A sensor-based hydrogen fuel cell tipping guard early warning processing method and system

Abstract

The application relates to the technical field of hydrogen fuel cells and discloses a hydrogen fuel cell toppling protection early warning processing method and system based on sensors, wherein a Kalman filtering algorithm is adopted to carry out denoising processing on collected data, and then normalization processing is carried out to obtain pretreatment data; the weights of a tilting angle, acceleration, material state and pipeline pressure are dynamically allocated based on a MOF material state, real-time comprehensive risk characteristic values are calculated according to the pretreatment data through a weighted fusion algorithm; the pretreatment data of the past 5 seconds are input into a Transform-TCN hybrid prediction model to obtain the tilting angle, acceleration and predicted comprehensive risk characteristic values of the future 1-2 seconds; a fuzzy control algorithm is adopted to dynamically adjust a safety threshold based on the MOF material state, and five grades of danger levels are divided; corresponding safety actions are triggered according to the danger levels, and safety action parameters are dynamically adjusted; the application realizes cooperative monitoring of equipment postures and reaction states, early warning of risks and graded dynamic protection.

Description

Technical Field

[0001] This invention relates to the field of hydrogen fuel cell technology, and specifically to a sensor-based method and system for preventing and warning of hydrogen fuel cell tipping. Background Technology

[0002] In the field of portable power generation in new energy, MOF water electrolysis hydrogen production technology has become the core technology path for miniaturized and mobile hydrogen fuel cell power supply equipment due to its advantages such as high hydrogen purity, strong reaction controllability and moderate energy density. Its working principle is: MOF material and water undergo a hydrolysis reaction under the action of a catalyst to generate hydrogen. After purification, the hydrogen is supplied to the fuel cell stack to achieve efficient conversion of chemical energy into electrical energy.

[0003] However, there are significant safety hazards in the MOF hydrolysis hydrogen production process: when the reaction system generates hydrogen, a large amount of foam (formed by the mixing of unreleased hydrogen with the reaction solution) is easily generated due to local reaction rate fluctuations and solution disturbances; if the equipment is tilted or subjected to a violent impact, the foam can easily enter the gas transmission pipeline and fuel cell stack with the hydrogen flow. At the same time, undissolved MOF particles in the reaction solution and metal ions generated by metal friction / corrosion inside the equipment will also mix into the hydrogen; solid impurities can clog pipelines and valves, leading to a decrease in hydrogen transmission efficiency and causing abnormal system pressure; metal ions, as catalyst poisons in fuel cell stacks, can be irreversibly adsorbed on the surface of platinum catalysts, resulting in reduced catalyst activity, decreased stack power, and in severe cases, direct stack scrapping.

[0004] In existing technologies, protection is triggered by a single parameter, such as static tilt angle or dynamic acceleration, without considering the differences in MOF material states. This leads to deviations in hazard level assessment; for example, at the same tilt angle, the spill risk of paste-like MOF is much higher than that of dry powder. Protection is only activated after the equipment has tilted or been impacted beyond the limit, making it impossible to predict risks in advance and easily missing the optimal protection opportunity. The parameters of the safety action are fixed and not dynamically adjusted according to real-time system pressure and reaction status, which can easily lead to over-protection or under-protection, affecting equipment stability and efficiency. Therefore, there is an urgent need for a tipping protection method that can integrate multi-dimensional parameters, has early warning capabilities, and adaptive protection actions to overcome the shortcomings of existing technologies. Summary of the Invention

[0005] The purpose of this invention is to solve the above-mentioned problems by designing a sensor-based method and system for preventing and warning about hydrogen fuel cell tipping.

[0006] The first aspect of this invention provides a sensor-based method for preventing and warning of hydrogen fuel cell tipping, the method comprising the following steps:

[0007] The tilt angle and acceleration data of the equipment are collected by the inertial measurement unit, and the MOF material state and pipeline pressure data are collected by the weight sensor, liquid level sensor and pressure sensor.

[0008] The Kalman filter algorithm is used to denoise the collected data, and then normalization is performed to obtain the preprocessed data.

[0009] Based on the dynamic allocation of tilt angle, acceleration, material state, and pipeline pressure according to MOF material state, the real-time comprehensive risk characteristic value is calculated through a weighted fusion algorithm based on preprocessed data;

[0010] The preprocessed data from the past 5 seconds is input into the Transformer-TCN hybrid prediction model to obtain the tilt angle, acceleration, and predicted comprehensive risk characteristic value for the next 1-2 seconds.

[0011] A fuzzy control algorithm is used to dynamically adjust the safety threshold based on the MOF material state, and the five hazard levels are divided by combining the real-time comprehensive risk characteristic value and the predicted comprehensive risk characteristic value.

[0012] The corresponding safety action is triggered based on the level of danger, and the safety action parameters are dynamically adjusted through a PID control algorithm.

[0013] Optionally, in a first implementation of the first aspect of the present invention, the step of using a Kalman filter algorithm to denoise the acquired data and then normalizing it to obtain preprocessed data includes:

[0014] A state equation is constructed based on the motion characteristics of the equipment, with tilt angle and acceleration as state variables, and sensor drift error is introduced as a noise term. Initial values ​​for process noise covariance and measurement noise covariance are set based on historical data.

[0015] The state estimate at the current moment is predicted based on the state equation, and then the Kalman gain is calculated by combining the collected data. The predicted value is corrected by the Kalman gain to obtain the denoised data.

[0016] Determine the upper and lower limits of the safe range of each parameter in the denoised data, and map each parameter value to the [0,1] interval to obtain standardized preprocessed data.

[0017] Optionally, in a second implementation of the first aspect of the present invention, the step of dynamically allocating the weights of tilt angle, acceleration, material state, and pipeline pressure based on MOF material state, and calculating the real-time comprehensive risk characteristic value based on preprocessed data using a weighted fusion algorithm, includes:

[0018] Based on the remaining mass collected by the weight sensor and the amount of catalyst solution collected by the liquid level sensor, combined with the preset state classification rules, the current state type of the MOF material is determined.

[0019] The four parameters of tilt angle, acceleration, MOF material state, and pipeline pressure are assigned corresponding weights. Each parameter in the preprocessed data is multiplied by its corresponding weight, and the products are summed to obtain a real-time comprehensive risk characteristic value that reflects the overall risk level of the current equipment.

[0020] Optionally, in a third implementation of the first aspect of the present invention, when the MOF material is a paste, the static tilt angle weight is 0.4 and the MOF material state weight is 0.25.

[0021] When the MOF material is a dry powder, the dynamic acceleration weight is 0.4 and the MOF material state weight is 0.2.

[0022] Optionally, in the fourth implementation of the first aspect of the present invention, the step of inputting the preprocessed data of the past 5 seconds into the Transformer-TCN hybrid prediction model to obtain the tilt angle, acceleration, and predicted comprehensive risk characteristic value for the next 1-2 seconds includes:

[0023] The preprocessed data from the past 5 seconds is input into the Transformer-TCN hybrid prediction model, which employs a dual-channel parallel processing architecture.

[0024] The TCN channel uses dilated convolutional layers with different dilation rates to extract local features at different time scales in parallel. The Transformer channel uses a multi-head self-attention mechanism to model global dependencies, focusing on the long-term trend of tilt angle changes and the periodic characteristics of acceleration fluctuations.

[0025] A cross-attention mechanism is used to deeply fuse the multi-scale local features extracted by the TCN channel with the global features extracted by the Transformer channel to obtain fused features;

[0026] Extract the subset of features most relevant to dumping risk from the fused features, and generate a predicted comprehensive risk feature value through weighted calculation;

[0027] The calculated predicted comprehensive risk feature value is output together with the predicted tilt angle and acceleration value. The predicted tilt angle and acceleration values ​​are converted into a continuous prediction sequence for the next 1-2 seconds by the decoder layer.

[0028] Optionally, in the fifth implementation of the first aspect of the present invention, the step of employing a fuzzy control algorithm to dynamically adjust the safety threshold based on the MOF material state, and combining real-time comprehensive risk characteristic values ​​and predicted comprehensive risk characteristic values ​​to classify 5 hazard levels, including:

[0029] Based on the safety operation standards for hydrogen fuel cells, the basic threshold values ​​for tilt angle, acceleration, and pipeline pressure parameters are preset.

[0030] A fuzzy control algorithm is used to determine the adjustment coefficient and material state coefficient based on the current MOF material state. The adaptive safety threshold is obtained by multiplying the base threshold by the adjusted coefficient.

[0031] The real-time comprehensive risk characteristic value and the predicted comprehensive risk characteristic value are compared with the adaptive safety threshold. Combined with the preset level classification rules, the equipment risk is divided into 5 levels: safe level, low risk level, medium risk level, high risk level and extremely high risk level.

[0032] Optionally, in a sixth implementation of the first aspect of the present invention, the step of triggering a corresponding safety action based on the hazard level and dynamically adjusting the safety action parameters through a PID control algorithm includes:

[0033] Based on the hazard level, the preset safety action library is invoked to determine the safety actions to be performed. The safety actions corresponding to low hazard level are: reducing the reaction rate; medium hazard level: reducing the reaction rate, briefly closing the hydrogen inlet valve of the fuel cell stack, and opening the pressure relief solenoid valve when the pressure is too high; high hazard level: stopping the injection of catalyst solution, closing the hydrogen inlet valve of the fuel cell stack, opening the pressure relief solenoid valve, and restarting after waiting for the temperature of the Mofu tank to drop; and extremely high hazard level: stopping the injection of catalyst solution, closing the hydrogen inlet valve of the fuel cell stack, opening the pressure relief solenoid valve, disallowing restarting, and reminding the user to replace the filter element.

[0034] Based on the target parameters for safe actions, the deviation between real-time parameters and target parameters is collected. The adjustment amount is calculated through proportional, integral, and derivative components. The actuator is driven to act according to the calculated adjustment amount. At the same time, the effect of the action is monitored in real time by sensors, and the feedback data is transmitted back to the PID controller to dynamically correct the adjustment amount until the equipment status meets the safety requirements.

[0035] A second aspect of the present invention provides a sensor-based hydrogen fuel cell tipping prevention and early warning system, the system comprising:

[0036] The data acquisition module is used to acquire the tilt angle and acceleration data of the equipment through the inertial measurement unit, and to acquire the MOF material state and pipeline pressure data in combination with the weight sensor, liquid level sensor and pressure sensor;

[0037] The preprocessing module is used to denoise the acquired data using the Kalman filter algorithm, and then normalize the data to obtain the preprocessed data.

[0038] The weighted fusion module is used to dynamically allocate weights for tilt angle, acceleration, material state, and pipeline pressure based on MOF material state, and calculates real-time comprehensive risk characteristic values ​​based on preprocessed data using a weighted fusion algorithm.

[0039] The prediction module is used to input the preprocessed data from the past 5 seconds into the Transformer-TCN hybrid prediction model to obtain the tilt angle, acceleration, and predicted comprehensive risk characteristic value for the next 1-2 seconds.

[0040] The classification module is used to dynamically adjust the safety threshold based on the MOF material state using a fuzzy control algorithm, and classify five hazard levels by combining real-time comprehensive risk characteristic values ​​and predicted comprehensive risk characteristic values.

[0041] The dynamic adjustment module is used to trigger corresponding safety actions based on the hazard level and dynamically adjust the safety action parameters through a PID control algorithm.

[0042] A third aspect of the present invention provides a sensor-based hydrogen fuel cell tipping protection and early warning processing device, the sensor-based hydrogen fuel cell tipping protection and early warning processing device comprising a memory and at least one processor, the memory storing instructions; the at least one processor calling the instructions in the memory to cause the sensor-based hydrogen fuel cell tipping protection and early warning processing device to perform the various steps of the sensor-based hydrogen fuel cell tipping protection and early warning processing method as described in any of the preceding claims.

[0043] A fourth aspect of the present invention provides a computer-readable storage medium storing instructions that, when executed by a processor, implement the steps of the sensor-based hydrogen fuel cell tipping prevention and early warning processing method as described in any of the preceding claims.

[0044] The technical solution provided by this invention collects tilt angle and acceleration data of the equipment through an inertial measurement unit, and collects MOF material state and pipeline pressure data through a weight sensor, liquid level sensor, and pressure sensor. A Kalman filter algorithm is used to denoise the collected data, followed by normalization to obtain preprocessed data. Weights are dynamically assigned to tilt angle, acceleration, material state, and pipeline pressure based on the MOF material state. A weighted fusion algorithm is used to calculate the real-time comprehensive risk characteristic value based on the preprocessed data. The preprocessed data from the past 5 seconds is input into the Transformer-TCN hybrid prediction model to obtain the tilt angle, acceleration, and predicted comprehensive risk characteristic value for the next 1-2 seconds. A fuzzy control algorithm is used to dynamically adjust the safety threshold based on the MOF material state, and the real-time and predicted comprehensive risk characteristic values ​​are combined to classify five hazard levels. Corresponding safety actions are triggered according to the hazard level, and the safety action parameters are dynamically adjusted through a PID control algorithm. This invention integrates multiple parameters such as equipment attitude, MOF material state, and pipeline pressure to solve the problem of single-parameter judgment bias, improving the accuracy of hazard level determination. It predicts the next 1-2 seconds... The second-level attitude change triggers protection earlier than traditional methods, effectively preventing foam and impurities from entering the fuel cell stack and extending its lifespan. Based on the dynamic adjustment of safety thresholds according to the MOF material state, combined with PID control to optimize protection action parameters, it avoids frequent start-ups and shutdowns of the equipment due to over-protection, while preventing safety accidents caused by insufficient protection. It achieves coordinated monitoring of equipment attitude and reaction status, early warning of risks, and graded dynamic protection, ensuring the safety and stability of the equipment in complex scenarios. Attached Figure Description

[0045] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention.

[0046] Figure 1 A flowchart illustrating a sensor-based hydrogen fuel cell tipping protection and early warning processing method provided in an embodiment of the present invention;

[0047] Figure 2 A schematic diagram of the structure of a sensor-based hydrogen fuel cell tilting protection and early warning system provided in an embodiment of the present invention;

[0048] Figure 3 This is a schematic diagram of the structure of a sensor-based hydrogen fuel cell tilting protection and early warning processing device provided in an embodiment of the present invention. Detailed Implementation

[0049] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” or “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, apparatus, product, or device that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0050] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 The flowchart of the sensor-based hydrogen fuel cell tipping protection and early warning method provided in this embodiment of the invention includes the following steps:

[0051] Step 101: Collect the tilt angle and acceleration data of the equipment through the inertial measurement unit, and collect the MOF material state and pipeline pressure data in combination with the weight sensor, liquid level sensor and pressure sensor;

[0052] Step 102: Use the Kalman filter algorithm to denoise the collected data, and then normalize it to obtain the preprocessed data;

[0053] In this embodiment, the hardware interface communicates with the sensor via the SPI communication protocol; the data acquisition frequency is set to 1kHz based on the sensor's output data rate (ODR), with a maximum support of 32kHz; the data format is as follows: the acquired accelerometer and gyroscope data will be stored in the raw data format and will be converted and processed as needed later.

[0054] In this embodiment, a state equation is constructed based on the motion characteristics of the equipment, with tilt angle and acceleration as state variables, and sensor drift error is introduced as a noise term. Initial values ​​for process noise covariance and measurement noise covariance are set based on historical data. The state estimate at the current moment is predicted based on the state equation, and the Kalman gain is calculated by combining the collected data. The predicted value is corrected by the Kalman gain to obtain the denoised data. The upper and lower limits of the safety range of each parameter in the denoised data are determined, and each parameter value is mapped to the [0,1] interval to obtain standardized preprocessed data.

[0055] Step 103: Based on the MOF material state, dynamically allocate the weights of tilt angle, acceleration, material state, and pipeline pressure, and calculate the real-time comprehensive risk characteristic value through a weighted fusion algorithm based on the preprocessed data;

[0056] In this embodiment, the current state type of MOF material is determined based on the remaining mass collected by the weight sensor and the amount of catalyst solution collected by the liquid level sensor, combined with the preset state classification rules. The four parameters of tilt angle, acceleration, MOF material state and pipeline pressure are assigned corresponding weights. Each parameter in the preprocessed data is multiplied by its corresponding weight, and the products are summed to obtain a real-time comprehensive risk characteristic value that reflects the overall risk level of the current equipment.

[0057] In this embodiment, when the MOF material is a paste, the static tilt angle weight is 0.4 and the MOF material state weight is 0.25; when the MOF material is a dry powder, the dynamic acceleration weight is 0.4 and the MOF material state weight is 0.2.

[0058] In this embodiment, the static tilt exceeding the standard judgment is: the pitch angle, roll angle or combined tilt angle of the device relative to the horizontal plane is calculated, and it is determined whether it continuously exceeds the preset safety threshold; the dynamic acceleration abnormal judgment is: the acceleration amplitude (composite acceleration or specific axial acceleration) of the device in a short period of time is calculated, and it is determined whether it exceeds the preset safety threshold.

[0059] Step 104: Input the preprocessed data from the past 5 seconds into the Transformer-TCN hybrid prediction model to obtain the tilt angle, acceleration, and predicted comprehensive risk characteristic value for the next 1-2 seconds;

[0060] In this embodiment, preprocessed data from the past 5 seconds is input into the Transformer-TCN hybrid prediction model, which employs a dual-channel parallel processing architecture. The TCN channel uses dilated convolutional layers with different dilation rates to extract local features at different time scales in parallel, while the Transformer channel utilizes a multi-head self-attention mechanism to model global dependencies, focusing on the long-term trend of tilt angle changes and the periodic characteristics of acceleration fluctuations. A cross-attention mechanism is used to deeply fuse the multi-scale local features extracted by the TCN channel with the global features extracted by the Transformer channel to obtain fused features. The subset of features most relevant to the tipping risk is extracted from the fused features, and a weighted calculation is used to generate a predicted comprehensive risk feature value. The calculated predicted comprehensive risk feature value is output along with the predicted tilt angle and acceleration values, where the predicted tilt angle and acceleration values ​​are converted into a continuous prediction sequence for the next 1-2 seconds through a decoder layer.

[0061] Step 105: Using a fuzzy control algorithm, the safety threshold is dynamically adjusted based on the state of MOF materials, and the five hazard levels are divided by combining the real-time comprehensive risk characteristic value and the predicted comprehensive risk characteristic value.

[0062] In this embodiment, when setting the basic thresholds for tilt angle, acceleration, and pipeline pressure parameters according to the safety operation standards of hydrogen fuel cells, it is necessary to first refer to the general specifications for the safe operation of hydrogen fuel cell equipment in the industry, such as hydrogen system design standards and safety guidelines for portable power generation equipment, and conduct special tests in combination with the structural characteristics of the equipment, such as the connection strength between the MOF tank and the stack and the pipeline pressure resistance level: by simulating the stability of the equipment at different tilt angles, such as whether the MOF solution begins to overflow when tilted to a certain angle, the basic threshold for tilt angle is determined; by recording the tolerance limits of equipment components such as valves and sensors through impact tests, the basic threshold for acceleration is determined; based on the burst pressure and sealing performance test results of the hydrogen transmission pipeline, the basic threshold for pipeline pressure is determined, and finally these tested and verified values ​​are solidified as the initial basic threshold parameters of the system;

[0063] When using a fuzzy control algorithm to determine the adaptive safety threshold, the following steps are required: First, define a fuzzy set of MOF material states, such as four fuzzy subsets: dry powder, saturated adsorption, paste slurry, and excessive dilution. Assign a membership function to each subset to determine the degree to which the current state belongs to a particular subset. Next, establish a fuzzy rule base. The rules specify that if the MOF material is in a certain state, the adjustment coefficient and material state coefficient should be a certain value. For example, if the rule is for paste slurry, the adjustment coefficient should be 0.25 and the material state coefficient 0.3. Then, input the current MOF material state data into the fuzzy controller. Fuzzy inference matches the rule in the rule base that best matches the current state to obtain the fuzzy output of the adjustment coefficient and material state coefficient. Finally, defuzzification, such as the centroid method, converts the fuzzy output into a precise value. The final adaptive safety threshold is then calculated by multiplying the base threshold by 1 - the adjustment coefficient × the material state coefficient.

[0064] The real-time comprehensive risk characteristic value and the predicted comprehensive risk characteristic value are compared with the adaptive safety threshold. Combined with the preset level classification rules, the equipment risk is divided into 5 levels: safe level, low risk level, medium risk level, high risk level and extremely high risk level.

[0065] Step 106: Trigger the corresponding safety action according to the hazard level, and dynamically adjust the safety action parameters through the PID control algorithm.

[0066] In this embodiment, based on the hazard level, a preset safety action library is invoked to determine the safety actions to be performed. The safety actions corresponding to the low hazard level are: reducing the reaction rate; the safety actions corresponding to the medium hazard level are: reducing the reaction rate, briefly closing the hydrogen inlet valve of the fuel cell stack, and opening the pressure relief solenoid valve when the pressure is too high; the safety actions corresponding to the high hazard level are: stopping the injection of catalyst solution, closing the hydrogen inlet valve of the fuel cell stack, opening the pressure relief solenoid valve, and restarting after waiting for the temperature of the Mofu tank to drop; the safety actions corresponding to the extremely high hazard level are: stopping the injection of catalyst solution, closing the hydrogen inlet valve of the fuel cell stack, opening the pressure relief solenoid valve, not allowing restarting, and reminding the user to replace the filter element.

[0067] The target parameters for safe actions are determined, such as stabilizing pipeline pressure or reducing the catalyst injection solution flow rate to a target value. Then, real-time parameters such as real-time pipeline pressure and real-time catalyst injection solution flow rate are collected using corresponding sensors, and the deviation between these real-time parameters and the target parameters is calculated. Next, the proportional component of the PID controller directly outputs a proportional adjustment based on the magnitude of the current deviation; the larger the deviation, the stronger the adjustment, quickly reducing the deviation. The integral component accumulates the total deviation over a period of time; when there is a persistent small deviation, it outputs an additional adjustment through the cumulative effect to gradually eliminate the systematic deviation. The derivative component analyzes the rate and trend of deviation change to predict the development of the deviation. The controller first outputs an adjustment amount in advance to suppress the expansion of deviation. Then, the controller integrates the outputs of the proportional, integral, and derivative components to obtain the final adjustment amount. This adjustment amount is then used to drive the corresponding actuator, such as adjusting the opening of the pressure relief valve, changing the speed of the catalyst injection pump, or controlling the on / off state of the hydrogen valve. At the same time, the sensor continuously monitors the changes in equipment parameters after the actuator's action and transmits the real-time feedback data back to the PID controller. The controller recalculates the new deviation and updates the adjustment amount again through the three components. This process is repeated to dynamically correct the actuator's action parameters until the deviation between the real-time equipment parameters and the target parameters is within the allowable range, meaning the equipment status meets safety requirements.

[0068] Please see Figure 2 A schematic diagram of the structure of a sensor-based hydrogen fuel cell tipping prevention and early warning system provided in this embodiment of the invention. The system includes:

[0069] The data acquisition module is used to acquire the tilt angle and acceleration data of the equipment through the inertial measurement unit, and to acquire the MOF material state and pipeline pressure data in combination with the weight sensor, liquid level sensor and pressure sensor;

[0070] The preprocessing module is used to denoise the acquired data using the Kalman filter algorithm, and then normalize the data to obtain the preprocessed data.

[0071] The weighted fusion module is used to dynamically allocate weights for tilt angle, acceleration, material state, and pipeline pressure based on MOF material state, and calculates real-time comprehensive risk characteristic values ​​based on preprocessed data using a weighted fusion algorithm.

[0072] The prediction module is used to input the preprocessed data from the past 5 seconds into the Transformer-TCN hybrid prediction model to obtain the tilt angle, acceleration, and predicted comprehensive risk characteristic value for the next 1-2 seconds.

[0073] The classification module is used to dynamically adjust the safety threshold based on the MOF material state using a fuzzy control algorithm, and classify five hazard levels by combining real-time comprehensive risk characteristic values ​​and predicted comprehensive risk characteristic values.

[0074] The dynamic adjustment module is used to trigger corresponding safety actions based on the hazard level and dynamically adjust the safety action parameters through a PID control algorithm.

[0075] Figure 3 This is a schematic diagram of a sensor-based hydrogen fuel cell tipping prevention and early warning processing device 300 provided in an embodiment of the present invention. The sensor-based hydrogen fuel cell tipping prevention and early warning processing device 300 can vary significantly due to different configurations or performance. It may include one or more central processing units (CPUs) 310 and memories 320, and one or more storage media 330 (e.g., one or more mass storage devices) storing application programs 333 or data 332. The memories 320 and storage media 330 can be short-term or long-term storage. The program stored in the storage media 330 may include one or more modules (not shown in the diagram), each module may include a series of instruction operations on the sensor-based hydrogen fuel cell tipping prevention and early warning processing device 300. Furthermore, the processor 310 may be configured to communicate with the storage media 330 and execute the series of instruction operations in the storage media 330 on the sensor-based hydrogen fuel cell tipping prevention and early warning processing device 300 to implement the method provided in the above embodiment.

[0076] The sensor-based hydrogen fuel cell tipping prevention and early warning system 300 may also include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input / output interfaces 360, and / or one or more operating systems 331, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will understand that... Figure 3The sensor-based hydrogen fuel cell tipping protection and early warning processing device structure shown does not constitute a limitation on the computer device provided by the present invention. It may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0077] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the various steps of the sensor-based hydrogen fuel cell tipping protection and early warning processing method provided in the above embodiments.

[0078] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the above-described equipment or apparatus / unit can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0079] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0080] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A sensor-based method for preventing and warning of tipping over of hydrogen fuel cells, characterized in that, The method includes the following steps: The tilt angle and acceleration data of the equipment are collected by the inertial measurement unit, and the MOF material state and pipeline pressure data are collected by the weight sensor, liquid level sensor and pressure sensor. The Kalman filter algorithm is used to denoise the collected data, and then normalization is performed to obtain the preprocessed data. Based on the dynamic allocation of tilt angle, acceleration, material state, and pipeline pressure according to MOF material state, the real-time comprehensive risk characteristic value is calculated through a weighted fusion algorithm based on preprocessed data; The preprocessed data from the past 5 seconds is input into the Transformer-TCN hybrid prediction model to obtain the tilt angle, acceleration, and predicted comprehensive risk characteristic value for the next 1-2 seconds. A fuzzy control algorithm is used to dynamically adjust the safety threshold based on the MOF material state, and the five hazard levels are divided by combining the real-time comprehensive risk characteristic value and the predicted comprehensive risk characteristic value. The corresponding safety action is triggered based on the level of danger, and the safety action parameters are dynamically adjusted through a PID control algorithm.

2. The sensor-based hydrogen fuel cell tipping prevention and early warning method as described in claim 1, characterized in that, The process involves using a Kalman filter algorithm to denoise the acquired data, followed by normalization to obtain preprocessed data, including: A state equation is constructed based on the motion characteristics of the equipment, with tilt angle and acceleration as state variables, and sensor drift error is introduced as a noise term. Initial values ​​for process noise covariance and measurement noise covariance are set based on historical data. The state estimate at the current moment is predicted based on the state equation, and then the Kalman gain is calculated by combining the collected data. The predicted value is corrected by the Kalman gain to obtain the denoised data. Determine the upper and lower limits of the safe range of each parameter in the denoised data, and map each parameter value to the [0,1] interval to obtain standardized preprocessed data.

3. The sensor-based hydrogen fuel cell tipping prevention and early warning method as described in claim 1, characterized in that, The method of dynamically assigning weights to tilt angle, acceleration, material state, and pipeline pressure based on MOF material state, and calculating real-time comprehensive risk characteristic values ​​using a weighted fusion algorithm based on preprocessed data, includes: Based on the remaining mass collected by the weight sensor and the amount of catalyst solution collected by the liquid level sensor, combined with the preset state classification rules, the current state type of the MOF material is determined. The four parameters of tilt angle, acceleration, MOF material state, and pipeline pressure are assigned corresponding weights. Each parameter in the preprocessed data is multiplied by its corresponding weight, and the products are summed to obtain a real-time comprehensive risk characteristic value that reflects the overall risk level of the current equipment.

4. The sensor-based hydrogen fuel cell tipping prevention and early warning method as described in claim 3, characterized in that, When the MOF material is a paste-like slurry, the static tilt angle weight is 0.4, and the MOF material state weight is 0.

25. When the MOF material is a dry powder, the dynamic acceleration weight is 0.4 and the MOF material state weight is 0.

2.

5. The sensor-based hydrogen fuel cell tipping prevention and early warning method as described in claim 1, characterized in that, The process involves inputting preprocessed data from the past 5 seconds into the Transformer-TCN hybrid prediction model to obtain the tilt angle, acceleration, and predicted comprehensive risk characteristic values ​​for the next 1-2 seconds, including: The preprocessed data from the past 5 seconds is input into the Transformer-TCN hybrid prediction model, which employs a dual-channel parallel processing architecture. The TCN channel uses dilated convolutional layers with different dilation rates to extract local features at different time scales in parallel. The Transformer channel uses a multi-head self-attention mechanism to model global dependencies, focusing on the long-term trend of tilt angle changes and the periodic characteristics of acceleration fluctuations. A cross-attention mechanism is used to deeply fuse the multi-scale local features extracted by the TCN channel with the global features extracted by the Transformer channel to obtain fused features; Extract the subset of features most relevant to dumping risk from the fused features, and generate a predicted comprehensive risk feature value through weighted calculation; The calculated predicted comprehensive risk feature value is output together with the predicted tilt angle and acceleration value. The predicted tilt angle and acceleration values ​​are converted into a continuous prediction sequence for the next 1-2 seconds by the decoder layer.

6. The sensor-based hydrogen fuel cell tipping prevention and early warning method as described in claim 1, characterized in that, The method employs a fuzzy control algorithm to dynamically adjust the safety threshold based on the MOF material state, and combines real-time comprehensive risk characteristic values ​​with predicted comprehensive risk characteristic values ​​to classify five hazard levels, including: Based on the safety operation standards for hydrogen fuel cells, the basic threshold values ​​for tilt angle, acceleration, and pipeline pressure parameters are preset. A fuzzy control algorithm is used to determine the adjustment coefficient and material state coefficient based on the current MOF material state. The adaptive safety threshold is obtained by multiplying the base threshold by the adjusted coefficient. The real-time comprehensive risk characteristic value and the predicted comprehensive risk characteristic value are compared with the adaptive safety threshold. Combined with the preset level classification rules, the equipment risk is divided into 5 levels: safe level, low risk level, medium risk level, high risk level and extremely high risk level.

7. The sensor-based hydrogen fuel cell tipping prevention and early warning method as described in claim 1, characterized in that, The process of triggering corresponding safety actions based on hazard levels and dynamically adjusting safety action parameters using a PID control algorithm includes: Based on the hazard level, the preset safety action library is invoked to determine the safety actions to be performed. The safety actions corresponding to low hazard level are: reducing the reaction rate; medium hazard level: reducing the reaction rate, briefly closing the hydrogen inlet valve of the fuel cell stack, and opening the pressure relief solenoid valve when the pressure is too high; high hazard level: stopping the injection of catalyst solution, closing the hydrogen inlet valve of the fuel cell stack, opening the pressure relief solenoid valve, and restarting after waiting for the temperature of the Mofu tank to drop; and extremely high hazard level: stopping the injection of catalyst solution, closing the hydrogen inlet valve of the fuel cell stack, opening the pressure relief solenoid valve, disallowing restarting, and reminding the user to replace the filter element. Based on the target parameters for safe actions, the deviation between real-time parameters and target parameters is collected. The adjustment amount is calculated through proportional, integral, and derivative components. The actuator is driven to act according to the calculated adjustment amount. At the same time, the effect of the action is monitored in real time by sensors, and the feedback data is transmitted back to the PID controller to dynamically correct the adjustment amount until the equipment status meets the safety requirements.

8. A sensor-based hydrogen fuel cell tipping protection and early warning system, characterized in that, The system includes: The data acquisition module is used to acquire the tilt angle and acceleration data of the equipment through the inertial measurement unit, and to acquire the MOF material state and pipeline pressure data in combination with the weight sensor, liquid level sensor and pressure sensor; The preprocessing module is used to denoise the acquired data using the Kalman filter algorithm, and then normalize the data to obtain the preprocessed data. The weighted fusion module is used to dynamically allocate weights for tilt angle, acceleration, material state, and pipeline pressure based on MOF material state, and calculates real-time comprehensive risk characteristic values ​​based on preprocessed data using a weighted fusion algorithm. The prediction module is used to input the preprocessed data from the past 5 seconds into the Transformer-TCN hybrid prediction model to obtain the tilt angle, acceleration, and predicted comprehensive risk characteristic value for the next 1-2 seconds. The classification module is used to dynamically adjust the safety threshold based on the MOF material state using a fuzzy control algorithm, and classify five hazard levels by combining real-time comprehensive risk characteristic values ​​and predicted comprehensive risk characteristic values. The dynamic adjustment module is used to trigger corresponding safety actions based on the hazard level and dynamically adjust the safety action parameters through a PID control algorithm.

9. A sensor-based hydrogen fuel cell tipping protection and early warning system, characterized in that, The sensor-based hydrogen fuel cell tipping protection early warning processing device includes a memory and at least one processor, wherein the memory stores instructions; the at least one processor invokes the instructions in the memory to cause the sensor-based hydrogen fuel cell tipping protection early warning processing device to perform each step of the sensor-based hydrogen fuel cell tipping protection early warning processing method as described in any one of claims 1-7.

10. A computer-readable storage medium storing instructions thereon, characterized in that, When the instructions are executed by the processor, they implement each step of the sensor-based hydrogen fuel cell tipping protection and early warning processing method as described in any one of claims 1-7.

Images