Hydrogen leakage prediction method and device based on artificial neural network
Through the hydrogen leakage prediction method based on artificial neural network, the problem of poor hydrogen concentration prediction time and accuracy in hydrogen powered cabin system is solved, and fast and efficient hydrogen concentration monitoring and alarm are achieved, enhancing the safety of the cabin system.
Patent Information
- Application Number
- CN202410984830.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-22
- Publication Date
- 2024-11-12
AI Technical Summary
In the prior art, hydrogen concentration prediction calculation takes a long time, poor accuracy and real-time performance, which affects the safety of the hydrogen-powered cabin system.
Using the hydrogen leakage prediction method based on artificial neural network, a simulation model of fuel cell cabin is established, a hydrogen leakage database is established, and an artificial neural network is trained to obtain a hydrogen leakage prediction model, predict the hydrogen concentration at the future moment, and based on the predicted concentration and current concentration Determine the hydrogen leakage and start the alarm system.
It improves the real-time and accuracy of hydrogen concentration prediction, realizes fast and efficient hydrogen concentration monitoring, timely alarm and emergency treatment, and enhances the safety of the hydrogen-powered cabin system.
Smart Images

Figure CN118940224A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the technical field of hydrogen leakage prediction, and in particular to a hydrogen leakage prediction method and device based on artificial neural network. Background Art
[0002] With the popularization of hydrogen energy technology, various types of ships around the world tend to be equipped with hydrogen fuel cell systems to achieve hydrogen fuel cells as the main propulsion power and various power requirements of ships. However, fuel cells use hydrogen as fuel, and hydrogen-powered cabins are equipped with hydrogen storage tanks and complex intake and exhaust pipes and circuit designs. Once hydrogen leaks, and the hydrogen leakage reaches a certain concentration, it is easy to spontaneously combust or explode when ignited by open flames. In recent years, explosion accidents caused by hydrogen leakage have been common in industry and transportation. Therefore, it is very necessary to reasonably predict the hydrogen concentration of hydrogen-powered cabin systems, determine whether hydrogen leakage has occurred, and take timely prognostic measures when hydrogen leakage occurs, in order to ensure the safe and stable operation of hydrogen-powered cabin systems.
[0003] In the prior art, hydrogen fuel cell groups are generally placed in the fuel cell cabin, which may cause safety accidents such as hydrogen leakage during operation. Since hydrogen is flammable, it is easy to cause safety accidents such as fire and explosion when it reaches a certain concentration. In addition, the working environment of the ship is on the water, and the working environment is more severe and changeable. When an accident occurs, it is relatively difficult to implement corresponding rescue measures. The existing hydrogen concentration prediction calculation is often time-consuming, and the prediction accuracy and real-time performance are poor, which affects the safety of the hydrogen-powered cabin system. Summary of the invention
[0004] The purpose of the embodiments of the present disclosure is to provide a hydrogen leakage prediction method and device based on an artificial neural network, so as to at least solve the technical problems existing in the prior art that the hydrogen concentration prediction calculation takes a long time, the prediction accuracy and real-time performance are poor, and the safety of the hydrogen-powered cabin system is affected.
[0005] In order to solve the above technical problems, the embodiments of the present disclosure adopt the following technical solutions:
[0006] In a first aspect, an embodiment of the present disclosure provides a hydrogen leakage prediction method based on an artificial neural network, comprising:
[0007] Establish a simulation model of the fuel cell cabin;
[0008] Establishing a hydrogen leakage database according to the simulation model, wherein the data in the hydrogen leakage database includes the hydrogen concentration of each hydrogen detection point under different hydrogen leakage conditions;
[0009] Training an artificial neural network according to the hydrogen leakage database to obtain a hydrogen leakage prediction model;
[0010] Predicting the hydrogen concentration at a future time according to the hydrogen leakage prediction model to obtain a predicted hydrogen concentration;
[0011] The hydrogen leakage situation is predicted based on the predicted hydrogen concentration and the current hydrogen concentration of the fuel cell cabin.
[0012] In some embodiments, establishing a hydrogen leakage database according to the simulation model includes:
[0013] Determining a hydrogen leakage condition according to at least one factor of an opening and closing state of a hatch, a ventilation intensity of a vent, and a hydrogen leakage duration in the simulation model;
[0014] Inputting the hydrogen leakage condition into the simulation model to obtain a simulation result;
[0015] A hydrogen leakage database is established according to the simulation results.
[0016] In some embodiments, an artificial neural network is trained according to the hydrogen leakage database to obtain a hydrogen leakage prediction model, including:
[0017] The hydrogen concentration, average hydrogen concentration, and hydrogen concentration change rate extracted from the hydrogen leakage database are used as inputs of the NARX neural network, the predicted value of the hydrogen concentration at time t+1 is used as the output of the NARX neural network, and the NARX neural network is trained to obtain a time series prediction model of hydrogen concentration;
[0018] Predicting the hydrogen concentration at a future time according to the hydrogen leakage prediction model includes:
[0019] The historical data of the hydrogen concentration, average hydrogen concentration and hydrogen concentration change rate are input into the time series prediction model for single-step recursive prediction to predict the hydrogen concentration at time t+1.
[0020] In some embodiments, when the hydrogen concentration time series prediction model is trained, the input delay number is 10, the output delay number is 5, the number of hidden layers is 1, the number of neurons is 10, and each logic gate inside the neuron uses a tanh activation function.
[0021] In some embodiments, the method further comprises:
[0022] According to the error between the predicted value and the measured value of the hydrogen concentration, a genetic algorithm is used to correct the hydrogen leakage prediction model.
[0023] In some embodiments, predicting hydrogen leakage according to the predicted hydrogen concentration and the current hydrogen concentration of the fuel cell cabin includes:
[0024] Determining a hydrogen leakage level according to the predicted hydrogen concentration, the current hydrogen concentration in the fuel cell cabin, and the duration of the hydrogen leakage;
[0025] The corresponding alarm system is activated according to the hydrogen leakage level to sound an alarm, and the alarm system includes an audible and visual alarm system and an emergency ventilation system.
[0026] In some embodiments, determining the hydrogen leakage level according to the predicted hydrogen concentration, the current hydrogen concentration of the fuel cell cabin, and the duration of the hydrogen leakage includes:
[0027] If both the current hydrogen concentration and the predicted hydrogen concentration do not exceed 0.5%, it is determined that there is no hydrogen leakage;
[0028] If the current hydrogen concentration or the predicted hydrogen concentration exceeds 0.5%, and both the current hydrogen concentration and the predicted hydrogen concentration do not exceed 4%, a primary hydrogen leak is determined;
[0029] If the current hydrogen concentration or the predicted hydrogen concentration exceeds 4%, and the leakage duration of the hydrogen exceeds 20 seconds, a secondary hydrogen leakage is determined.
[0030] In some embodiments, starting a corresponding alarm system to sound an alarm according to the hydrogen leakage level includes:
[0031] If the hydrogen is not leaking, the green light of the sound and light alarm system lights up to provide a safety reminder;
[0032] If the hydrogen leaks at the first level, the yellow light of the sound and light alarm system lights up to give an alarm, and the forced ventilation of the emergency ventilation system is started;
[0033] If the hydrogen leaks at the second level, the yellow light of the sound and light alarm system will light up, the buzzer will start to alarm, and the forced ventilation of the emergency ventilation system will be started.
[0034] In some embodiments, after determining the hydrogen leak, the method further comprises:
[0035] The concentration of the hydrogen is predicted within a preset time range.
[0036] In a second aspect, the present disclosure provides a hydrogen leakage prediction device based on an artificial neural network, comprising:
[0037] A first building module is configured to build a simulation model of a fuel cell cabin;
[0038] A second establishing module is configured to establish a hydrogen leakage database according to the simulation model, wherein the data in the hydrogen leakage database includes the hydrogen concentration of each hydrogen detection point under different hydrogen leakage conditions;
[0039] A training module, configured to train an artificial neural network according to the hydrogen leakage database to obtain a hydrogen leakage prediction model;
[0040] A hydrogen concentration prediction module, configured to predict the hydrogen concentration at a future time according to the hydrogen leakage prediction model to obtain a predicted hydrogen concentration;
[0041] The hydrogen leakage prediction module is configured to predict the hydrogen leakage situation according to the predicted hydrogen concentration and the current hydrogen concentration of the fuel cell cabin.
[0042] In a third aspect, an embodiment of the present disclosure provides an electronic device, comprising at least a memory and a processor, wherein the memory stores a computer program, and the processor implements the above-mentioned hydrogen leakage prediction method based on artificial neural network when executing the computer program on the memory.
[0043] In a fourth aspect, an embodiment of the present disclosure provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the above-mentioned hydrogen leakage prediction method based on artificial neural network is implemented.
[0044] The embodiments of the present disclosure provide a hydrogen leakage prediction method, device, electronic device and storage medium based on artificial neural network, which establish a simulation model of a fuel cell cabin; establish a hydrogen leakage database according to the simulation model; train the artificial neural network according to the hydrogen leakage database to obtain a hydrogen leakage prediction model; predict the hydrogen concentration at a future time according to the hydrogen leakage prediction model to obtain a predicted hydrogen concentration; predict the hydrogen leakage situation according to the predicted hydrogen concentration and the current hydrogen concentration of the fuel cell cabin, and can realize fast and efficient prediction of hydrogen concentration at a future time based on the hydrogen concentration database established by the simulation model of the fuel cell cabin and using artificial neural network, improve the real-time and accuracy of hydrogen concentration prediction, provide actual value and predicted value of hydrogen concentration for analysis and judgment, accurately monitor hydrogen leakage situation, and improve the accuracy of hydrogen leakage prediction; at the same time, the artificial neural network model is used to increase the robustness of the hydrogen concentration prediction model and improve the accuracy of prediction results. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] In order to more clearly illustrate the embodiments of the present disclosure or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0046] Figure 1is a flow chart of a hydrogen leakage prediction method based on an artificial neural network according to an embodiment of the present disclosure;
[0047] Figure 2 is another flow chart of the hydrogen leakage prediction method based on artificial neural network according to an embodiment of the present disclosure;
[0048] Figure 3 A schematic diagram of a simulation model of a fuel cell cabin established for an embodiment of the present disclosure;
[0049] Figure 4 for Figure 3 A top view of the simulation model of the fuel cell cabin established in ;
[0050] Figure 5 A flowchart of establishing a simulation model of a fuel cell cabin and establishing a hydrogen leakage database according to an embodiment of the present disclosure;
[0051] Figure 6 is a structural diagram of the NARX neural network of an embodiment of the present disclosure;
[0052] Figure 7 A flow chart of hydrogen concentration prediction according to an embodiment of the present disclosure;
[0053] Figure 8 This is a schematic diagram of an embodiment of the present disclosure for alarming according to hydrogen leakage;
[0054] Fig. 9 It is a schematic diagram of the structure of a hydrogen leakage prediction device based on an artificial neural network according to an embodiment of the present disclosure. DETAILED DESCRIPTION
[0055] Various aspects and features of the present disclosure are described herein with reference to the drawings.
[0056] It should be understood that various modifications may be made to the embodiments of the present application. Therefore, the above description should not be considered as limiting, but only as an example of an embodiment. Other modifications within the scope and spirit of the present disclosure will occur to those skilled in the art.
[0057] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate embodiments of the present disclosure and, together with the general description of the present disclosure given above and the detailed description of the embodiments given below, serve to explain the principles of the present disclosure.
[0058] These and other characteristics of the present disclosure will become apparent from the following description of a preferred form of embodiment given as a non-limiting example, with reference to the attached drawings.
[0059] It should also be understood that, although the present disclosure has been described with reference to some specific examples, those skilled in the art will be able to realize many other equivalent forms of the present disclosure that have the features described in the claims and are therefore within the scope of protection defined thereby.
[0060] The above and other aspects, features and advantages of the present disclosure will become more apparent from the following detailed description when taken in conjunction with the accompanying drawings.
[0061] Specific embodiments of the present disclosure are described hereinafter with reference to the accompanying drawings; however, it should be understood that the embodiments claimed are merely examples of the present disclosure, which may be implemented in a variety of ways. Well-known and / or repeated functions and structures are not described in detail to avoid obscuring the present disclosure with unnecessary or redundant details. Therefore, the specific structural and functional details claimed herein are not intended to be limiting, but merely serve as a basis and representative basis for the claims to teach those skilled in the art to use the present disclosure in a variety of ways with substantially any suitable detailed structure.
[0062] This specification may use the phrases "in one embodiment," "in another embodiment," "in a further embodiment," or "in other embodiments," all of which may refer to one or more of the same or different embodiments according to the present disclosure.
[0063] Embodiment 1
[0064] Figure 1 and Figure 2 A flowchart of a hydrogen leakage prediction method based on an artificial neural network according to an embodiment of the present disclosure is shown. Figure 1 and Figure 2 As shown, an embodiment of the present disclosure provides a hydrogen leakage prediction method based on an artificial neural network, comprising:
[0065] S101: Establish a simulation model of a fuel cell cabin.
[0066] A fuel cell cabin (also known as a hydrogen-powered cabin) usually consists of three parts: a battery compartment, a control cabin, and a passenger cabin. According to the research objectives, a simulation model of the fuel cell cabin is established, in which the length, width, and height of the fuel cell cabin are 7m, 5m, and 2.74m, respectively. The cabin has three exhaust ports, including two rear exhaust ports and one upper exhaust port; the cabin has two doors, a front door and a rear door, in which the rear door leads to the stern and the front door leads to the control room. The hydrogen supply pipeline is a 3 / 4-inch 316 steel pipe with an inner diameter of 20mm. The air pressure in the pipeline is about 1MPa. Four fuel cell stacks are arranged in the fuel cell cabin. In the specific implementation, the following is established in the Simcape modeling and simulation platform: Figure 3 The simplified model of the fuel cell cabin is shown in Figure 1. The hydrogen concentration monitoring point is set at the following locations: Figure 4As shown, a hydrogen concentration monitoring point is set at each of the four corners in the fuel cell compartment.
[0067] like Figure 5 As shown, in this embodiment, the hydrogen leakage mode is mainly the continuous leakage of hydrogen due to the rupture of the hydrogen supply pipeline. The hydrogen uses an ideal gas model, the outlet flow state is supersonic flow, and the hydrogen mass flow rate is calculated. During the leakage and diffusion process of hydrogen, the gas flows continuously, satisfies the conservation of mass and energy, and each component satisfies the component transport equation.
[0068] Furthermore, the boundary conditions of the simulation model are determined. The boundary conditions include the cabin environment pressure and temperature. The specific boundary conditions can be: the environment pressure is set to standard atmospheric pressure (101.325 kPa), and the temperature is set to 300 K. K is degrees Kelvin, and degrees Kelvin = degrees Celsius + 273.15 ° C. Considering the influence of gravity, the gravity acceleration is set to 9.8 m / s 2 , vertically downward. Assume that the hydrogen supply pipeline of the first fuel cell stack on the right side is completely broken, the pressure inside the pipe is 1MPa, and the fracture surface is a circular hole with a diameter of 20mm. Fluent simulation software is used to perform fluid simulation and establish a fluid model. Afterwards, a simulation model of the fuel cell cabin is established based on the above-established gas model, the determined boundary conditions, and the established fluid model.
[0069] S102: Establishing a hydrogen leakage database according to the simulation model, wherein the data in the hydrogen leakage database includes the hydrogen concentration of each hydrogen detection point under different hydrogen leakage conditions.
[0070] In this step, the above-mentioned simulation model can be used to set a variety of hydrogen leakage conditions, and then the hydrogen concentration of each hydrogen detection point is simulated and monitored according to the above-mentioned hydrogen leakage conditions to obtain the hydrogen concentration of each hydrogen detection point at a certain moment or a certain time period and the time series data of the hydrogen concentration (data of the change of hydrogen concentration over time), thereby establishing a hydrogen leakage database under different hydrogen leakage conditions.
[0071] S103: Training an artificial neural network according to the hydrogen leakage database to obtain a hydrogen leakage prediction model.
[0072] After the hydrogen leakage database is established, a certain amount of data is extracted from it as training data, which is input into a preset artificial neural network, and the artificial neural network is trained to obtain a hydrogen leakage prediction model.
[0073] S104: Predicting the hydrogen concentration at a future time according to the hydrogen leakage prediction model to obtain a predicted hydrogen concentration.
[0074] After the hydrogen leakage prediction model is trained, the historical data of hydrogen concentration can be input into the hydrogen leakage prediction model to predict the hydrogen concentration at a certain moment in the future to obtain the predicted hydrogen concentration.
[0075] S105: Predicting hydrogen leakage according to the predicted hydrogen concentration and the current hydrogen concentration of the fuel cell cabin.
[0076] In this step, the hydrogen leakage situation can be predicted based on the predicted hydrogen concentration and the current hydrogen concentration actually detected in the fuel cell cabin to determine whether there is a risk of hydrogen leakage.
[0077] The hydrogen leakage prediction method based on artificial neural network provided by the embodiment of the present disclosure establishes a simulation model of a fuel cell cabin; establishes a hydrogen leakage database according to the simulation model; trains an artificial neural network according to the hydrogen leakage database to obtain a hydrogen leakage prediction model; predicts the hydrogen concentration at a future time according to the hydrogen leakage prediction model to obtain a predicted hydrogen concentration; predicts the hydrogen leakage situation according to the predicted hydrogen concentration and the current hydrogen concentration of the fuel cell cabin. It can realize fast and efficient prediction of the hydrogen concentration at a future time based on the hydrogen concentration database established by the simulation model of the fuel cell cabin and by using artificial neural network, improve the real-time and accuracy of the hydrogen concentration prediction, provide the actual value and predicted value of the hydrogen concentration for analysis and judgment, accurately monitor the hydrogen leakage situation, and improve the accuracy of hydrogen leakage prediction; at the same time, the artificial neural network model is used to increase the robustness of the hydrogen concentration prediction model and improve the accuracy of the prediction result.
[0078] In some embodiments, in step S102, establishing a hydrogen leakage database according to the simulation model includes:
[0079] S1021: determining a hydrogen leakage condition according to at least one factor of an opening and closing state of a hatch, a ventilation intensity of a vent, and a hydrogen leakage duration in the simulation model;
[0080] S1022: inputting the hydrogen leakage condition into the simulation model to obtain a simulation result;
[0081] S1023: Establishing a hydrogen leakage database according to the simulation results.
[0082] like Figure 5 As shown, according to the established fuel cell cabin model, various leakage condition combinations are set according to the opening and closing state of the cabin door, the ventilation intensity of the vents, and the duration of hydrogen leakage. In the Fluent simulation software, it is set to pressurize the pressure outlet and the gauge pressure is 0 MPa.
[0083] Since the front door leads to the cabin control room, in order to ensure that there is not too much hydrogen leakage into the control room and cabin, the front door is half open when opened, and its width and height are 0.3 meters and 2 meters respectively. The exhaust conditions of the vents can be set to natural ventilation and forced ventilation. The exhaust efficiency of forced ventilation is the mass flow rate of the gas, which is set as the mass flow outlet in the Fluent simulation software. According to the set leakage conditions, the boundary conditions are set. During the transient numerical calculation process, the time step is fixed to 0.002s. A hydrogen leakage database is established based on the simulation results to obtain the change of hydrogen concentration over time at each hydrogen detection point under various conditions.
[0084] Through the above steps S1021 to S1023, various hydrogen leakage conditions that affect hydrogen leakage are comprehensively considered, the hydrogen leakage conditions are input into the simulation model for simulation, and a comprehensive and accurate hydrogen leakage database is established.
[0085] In some embodiments, in step S103, an artificial neural network is trained according to the hydrogen leakage database to obtain a hydrogen leakage prediction model, including:
[0086] S1031: extracting hydrogen concentration, average hydrogen concentration, and hydrogen concentration change rate from the hydrogen leakage database as inputs of the NARX neural network, taking the predicted value of hydrogen concentration at time t+1 as output of the NARX neural network, training the NARX neural network, and obtaining a time series prediction model for hydrogen concentration;
[0087] In step S104, the hydrogen concentration at a future time is predicted according to the hydrogen leakage prediction model to obtain a predicted hydrogen concentration, including:
[0088] S1041: Inputting the historical data of the hydrogen concentration, average hydrogen concentration, and hydrogen concentration change rate into the time series prediction model for single-step recursive prediction to obtain a predicted value of the hydrogen concentration at time t+1.
[0089] In steps S1031 and S1041, the prediction of hydrogen concentration at future moments is realized based on the NARX neural network. The NARX neural network (Nonlinear Auto-Regressive with eXogenous Inputs) is a neural network based on an autoregressive model, which can predict and control time series data by learning the time dependency of input data and the effects of exogenous factors.
[0090] like Figure 6 and Figure 7 As shown, first, the NARX neural network is structurally initialized with the hydrogen concentration c at time t. t , average hydrogen concentration c average, hydrogen concentration change rate a c The three features are used as the input of the NARX neural network, and the predicted value of hydrogen concentration ct+1 at time t+1 is used as the output of the NARX neural network.
[0091] like Figure 6 As shown, the NARX neural network in this embodiment includes an input layer, a hidden layer and an output layer. The input layer accepts input data, the hidden layer generates a new feature representation by weighting the input and processing the activation function, and the output layer weights the output of the hidden layer and processes the activation function to obtain the final output result. w1, w2, and w3 are the weights of the hidden layer, corresponding to the above three inputs respectively, b1 is the bias of the hidden layer; w4 is the weight of the output layer, and b2 is the bias of the output layer.
[0092] Optionally, when training the time series prediction model of hydrogen concentration, the input delay number is 10, the output delay number is 5, the number of hidden layers is 1, the number of neurons is 10, and each logic gate inside the neuron uses a tanh activation function.
[0093] Then, based on the hydrogen leakage database established by the simulation model, the historical concentration data of hydrogen leakage is extracted, and the NARX training samples are constructed, that is, each training sample uses the average hydrogen concentration and the actual value of the hydrogen concentration change rate corresponding to the time t-10 to the time t, and the actual value of the hydrogen concentration corresponding to the time t-4 to the time t as the input of the NARX neural network, and the actual value of the hydrogen concentration at the time t+1 is used as the output of the NARX neural network. After the construction of the training samples is completed, the NARX neural network is trained to obtain a time series prediction model of hydrogen concentration, which is a hydrogen leakage prediction model. After the training of the NARX neural network is completed, the historical data corresponding to the hydrogen concentration, the average hydrogen concentration, and the hydrogen concentration change rate are input into the time series prediction model for a single-step forward recursive prediction to obtain the predicted value of the hydrogen concentration corresponding to the time t+1, thereby obtaining the predicted hydrogen concentration at the future time.
[0094] In a specific implementation, training parameters such as the input delay number, the output delay number, and the activation function can be determined according to actual needs, and are not specifically limited in the present disclosure. For example, the NARX neural network can also use activation functions such as Sigmoid and Softsign.
[0095] It can be understood that in this embodiment, in the time series prediction model of hydrogen concentration obtained by training, the historical data corresponding to the hydrogen concentration, average hydrogen concentration, and hydrogen concentration change rate as the model input are the hydrogen concentration, average hydrogen concentration, and hydrogen concentration change rate at time t. In specific implementations, the historical data corresponding to the hydrogen concentration, average hydrogen concentration, and hydrogen concentration change rate as the model input can also be the hydrogen concentration, average hydrogen concentration, and hydrogen concentration change rate at other times or other time periods before time t. The specific input features used for hydrogen concentration prediction are determined based on the constructed NARX neural network and the training results of the model, as long as they meet the requirements of predicting future time data based on historical data.
[0096] In this embodiment, a hydrogen leakage prediction model is obtained by training a NARX neural network, and a delay and feedback mechanism is added so that the hydrogen concentration alarm limit is reached in advance, thereby shortening the hydrogen leakage monitoring alarm time and providing an early warning of the danger of hydrogen leakage.
[0097] In other embodiments, the artificial neural network may also include auto-regressive integrated moving averages (ARIMA), recurrent neural networks (RNN), attention mechanisms or Transformer neural networks, etc. In specific implementations, the target parameters, characteristic parameters of the database, sample capacity, model complexity, computing speed, etc. can be determined according to the hydrogen concentration prediction scenario and purpose, and then the corresponding artificial neural network is selected for training to obtain a hydrogen leakage prediction model. The NARX neural network model selected in this embodiment has a relatively simple structure, a relatively mature algorithm, is widely used, and has good prediction results. It can also warn of the danger of hydrogen leakage and the safety of the fuel cell cabin in advance.
[0098] In some embodiments, the method further comprises:
[0099] S201: According to the error between the predicted value and the measured value of the hydrogen concentration, the hydrogen leakage prediction model is corrected using a genetic algorithm.
[0100] Genetic Algorithms (GA) seek the optimal solution to a problem by iteratively searching multiple individuals. Genetic algorithms can be used to optimize the parameters of machine learning models and improve the performance of the models. For example, the training process of models such as neural networks and support vector machines.
[0101] like Figure 7As shown in the figure, after the hydrogen leakage prediction model is constructed, the error between the predicted hydrogen concentration (predicted value) predicted by the above hydrogen leakage prediction model and the hydrogen concentration (measured value) actually monitored in the fuel cell cabin can be used to optimize the hyperparameters of the neural network by using a genetic algorithm, and the model can be continuously corrected so that the error between the predicted value and the measured value of the hydrogen concentration is within the preset error range, which can ensure the convergence speed of the network weight coefficient and prevent overfitting, thereby improving the accuracy of hydrogen concentration prediction.
[0102] In specific implementation, different optimization algorithms can be selected to optimize the artificial neural network for the purpose of improving model performance according to different prediction effects of the model. For example, when the model prediction result is underfitting, the method of increasing the number of training times or changing the complex model is adopted; when the model prediction result is overfitting, Dropout regularization, L1 regularization, L2 regularization and other methods can be used for improvement; if the model parameters are adjusted to minimize the loss function, the whale optimization algorithm, the above-mentioned genetic algorithm, ant colony algorithm, etc. can be used. The genetic algorithm has a strong global optimization capability and is suitable for situations with complex mathematical representations, which can effectively improve the accuracy of hydrogen concentration prediction.
[0103] In some embodiments, in step S105, predicting hydrogen leakage according to the predicted hydrogen concentration and the current hydrogen concentration of the fuel cell cabin includes:
[0104] S1051: Determine the hydrogen leakage level according to the predicted hydrogen concentration, the current hydrogen concentration of the fuel cell cabin, and the duration of the hydrogen leakage;
[0105] S1052: Activate a corresponding alarm system according to the hydrogen leakage level to generate an alarm, wherein the alarm system includes an audible and visual alarm system and an emergency ventilation system.
[0106] Among them, the current hydrogen concentration of the fuel cell cabin is the hydrogen concentration actually monitored in the fuel cell cabin, that is, the measured value of the hydrogen concentration; the predicted hydrogen concentration is the predicted value of the hydrogen concentration.
[0107] like Figure 2 and Figure 8 As shown, hydrogen sensors and timers can be set at various locations in the cabin to monitor the current concentration of hydrogen in the cabin in real time, and the hydrogen concentration at a future time (e.g., the hydrogen concentration at time t+1) can be predicted using the above hydrogen leakage model based on the current hydrogen concentration monitored in real time (e.g., the hydrogen concentration at time t+1) to obtain the predicted hydrogen concentration. Then, the predicted hydrogen concentration, the current hydrogen concentration of the fuel cell cabin, and the duration of the hydrogen leakage are analyzed by the central processing system (e.g., CPU), the hydrogen concentration leakage situation under different dangerous situations is judged, and then corresponding instructions are issued to the alarm system to perform different alarm prompts.
[0108] The existing technology is relatively simple in monitoring or risk prevention of hydrogen leakage, and often lacks reasonable and reliable prognostic measures. In this embodiment, the current concentration of hydrogen, the predicted concentration at a future time, and the duration of hydrogen leakage are comprehensively considered, and the concentration limit of hydrogen combustion and explosion is combined to set different hydrogen leakage concentration levels that trigger the alarm mechanism, and alarm through different alarm systems, that is, a hydrogen leakage concentration graded alarm mechanism is set, taking into account the randomness and uncertainty of actual hydrogen leakage, and taking reasonable alarm methods and prognostic measures to ensure that corresponding emergency plans are provided in different hydrogen leakage scenarios, so as to realize hydrogen safety monitoring, alarm, emergency treatment, etc. in the cabin, avoid the occurrence of hydrogen safety hazards (avoid the occurrence of potential dangers), and ensure the safety of the cabin.
[0109] Furthermore, in step S1051, the hydrogen leakage level is determined according to the predicted hydrogen concentration, the current hydrogen concentration of the fuel cell cabin, and the duration of the hydrogen leakage, including:
[0110] S301: If the current hydrogen concentration and the predicted hydrogen concentration are both less than 0.5%, it is determined that there is no hydrogen leakage;
[0111] S302: If the current hydrogen concentration or the predicted hydrogen concentration exceeds 0.5%, and both the current hydrogen concentration and the predicted hydrogen concentration do not exceed 4%, a primary hydrogen leak is determined;
[0112] S303: If the current hydrogen concentration or the predicted hydrogen concentration exceeds 4%, and the hydrogen leakage duration exceeds 20 seconds, a secondary hydrogen leakage is determined.
[0113] In this embodiment, different hydrogen leakage concentration levels (safety risk levels) are set by combining the current hydrogen concentration monitored by sensors at various locations with the predicted hydrogen concentration predicted by the hydrogen leakage prediction model, so as to quickly and efficiently achieve the prediction of safety risk level and leakage range. For example, the above-mentioned non-leakage of hydrogen indicates that the cabin is safe, the first-level leakage of hydrogen indicates that the hydrogen leakage is small and the cabin is at a relatively safe level; the first-level leakage of hydrogen indicates that the hydrogen leakage is large and the cabin is at a relatively dangerous level.
[0114] Furthermore, in step S1052, a corresponding alarm system is activated to sound an alarm according to the hydrogen leakage level, including:
[0115] S401: If the hydrogen is not leaking, the green light of the sound and light alarm system lights up to provide a safety reminder;
[0116] S402: If the hydrogen leaks at the first level, the yellow light of the sound and light alarm system lights up to give an alarm, and the forced ventilation of the emergency ventilation system is started;
[0117] S403: If the hydrogen leaks at the second level, the yellow light of the sound and light alarm system lights up, the buzzer starts to alarm, and the forced ventilation of the emergency ventilation system is started.
[0118] In this embodiment, an alarm system corresponding to the design of the hydrogen leakage level is used to give an alarm, which can promptly and effectively warn of the hidden dangers of hydrogen leakage. For example, when the hydrogen leakage is large and is in a secondary hydrogen leakage state, the sound and light alarm system not only gives a warning by lighting a red light, but also starts a buzzer alarm to warn people who cannot observe the alarm light. At the same time, forced ventilation emergency measures are started to prevent hydrogen accumulation and explosion through ventilation before the leaked hydrogen is effectively dealt with.
[0119] In this embodiment, different hydrogen leakage levels are clearly set according to the predicted hydrogen concentration, the current hydrogen concentration of the fuel cell cabin, and the duration of hydrogen leakage, and reasonable and effective prognostic measures are set in combination with hydrogen leakage theory and actual prevention and control experience. Corresponding measures are promptly implemented according to the danger level of the system response to avoid explosion or combustion caused by hydrogen leakage, eliminate safety hazards, and effectively ensure the safety of hydrogen-powered cabins.
[0120] In some embodiments, after determining the hydrogen leak, the method further comprises:
[0121] S501: Predicting the concentration of the hydrogen within a preset time range.
[0122] like Figure 8 As shown, after determining the hydrogen leakage, the above hydrogen leakage model can also be used to predict the hydrogen concentration within the future preset time range, continuously predict the hydrogen concentration, predict the leakage range of hydrogen, and further warn of the danger of hydrogen leakage in advance. For example, when it is determined that the current hydrogen leakage state is in the first level of hydrogen leakage, if it is predicted that the hydrogen may be in the second level of hydrogen leakage in the future preset time range, the second level of hydrogen leakage is warned in advance, and corresponding emergency measures are taken, for example, all vents are opened to prevent the expansion of hydrogen leakage; and when it is determined that the current hydrogen leakage state is in the second level of hydrogen leakage, after taking corresponding emergency measures, the hydrogen concentration within the future preset time range can be continuously predicted to determine whether the treatment of hydrogen leakage is completed. For example, when it is predicted that the hydrogen leakage state is in the second level of hydrogen leakage, it can be determined that the treatment measures are reasonable, and the hydrogen concentration is continuously predicted, and the hydrogen leakage is treated accordingly until the hydrogen does not leak and the cabin is in a safe state. Through the continuous prediction of the hydrogen concentration in the above step S501, the hydrogen concentration of the fuel cell cabin can be monitored in real time and accurately, and the safety of the fuel cell cabin can be effectively guaranteed.
[0123] Embodiment 2
[0124] Fig. 9 The structure diagram of the hydrogen leakage prediction device based on artificial neural network in the embodiment of the present disclosure is shown as follows: Fig. 9 As shown, the embodiment of the present disclosure provides a hydrogen leakage prediction device based on an artificial neural network, comprising:
[0125] A first building module 10 is configured to build a simulation model of a fuel cell cabin;
[0126] A second establishing module 20 is configured to establish a hydrogen leakage database according to the simulation model, wherein the data in the hydrogen leakage database includes the hydrogen concentration of each hydrogen detection point under different hydrogen leakage conditions;
[0127] A training module 30 is configured to train an artificial neural network according to the hydrogen leakage database to obtain a hydrogen leakage prediction model;
[0128] A hydrogen concentration prediction module 40 is configured to predict the hydrogen concentration at a future time according to the hydrogen leakage prediction model to obtain a predicted hydrogen concentration;
[0129] The hydrogen leakage prediction module 50 is configured to predict the hydrogen leakage situation according to the predicted hydrogen concentration and the current hydrogen concentration of the fuel cell cabin.
[0130] In some embodiments, the second establishing module 20 is further configured to:
[0131] Determining a hydrogen leakage condition according to at least one factor of an opening and closing state of a hatch, a ventilation intensity of a vent, and a hydrogen leakage duration in the simulation model;
[0132] Inputting the hydrogen leakage condition into the simulation model to obtain a simulation result;
[0133] A hydrogen leakage database is established according to the simulation results.
[0134] In some embodiments, the training module 30 is further configured to:
[0135] Extracting hydrogen concentration, average hydrogen concentration, and hydrogen concentration change rate from the hydrogen leakage database as inputs of a NARX neural network, taking the predicted value of hydrogen concentration at time t+1 as output of the NARX neural network, and training the NARX neural network to obtain a time series prediction model for hydrogen concentration;
[0136] The hydrogen concentration prediction module 40 is further configured as follows:
[0137] The historical data of the hydrogen concentration, average hydrogen concentration and hydrogen concentration change rate are input into the time series prediction model for single-step recursive prediction to predict the hydrogen concentration at time t+1.
[0138] In some embodiments, when the hydrogen concentration time series prediction model is trained, the input delay number is 10, the output delay number is 5, the number of hidden layers is 1, the number of neurons is 10, and each logic gate inside the neuron uses a tanh activation function.
[0139] In some embodiments, a correction module is further included, configured to:
[0140] According to the error between the predicted value and the measured value of the hydrogen concentration, a genetic algorithm is used to correct the hydrogen leakage prediction model.
[0141] In some embodiments, the hydrogen leakage prediction module 50 is further configured to:
[0142] Determining a hydrogen leakage level according to the predicted hydrogen concentration, the current hydrogen concentration in the fuel cell cabin, and the duration of the hydrogen leakage;
[0143] The corresponding alarm system is activated according to the hydrogen leakage level to sound an alarm, and the alarm system includes an audible and visual alarm system and an emergency ventilation system.
[0144] In some embodiments, the hydrogen leakage prediction module 50 is further configured to:
[0145] If both the current hydrogen concentration and the predicted hydrogen concentration do not exceed 0.5%, it is determined that there is no hydrogen leakage;
[0146] If the current hydrogen concentration or the predicted hydrogen concentration exceeds 0.5%, and both the current hydrogen concentration and the predicted hydrogen concentration do not exceed 4%, a primary hydrogen leak is determined;
[0147] If the current hydrogen concentration or the predicted hydrogen concentration exceeds 4%, and the leakage duration of the hydrogen exceeds 20 seconds, a secondary hydrogen leakage is determined.
[0148] In some embodiments, the hydrogen leakage prediction module 50 is further configured to:
[0149] If the hydrogen is not leaking, the green light of the sound and light alarm system lights up to provide a safety reminder;
[0150] If the hydrogen leaks at the first level, the yellow light of the sound and light alarm system lights up to give an alarm, and the forced ventilation of the emergency ventilation system is started;
[0151] If the hydrogen leaks at the second level, the yellow light of the sound and light alarm system will light up, the buzzer will start to alarm, and the forced ventilation of the emergency ventilation system will be started.
[0152] In some embodiments, the hydrogen leakage prediction module 50 is further configured to:
[0153] The concentration of the hydrogen is predicted within a preset time range.
[0154] The hydrogen leakage prediction device based on artificial neural network provided in the embodiment of the present disclosure corresponds to the hydrogen leakage prediction method based on artificial neural network in the above-mentioned embodiment. Any optional items in the embodiment of the hydrogen leakage prediction method based on artificial neural network are also applicable to the embodiment of the hydrogen leakage prediction device based on artificial neural network, which will not be repeated here.
[0155] Embodiment 3
[0156] An embodiment of the present disclosure also provides an electronic device, comprising at least a memory and a processor, wherein the memory stores a computer program, and the processor implements the above-mentioned hydrogen leakage prediction method based on artificial neural network when executing the computer program on the memory.
[0157] The above-mentioned electronic equipment is mainly used to analyze and transmit the measurement data detected by the quantum detector to obtain accurate gas absorption rate.
[0158] In some embodiments, the processor executing the computer program may be a processing device including one or more general-purpose processing devices, such as a microprocessor, a central processing unit (CPU), a graphics processing unit (GPU), etc. More specifically, the processor may be a complex instruction set computing (CISC) microprocessor, a reduced instruction set computing (RISC) microprocessor, a very long instruction word (VLIW) microprocessor, a processor running other instruction sets, or a processor running a combination of instruction sets. The processor may also be one or more special-purpose processing devices, such as an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), a digital signal processor (DSP), a system on a chip (SoC), etc.
[0159] The memory may be a read-only memory (ROM), a random access memory (RAM), a phase-change random access memory (PRAM), a static random access memory (SRAM), a dynamic random access memory (DRAM), an electrically erasable programmable read-only memory (EEPROM), other types of random access memory (RAM), a flash disk or other form of flash memory, a cache, a register, a static memory, a compact disk read-only memory (CD-ROM), a digital versatile disk (DVD) or other optical storage, a cassette or other magnetic storage device, or any other possible non-temporary medium used to store information or instructions that can be accessed by a computer device.
[0160] The electronic devices of the embodiments of the present disclosure may include but are not limited to fixed terminal devices such as servers, desktop computers, digital TVs, etc., and mobile terminal devices such as vehicle-mounted devices (such as head-up display devices), handheld devices (such as mobile phones, tablet computers, etc.), wearable devices (such as smart watches, smart bracelets, etc.), etc.
[0161] Embodiment 4
[0162] The embodiment of the present disclosure also provides a computer-readable storage medium, wherein the computer-readable medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned hydrogen leakage prediction method based on artificial neural network is implemented.
[0163] The computer-readable storage medium of the embodiments of the present disclosure may adopt any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. In the embodiments of the present disclosure, the computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, device or device, for example, the memory described above.
[0164] The computer programs of the embodiments of the present disclosure may be organized into one or more computer executable components or modules. Any number and combination of such components or modules may be used to implement the various aspects of the present disclosure. For example, the various aspects of the present disclosure are not limited to the specific computer executable instructions or specific components or modules shown in the drawings and described herein. Other embodiments may include different computer executable instructions or components with more or less functions than those shown and described herein.
[0165] The above description is only a preferred embodiment of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present disclosure is not limited to the technical solutions formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in the present disclosure (but not limited to) by each other to form a technical solution.
Claims
1. A hydrogen leakage prediction method based on artificial neural network, characterized in that: include: Establish a simulation model of the fuel cell cabin; Establishing a hydrogen leakage database according to the simulation model, wherein the data in the hydrogen leakage database includes the hydrogen concentration of each hydrogen detection point under different hydrogen leakage conditions; Training an artificial neural network according to the hydrogen leakage database to obtain a hydrogen leakage prediction model; Predicting the hydrogen concentration at a future time according to the hydrogen leakage prediction model to obtain a predicted hydrogen concentration; The hydrogen leakage situation is predicted based on the predicted hydrogen concentration and the current hydrogen concentration of the fuel cell cabin.
2. The method according to claim 1, characterized in that A hydrogen leakage database is established according to the simulation model, including: Determining a hydrogen leakage condition according to at least one factor of an opening and closing state of a hatch, a ventilation intensity of a vent, and a hydrogen leakage duration in the simulation model; Inputting the hydrogen leakage condition into the simulation model to obtain a simulation result; A hydrogen leakage database is established according to the simulation results.
3. The method according to claim 1, characterized in that The artificial neural network is trained according to the hydrogen leakage database to obtain a hydrogen leakage prediction model, including: The hydrogen concentration, average hydrogen concentration, and hydrogen concentration change rate extracted from the hydrogen leakage database are used as inputs of the NARX neural network, the predicted value of the hydrogen concentration at time t+1 is used as the output of the NARX neural network, and the NARX neural network is trained to obtain a time series prediction model of hydrogen concentration; Predicting the hydrogen concentration at a future time according to the hydrogen leakage prediction model includes: The historical data of the hydrogen concentration, average hydrogen concentration and hydrogen concentration change rate are input into the time series prediction model for single-step recursive prediction to predict the hydrogen concentration at time t+1.
4. The method according to claim 3, characterized in that When the hydrogen concentration time series prediction model is trained, the input delay number is 10, the output delay number is 5, the number of hidden layers is 1, the number of neurons is 10, and each logic gate inside the neuron uses a tanh activation function.
5. The method according to claim 1, characterized in that The method further comprises: According to the error between the predicted value and the measured value of the hydrogen concentration, a genetic algorithm is used to correct the hydrogen leakage prediction model.
6. The method according to claim 1, characterized in that Predicting hydrogen leakage according to the predicted hydrogen concentration and the current hydrogen concentration of the fuel cell cabin includes: Determining a hydrogen leakage level according to the predicted hydrogen concentration, the current hydrogen concentration in the fuel cell cabin, and the duration of the hydrogen leakage; The corresponding alarm system is activated according to the hydrogen leakage level to sound an alarm, and the alarm system includes an audible and visual alarm system and an emergency ventilation system.
7. The method according to claim 6, characterized in that Determining the hydrogen leakage level according to the predicted hydrogen concentration, the current hydrogen concentration of the fuel cell cabin, and the duration of hydrogen leakage includes: If both the current hydrogen concentration and the predicted hydrogen concentration do not exceed 0.5%, it is determined that there is no hydrogen leakage; If the current hydrogen concentration or the predicted hydrogen concentration exceeds 0.5%, and both the current hydrogen concentration and the predicted hydrogen concentration do not exceed 4%, a primary hydrogen leak is determined; If the current hydrogen concentration or the predicted hydrogen concentration exceeds 4%, and the leakage duration of the hydrogen exceeds 20 seconds, a secondary hydrogen leakage is determined.
8. The method according to claim 7, characterized in that According to the hydrogen leakage level, the corresponding alarm system is activated to alarm, including: If the hydrogen is not leaking, the green light of the sound and light alarm system lights up to provide a safety reminder; If the hydrogen leaks at the first level, the yellow light of the sound and light alarm system lights up to give an alarm, and the forced ventilation of the emergency ventilation system is started; If the hydrogen leaks at the second level, the yellow light of the sound and light alarm system will light up, the buzzer will start to alarm, and the forced ventilation of the emergency ventilation system will be started.
9. The method according to claim 7, characterized in that: After determining the hydrogen leak, the method further includes: The concentration of the hydrogen is predicted within a preset time range.
10. A hydrogen leakage prediction device based on artificial neural network, characterized in that: include: A first building module is configured to build a simulation model of a fuel cell cabin; A second establishing module is configured to establish a hydrogen leakage database according to the simulation model, wherein the data in the hydrogen leakage database includes the hydrogen concentration of each hydrogen detection point under different hydrogen leakage conditions; A training module, configured to train an artificial neural network according to the hydrogen leakage database to obtain a hydrogen leakage prediction model; A hydrogen concentration prediction module, configured to predict the hydrogen concentration at a future time according to the hydrogen leakage prediction model to obtain a predicted hydrogen concentration; The hydrogen leakage prediction module is configured to predict the hydrogen leakage situation according to the predicted hydrogen concentration and the current hydrogen concentration of the fuel cell cabin.
Citation Information
Cited By
Rapid deduction method and system for hydrogen leakage consequence of hydrogen refueling station
CN119989901A
