the main area includes: near the leakage hole, the upper area in the cabin and the passenger breathing area; Selecting different grid densities to carry out independence verification, and comparing the simulation results with experimental or literature data for verification; Using Fluent software to carry out transient simulation calculation.

4. The method of claim 1, wherein, Determination of the optimal sensor arrangement position includes: Post-processing the simulation calculation result; Based on the post-processing result, the coverage effective degree of each monitoring point in the space, time and frequency dimensions is calculated respectively; Comprehensively considering the comprehensive effective degree of space, time and frequency three dimensions, the optimal sensor arrangement position is determined.

5. The method of claim 4, wherein, The post-processing of the simulation calculation result includes: Based on the simulation calculation result, hydrogen molar fraction cloud chart at different time, monitoring point concentration change curve and the volume and mass of combustible gas cloud in the passenger cabin at different time are generated.

6. The method of claim 4, wherein, Respectively calculating the coverage effective degree of each monitoring point in the space, time and frequency dimensions includes: For different leakage working conditions, for each monitoring point, the proportion of being covered by different molar fraction threshold contours at different time is calculated, and the spatial coverage effective degree of each monitoring point is calculated by weighted summation; According to the concentration change curve of each monitoring point under different working conditions, the proportion of the time period that can monitor each threshold to the total leakage time is calculated, and the time coverage effective degree of each monitoring point is calculated by weighted summation; According to the concentration change curve of the monitoring point under different working conditions, the number of times that exceeds a certain threshold is counted, and the frequency coverage effective degree of each monitoring point is calculated by weighted summation.

7. A method for emergency response in a hydrogen energy vehicle passenger cabin, characterized in that, The method comprises the following steps: Based on the optimal sensor arrangement position determined by the hydrogen energy vehicle passenger cabin sensor optimization layout method of any one of claims 1-6, arranging sensors to collect concentration signals in real time; The concentration signals collected in real time are subjected to moving average filtering processing; Based on the filtered signal, a multi-stage leakage judgment strategy is adopted to judge the leakage by identifying the nonlinear relationship between the time sequence characteristics of the sensor signal and the environmental state under different working conditions; According to the determination result, a hydrogen leakage alarm decision is determined; Based on the hydrogen leakage alarm decision, a multi-level emergency alarm response is output, and adaptive release and continuous monitoring are performed.

8. The hydrogen energy vehicle passenger cabin emergency response method according to claim 7, characterized by, The multi-level leakage determination strategy includes: First-level rapid leakage determination: rapid preliminary judgment is made according to the concentration threshold and change rate of multiple sensors and multi-point consistency; Second-level confidence fusion determination: based on the fusion of multi-point detection results and spatial arrangement differences, a weighted Bayesian confidence model is established to calculate the local leakage probability of each sensor point and output the confidence leakage signal; Third-level intelligent correction: based on the first-level rapid leakage determination and the second-level confidence fusion determination, a long short-term memory neural network model is further introduced to dynamically correct and optimize the leakage confidence.

9. The hydrogen energy vehicle passenger cabin emergency response method according to claim 7, characterized by, The long short-term memory neural network model includes: Input variable design: the model input variables include concentration signal type, dynamic feature type, environmental state type, and driving state type; Training sample construction: the training samples include: numerical simulation data and experimental data fusion; wherein, the numerical simulation data is based on the leakage diffusion transient simulation results, and the hydrogen concentration time series of each monitoring point under different states is extracted, and the experimental data is obtained based on the typical cabin leakage test bench; Model training and deployment: the model adopts a double-layer LSTM structure, sets the number of hidden nodes of each layer, optimizes the algorithm, and sets the adaptive learning rate; the loss function adopts cross-entropy or mean square error; after the model is trained in an offline environment, the mapping function is obtained and exported as a lightweight model for deployment.

10. The hydrogen energy vehicle passenger cabin emergency response method according to claim 7, characterized by, The output multi-level emergency alarm response includes: First-level alarm response: when receiving the leakage established signal, the vehicle-mounted alarm system is immediately activated, including starting the instrument panel warning light and buzzer, and popping up prompt information on the vehicle display screen or central control interface; at the same time, the sensor data frame in the recent preset time period is frozen for subsequent fault tracing and data reporting; Second-level ventilation response: after the alarm is issued, the vehicle operating state is determined, and the preset ventilation action is automatically executed according to the vehicle operating state; Third-level linkage control stage: when the air conditioner is in the supply air or internal circulation mode, the air conditioner supply air outlet and internal circulation valve are immediately closed; if the external circulation mode is opened, the state is temporarily maintained; the leakage time, sensor number, and maximum concentration value information are packaged and uploaded to the vehicle safety control center or cloud monitoring platform.

Images