Hydrogen leak detection methods, detection systems, detection devices, and storage media

Abstract

This application provides a hydrogen leak detection method, detection system, detection device, and storage medium. The method involves uniformly arranging hydrogen sensors directly above a predetermined area, and also arranging hydrogen sensors at the edge of the predetermined area above it. This enhances the monitoring of hydrogen leaks at the edge of the predetermined area, thereby achieving comprehensive coverage of hydrogen leak monitoring across the entire predetermined area. Based on the data from all groups of sensors, it determines whether a hydrogen leak has occurred in the predetermined area. This shortens the reaction time for hydrogen leak detection and improves the timeliness of hydrogen leak detection, thus solving the problem of low timeliness in traditional hydrogen leak detection methods that typically rely on data from a single or a few sensors.

Description

Technical Field

[0001] This invention relates to the field of hydrogen leak detection technology, and more specifically, to a hydrogen leak detection method, a hydrogen leak detection system, a hydrogen leak detection device, a computer-readable storage medium, and a computer program product. Background Technology

[0002] The problems with traditional hydrogen leak detection methods are as follows:

[0003] Typically relying on data from a single or a few sensors, hydrogen leaks have a slow response time and weak detection capabilities in areas far from the leak point, resulting in low timeliness of hydrogen leak detection.

[0004] Ignoring the spatial connections between sensors and failing to effectively utilize the spatial layout information of the sensor network limits the accuracy of hydrogen leak location.

[0005] Currently, there is no solution to the above problems. Summary of the Invention

[0006] The main objective of this application is to provide a hydrogen leak detection method, a hydrogen leak detection system, a hydrogen leak detection device, a computer-readable storage medium, and a computer program product, so as to at least solve the problem that conventional hydrogen leak detection methods in the prior art usually rely on data from a single or a few sensors, resulting in low timeliness of hydrogen leak detection.

[0007] To achieve the above objectives, according to one aspect of this application, a hydrogen leak detection method is provided. The hydrogen leak detection system includes: multiple hydrogen sensors; a predetermined area composed of nine identical quadrilateral areas arranged in a 3x3 grid, wherein the quadrilateral areas are numbered sequentially in a serpentine order; a hydrogen sensor is disposed at a first predetermined distance directly above each vertex of the central quadrilateral area; a hydrogen sensor is disposed at a second predetermined distance directly above each first target point; the first target point is the intersection of any two quadrilateral areas other than the central quadrilateral area, and the first target point is located on the outer contour of the predetermined area; a hydrogen sensor is disposed at the second predetermined distance directly above the center point of the central quadrilateral area; and a hydrogen sensor is disposed at a third predetermined distance directly above each second target point. The hydrogen sensor is set at a certain distance. The second target point is the center point of the quadrilateral region that intersects with other quadrilateral regions on only two sides. A hydrogen sensor is set at a third predetermined distance directly above the center point of the quadrilateral region at the very center. The second predetermined distance is greater than the first predetermined distance, and the third predetermined distance is greater than the second predetermined distance. The method includes: acquiring multiple sets of sensor data in the current time period. Each set of sensor data includes: the three-dimensional coordinates of the hydrogen sensor in the predetermined coordinate system, the hydrogen pressure value, and the hydrogen concentration value. Each set of sensor data corresponds to one hydrogen sensor. The hydrogen pressure value and hydrogen concentration value in each set of sensor data are collected by the corresponding hydrogen sensor. Based on all sets of sensor data, it is determined whether a hydrogen leak has occurred in the predetermined area.

[0008] Optionally, determining whether hydrogen leakage has occurred in the predetermined area based on the sensor data of all groups includes: determining multiple leakage probabilities based on the sensor data of all groups, wherein each leakage probability corresponds one-to-one with a quadrilateral area, and the leakage probability is the probability of hydrogen leakage occurring in the corresponding quadrilateral area; determining whether each leakage probability satisfies a predetermined condition, wherein the predetermined condition is that the leakage probability is greater than a predetermined probability; and determining that hydrogen leakage has occurred in the predetermined area if at least one leakage probability satisfies the predetermined condition.

[0009] Optionally, after determining that a hydrogen leak has occurred in the predetermined area, the method further includes: obtaining the number of the target area, the target area being the quadrilateral area whose leakage probability satisfies the predetermined condition; generating a prompt message based on the number of the target area, the prompt message being used to characterize that a hydrogen leak has occurred in the target area, the prompt message including: the number of the target area.

[0010] Optionally, based on the sensor data from all groups, multiple leakage probabilities are determined, including: normalizing the sensor data from each group to obtain multiple sets of normalized sensor data, where each set of normalized sensor data corresponds to a set of sensor data; constructing a graph structure consisting of multiple nodes and multiple edges, where each node corresponds one-to-one with a hydrogen sensor, and each node represents the corresponding hydrogen sensor, each node including a node feature vector, the node feature vector including a set of normalized sensor data corresponding to the node, any two nodes being connected by an edge, the edge including an edge feature vector, the edge feature vector including the direction vectors of the two nodes connected by the edge and the distance between the two nodes connected by the edge; inputting the graph structure into a hydrogen leakage detection model to obtain multiple leakage probabilities, the leakage detection model being obtained by training a GCN model using multiple sets of training data, each set of training data including: the graph structure and the leakage probabilities acquired during historical periods.

[0011] Optionally, before inputting the graph structure into the hydrogen leak detection model to obtain multiple leak probabilities, the method further includes: establishing a target CFD model during the historical period, wherein the target CFD model is a three-dimensional CFD model of a predetermined space and a three-dimensional CFD model of the hydrogen leak detection system, the predetermined space being the space between the predetermined region and the top region, the top region having the same shape as the predetermined region, and the top region being located directly above the predetermined region and parallel to the predetermined region, and the distance between the top region and the predetermined region being the third predetermined distance; and setting multiple hydrogen leak probabilities. The hydrogen leak scenario includes at least: the three-dimensional coordinates of the point where the hydrogen leak occurs in the predetermined coordinate system, the rate of hydrogen leak, and the leakage probability of all the quadrilateral regions; in the target CFD model, each hydrogen leak scenario is run, and multiple hydrogen data sets are acquired. One hydrogen data set corresponds to one hydrogen leak scenario, and one hydrogen data set includes multiple sets of hydrogen data. One set of hydrogen data includes: hydrogen pressure value and hydrogen concentration value. One set of hydrogen data corresponds to one hydrogen sensor, and one set of hydrogen data is collected by the corresponding hydrogen sensor.

[0012] Optionally, inputting the graph structure into a hydrogen leak detection model to obtain multiple leak probabilities includes: inputting the graph structure into an input layer to obtain an initial node feature matrix, the initial node feature matrix being composed of all the node feature vectors; inputting the initial node feature matrix into a graph convolutional network to obtain a final node feature matrix, wherein the graph convolutional network is composed of N sequentially connected graph convolutional layers, the final node feature matrix being the node feature matrix output by the Nth graph convolutional layer, and the (l+1)th graph convolutional layer being: l is an integer, and -1 < l < N, H (0) H is the initial node feature matrix. (l+1) H is the node feature matrix output by the (l+1)th graph convolutional layer. (l) Let W be the node feature matrix output by the l-th graph convolutional layer, σ be the ReLU activation function, and W be the node feature matrix output by the l-th graph convolutional layer. (l) It is the weight matrix of the l-th graph convolutional layer. A is the adjacency matrix, A ij Let A be an element in A, and A ij Let 1 represent the communication connection between the i-th node and the j-th node, and A ij A value of 0 indicates that the i-th node and the j-th node are not connected by communication. D is a degree matrix, and the degree matrix is ​​a diagonal matrix. ii The number of nodes that are communicatively connected to the i-th node. It is e ij The weight matrix, e ij The edge feature vector is the edge feature vector contained in the edge connecting the i-th node and the j-th node, where N(i) is the set of nodes communicatively connected to the i-th node; the final node feature matrix is ​​input into a fully connected layer to obtain a classification feature vector, wherein the fully connected layer is: Z = Pooling(H (N) W fc +b fc Where Pooling() is global average pooling, W fc Let b be the weight matrix of the fully connected layer. fc H is the bias matrix of the fully connected layer. (N) Let Z be the node feature matrix output by the Nth graph convolutional layer, and Z be the classification feature vector. The classification feature vector is input into the output layer to obtain a probability vector, which is composed of all the leakage probabilities. The output layer is composed of a Softmax activation function, where the Softmax activation function is... in, Let Z be the probability vector. i Let M be the i-th element in the classification feature vector, and M be the total number of quadrilateral regions.

[0013] According to another aspect of this application, a hydrogen leak detection system is provided. The system includes: multiple hydrogen sensors, any two of which are communicatively connected. The hydrogen sensors are used to collect hydrogen pressure and concentration values. A predetermined area is composed of nine identical quadrilateral regions arranged in a 3x3 grid, with the quadrilateral regions numbered sequentially in a serpentine order. A hydrogen sensor is positioned at a first predetermined distance directly above each vertex of the central quadrilateral region. A hydrogen sensor is positioned at a second predetermined distance directly above each first target point. The first target point is the intersection of any two quadrilateral regions other than the central quadrilateral region, and the first target point is located on the outer contour of the predetermined area. The hydrogen sensors are positioned at the second predetermined distance directly above the center point of the central quadrilateral region. A hydrogen sensor is placed at a third predetermined distance directly above each second target point. The second target point is the center point of a quadrilateral region whose only two sides intersect with other quadrilateral regions. A hydrogen sensor is placed at the third predetermined distance directly above the center point of the quadrilateral region at the very center. The second predetermined distance is greater than the first predetermined distance, and the third predetermined distance is greater than the second predetermined distance. The origin of the predetermined coordinate system is the center point of the predetermined region. The horizontal axis of the predetermined coordinate system is parallel to one central axis of the predetermined region, the vertical axis of the predetermined coordinate system is parallel to the other central axis of the predetermined region, the vertical axis of the predetermined coordinate system is perpendicular to the horizontal axis of the predetermined coordinate system, and the vertical axis of the predetermined coordinate system is perpendicular to the vertical axis of the predetermined coordinate system.

[0014] According to another aspect of this application, a hydrogen leak detection device is provided. The hydrogen leak detection system includes: multiple hydrogen sensors; a predetermined area composed of nine identical quadrilateral areas arranged in a 3x3 grid, with the quadrilateral areas numbered sequentially in a serpentine order; a hydrogen sensor is disposed at a first predetermined distance directly above each vertex of the central quadrilateral area; a hydrogen sensor is disposed at a second predetermined distance directly above each first target point; the first target point is the intersection of any two quadrilateral areas excluding the central quadrilateral area, and the first target point is located on the outer contour of the predetermined area; a hydrogen sensor is disposed at the second predetermined distance directly above the center point of the central quadrilateral area; and a hydrogen sensor is disposed at a third predetermined distance directly above each second target point. The second target point is the center point of the quadrilateral region that intersects with other quadrilateral regions on only two sides. A hydrogen sensor is installed at a third predetermined distance directly above the center point of the quadrilateral region at its center. The second predetermined distance is greater than the first predetermined distance, and the third predetermined distance is greater than the second predetermined distance. The device includes: a first acquisition unit, used to acquire multiple sets of sensor data during the current time period. Each set of sensor data includes: the three-dimensional coordinates of the hydrogen sensor in the predetermined coordinate system, the hydrogen pressure value, and the hydrogen concentration value. Each set of sensor data corresponds to one hydrogen sensor, and the hydrogen pressure value and hydrogen concentration value in each set of sensor data are collected by the corresponding hydrogen sensor; and a determination unit, used to determine whether a hydrogen leak has occurred in the predetermined area based on all sets of sensor data.

[0015] According to another aspect of this application, a computer-readable storage medium is provided, the computer-readable storage medium including a stored program, wherein, when the program is executed, it controls the device where the computer-readable storage medium is located to perform any of the described hydrogen leak detection methods.

[0016] According to one aspect of this application, a computer program product is provided, including a computer program / instructions that, when executed by a processor, implement any of the described hydrogen leak detection methods.

[0017] By applying the technical solution of this application, hydrogen sensors are uniformly arranged directly above a predetermined area, and hydrogen sensors are also arranged at the edge directly above the predetermined area to enhance the monitoring of hydrogen leakage at the edge of the predetermined area. This achieves comprehensive coverage of hydrogen leakage monitoring for the entire predetermined area. Based on the data from all groups of sensors, it is determined whether a hydrogen leak has occurred in the predetermined area. This shortens the reaction time for hydrogen leaks and improves the timeliness of hydrogen leak detection. This solves the problem that traditional hydrogen leak detection methods in the prior art usually rely on data from a single or a few sensors, resulting in low timeliness of hydrogen leak detection. Attached Figure Description

[0018] Figure 1(a) shows a schematic diagram of the arrangement of the bottom hydrogen sensor according to an embodiment of the present application;

[0019] Figure 1(b) shows a schematic diagram of the arrangement of the hydrogen sensor in the middle according to an embodiment of this application;

[0020] Figure 1(c) shows a schematic diagram of the arrangement of the top hydrogen sensor according to an embodiment of the present application;

[0021] Figure 2 A schematic flowchart of a hydrogen leak detection method according to an embodiment of this application is shown;

[0022] Figure 3 A schematic diagram of a graph structure provided according to an embodiment of this application is shown;

[0023] Figure 4 A structural block diagram of a hydrogen leak detection model provided according to an embodiment of this application is shown;

[0024] Figure 5 A schematic diagram of a target CFD model provided according to an embodiment of this application is shown;

[0025] Figure 6 A structural block diagram of a hydrogen leak detection device provided according to an embodiment of this application is shown. Detailed Implementation

[0026] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0027] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0028] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus 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 apparatus.

[0029] For ease of description, the following explains some of the nouns or terms used in the embodiments of this application:

[0030] Hydrogen sensor: Hydrogen sensors typically integrate multiple sensing functions and can monitor the pressure and concentration of hydrogen in real time.

[0031] GCN (Graph Convolutional Network) model: It is a deep learning model for graph data analysis. It uses the structure of graph neural networks and convolutional operations to learn features and predict graph data. GCN model can effectively capture the relationships and local structure between nodes, thus achieving good performance in tasks such as node classification and link prediction.

[0032] CFD (Computational Fluid Dynamics) models are numerical simulation methods used to study the motion and heat transfer processes of fluids. By discretizing and numerically solving the fluid dynamics equations, CFD models can simulate the flow, heat transfer, and mass transfer processes of fluids in complex geometries, thereby analyzing information such as the flow field characteristics, temperature distribution, and pressure distribution of the fluid.

[0033] Computational Fluid Dynamics (ANSYS Fluent): Developed by ANSYS, this fluid dynamics simulation software is used to simulate and analyze various fluid flow phenomena. It provides powerful computing capabilities and diverse simulation tools, and can be used to solve problems related to fluid flow, heat transfer, and mass transfer of liquids, gases, etc. In ANSYS Fluent, CFD models are used to build mathematical models and solve fluid dynamics problems.

[0034] As described in the background section, conventional hydrogen leak detection methods in the prior art typically rely on data from a single or a few sensors, resulting in low timeliness of hydrogen leak detection. To address this problem, embodiments of this application provide a hydrogen leak detection method, a hydrogen leak detection system, a hydrogen leak detection device, a computer-readable storage medium, and a computer program product.

[0035] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0036] This embodiment provides a hydrogen leak detection system, the system comprising:

[0037] Multiple hydrogen sensors are provided, with any two sensors being communicatively connected. These sensors are used to collect hydrogen pressure and concentration values. The predetermined area consists of nine identical quadrilateral regions arranged in a 3x3 grid, with the quadrilateral regions numbered sequentially in a serpentine pattern. A hydrogen sensor is positioned at a first predetermined distance directly above each vertex of the central quadrilateral region, and a hydrogen sensor is positioned at a second predetermined distance directly above each first target point. The first target point is the intersection of any two quadrilateral regions other than the central one, and it lies on the outer contour of the predetermined area. A hydrogen sensor is positioned at the second predetermined distance directly above the center point of the central quadrilateral region. The hydrogen sensor is installed at a third predetermined distance directly above the target point. The second target point is the center point of the quadrilateral region that intersects with other quadrilateral regions on only two sides. A hydrogen sensor is installed at the third predetermined distance directly above the center point of the quadrilateral region at the center. The second predetermined distance is greater than the first predetermined distance, and the third predetermined distance is greater than the second predetermined distance. The origin of the predetermined coordinate system is the center point of the predetermined region. The horizontal axis of the predetermined coordinate system is parallel to one of the central axes of the predetermined region, the vertical axis of the predetermined coordinate system is parallel to the other central axis of the predetermined region, the vertical axis of the predetermined coordinate system is perpendicular to the horizontal axis of the predetermined coordinate system, and the vertical axis of the predetermined coordinate system is perpendicular to the vertical axis of the predetermined coordinate system.

[0038] Specifically, as shown in Figures 1(a), 1(b), and 1(c), in some embodiments, the predetermined area is the ground area of ​​the hydrogen fuel cell truck parking garage, which is a 60*60*8 meter area. This predetermined area is divided into 3 rows and 3 columns of quadrilateral areas, numbered sequentially from 1 to 9. Hydrogen leakage can occur anywhere within the hydrogen fuel cell truck parking garage, and it is not necessarily a single point leak; it may occur in combination. Considering the low density of hydrogen, it will rise rapidly when a leak occurs. Therefore, a hydrogen sensor in the height direction is necessary. For this purpose, three heights are defined: bottom (1 meter), middle (4 meters), and top (8 meters). The hydrogen sensor arrangement scheme is as follows: As shown in Figure 1(a), at the bottom height, the square indicates the position of the hydrogen sensor. One hydrogen sensor is arranged at each vertex of the quadrilateral area numbered 5. As shown in Figure 1(b), at the middle height, the triangle indicates the position of the hydrogen sensor. One hydrogen sensor is arranged at each first target point, and one hydrogen sensor is arranged at the center point of the quadrilateral area numbered 5. As shown in Figure 1(c), at the top height, the circle indicates the position of the hydrogen sensor. One hydrogen sensor is arranged at each second target point, and one hydrogen sensor is arranged at the center point of the quadrilateral area numbered 5.

[0039] Through the above embodiments, hydrogen sensors are uniformly arranged directly above the predetermined area, and hydrogen sensors are arranged at the edge directly above the predetermined area to enhance the monitoring of hydrogen leakage at the edge of the predetermined area, thereby achieving comprehensive coverage of hydrogen leakage monitoring of the entire predetermined area. This shortens the reaction time of hydrogen leakage and improves the timeliness of hydrogen leakage detection, thus solving the problem that traditional hydrogen leakage detection methods in the prior art usually rely on data from a single or a few sensors, resulting in low timeliness of hydrogen leakage detection.

[0040] This embodiment provides a method for detecting hydrogen leaks. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Also, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0041] Figure 2 This is a flowchart of a hydrogen leak detection method according to an embodiment of this application. Figure 2 As shown, the method includes the following steps:

[0042] Step S201: In the current time period, acquire multiple sets of sensor data. Each set of sensor data includes: the three-dimensional coordinates of the hydrogen sensor in the predetermined coordinate system, the pressure value of hydrogen, and the concentration value of hydrogen. Each set of sensor data corresponds to one hydrogen sensor. The pressure value of hydrogen and the concentration value of hydrogen in each set of sensor data are acquired by the corresponding hydrogen sensor.

[0043] Step S202: Based on the sensor data from all groups, determine whether a hydrogen leak has occurred in the predetermined area.

[0044] In one alternative approach, step S202 above can be implemented as follows:

[0045] Step S2021: Based on the sensor data of all groups, determine multiple leakage probabilities. The leakage probabilities correspond one-to-one with the quadrilateral regions. The leakage probability is the probability of hydrogen leakage occurring in the corresponding quadrilateral region.

[0046] In one alternative approach, step S2021 above can be implemented as follows:

[0047] Step S20211: Normalize each set of sensor data to obtain multiple sets of normalized sensor data, with each set of normalized sensor data corresponding to a set of sensor data.

[0048] Specifically, the three-dimensional coordinates of the hydrogen sensor in the predetermined coordinate system, the hydrogen pressure value, and the hydrogen concentration value in each set of sensor data are normalized to eliminate the influence of dimensions and unify the data scale. The normalization process can be min-maximum normalization or hydrogen leak detection score normalization.

[0049] Step S20212: Construct a graph structure. The graph structure consists of multiple nodes and multiple edges. Each node corresponds to a hydrogen sensor, and each node represents the corresponding hydrogen sensor. Each node includes a node feature vector, which includes a set of normalized sensor data corresponding to the node. Any two nodes are connected by an edge, and each edge includes an edge feature vector, which includes the direction vectors of the two nodes connected by the edge and the distance between the two nodes connected by the edge.

[0050] Specifically, the graph structure is as follows: Figure 3 As shown, a circle represents a node, a line segment represents an edge, and the node feature vector is N. i =[(x i ,y i ,z i ),F i ], where x iLet y be the x-coordinate of the i-th hydrogen detection sensor in the predetermined coordinate system. i Let z be the ordinate of the i-th hydrogen detection sensor in the predetermined coordinate system. i Let F be the vertical coordinate of the i-th hydrogen detection sensor in the predetermined coordinate system. i =[p i (tT),...,p i (t); c i (tT),...,c i [(t)], where T is the length of the current time period, and c i (tT) represents the hydrogen concentration value collected by the i-th hydrogen detection sensor at time tT, c i (t) represents the hydrogen concentration value collected by the i-th hydrogen detection sensor at time t, p i (tT) represents the hydrogen pressure value collected by the i-th hydrogen detection sensor at time tT, p i (t) represents the hydrogen pressure value collected by the i-th hydrogen detection sensor at time t. The connection between nodes is determined by the distance and direction vector between hydrogen sensors. All nodes in this application are interconnected, and the edge feature vector is e. ij =(v ij ,|v ij |), v ij =(x i -x j ,y i -y j ,z i -z j ) represents the direction vector from the i-th node to the j-th node, d ij =||v ij || represents the distance between the i-th node and the j-th node, x j Let y be the x-coordinate of the j-th hydrogen detection sensor in the predetermined coordinate system. j Let z be the ordinate of the j-th hydrogen detection sensor in the predetermined coordinate system. j Let be the vertical coordinate of the j-th hydrogen detection sensor in the predetermined coordinate system.

[0051] Step S20213: Input the above graph structure into the hydrogen leak detection model to obtain multiple leak probabilities. The leak detection model is obtained by training the GCN model with multiple sets of training data. Each set of training data includes the above graph structure and the above leak probabilities obtained in historical time periods.

[0052] Specifically, by utilizing a graph structure, the spatial relationships between hydrogen sensors (i.e., the nodes and edges of the graph structure) are explicitly utilized, allowing the GCN model to improve the accuracy of localization through spatial interactions. In this way, even if a hydrogen sensor in a quadrilateral region does not directly detect a hydrogen leak, the GCN model can infer the possibility of a hydrogen leak from the data collected by hydrogen sensors in neighboring quadrilateral regions. This solves the problem that traditional hydrogen leak detection methods in the prior art ignore the spatial connections between sensors, cannot effectively utilize the spatial layout information of the sensor network, and limit the accuracy of hydrogen leak localization.

[0053] In an alternative approach, step S20213 above can be implemented as follows:

[0054] The above graph structure is input into the input layer to obtain the initial node feature matrix, which is composed of all the above node feature vectors.

[0055] The initial node feature matrix is ​​input into the graph convolutional network to obtain the final node feature matrix. The graph convolutional network consists of N sequentially connected graph convolutional layers. The final node feature matrix is ​​the node feature matrix output by the Nth graph convolutional layer, and the (l+1)th graph convolutional layer is: l is an integer, and -1 < l < N, H (0) Let H be the initial node feature matrix mentioned above. (l+1) H is the node feature matrix output by the (l+1)th convolutional layer of the above graph. (l) Let W be the node feature matrix output by the l-th convolutional layer of the above graph, where σ is the ReLU activation function, and W is the node feature matrix. (l) It is the weight matrix of the l-th convolutional layer of the above graph. A is the adjacency matrix, A ij Let A be an element in A, and A ij Let 1 represent the communication connection between the i-th node and the j-th node, and A ij A value of 0 indicates that the i-th node and the j-th node are not connected. D is a degree matrix, and the degree matrix is ​​a diagonal matrix. ii The number of nodes that are connected to communicate with the i-th node. It is e ij The weight matrix, e ij The edge feature vector is the edge that connects the i-th node and the j-th node, and N(i) is the set of nodes that are communicatively connected to the i-th node.

[0056] Specifically, in this application, each hydrogen sensor is connected to all other hydrogen sensors, as shown in Figures 1(a), 1(b), and 1(c). Eighteen hydrogen sensors are set up. At this time, the adjacency matrix is ​​an 18*18 matrix, and all elements except the diagonal are 1.

[0057] The final node feature matrix is ​​input into a fully connected layer to obtain the classification feature vector, where the fully connected layer is: Z = Pooling(H (N) W fc +b fc Where Pooling() is global average pooling, W fc Let b be the weight matrix of the fully connected layer mentioned above. fc H is the bias matrix of the fully connected layer mentioned above. (N) Z is the node feature matrix output by the Nth graph convolutional layer mentioned above, and Z is the classification feature vector mentioned above.

[0058] The aforementioned classification feature vector is input into the output layer to obtain a probability vector, which consists of all the aforementioned leakage probabilities. The output layer is composed of a Softmax activation function, which is... in, Let Z be the probability vector mentioned above. i Let be the i-th element in the above classification feature vector, and M be the total number of the above quadrilateral regions.

[0059] Specifically, such as Figure 4 As shown in Figures 1(a), 1(b), and 1(c), the hydrogen leak detection model consists of an input layer, a first graph convolutional layer, a second graph convolutional layer, a fully connected layer, and an output layer. With 18 hydrogen sensors, the initial node feature matrix output by the input layer consists of 18 node feature vectors, with a dimension of 18×(3+2T). The output layer outputs the initial node feature matrix to the graph convolutional network. The node feature matrix output by the first graph convolutional layer has a dimension of 18×64, and the node feature matrix output by the second graph convolutional layer has a dimension of 18×32, resulting in a final node feature matrix of 18×32. The graph convolutional network outputs the final node feature matrix to the fully connected layer. The classification feature vector output by the fully connected layer has a dimension of 1×9. The fully connected layer then outputs the classification feature vector to the output layer to obtain a probability vector, also with a dimension of 1×9. Let be the leakage probability of quadrilateral region numbered 1. Let be the leakage probability of quadrilateral region number 2. This represents the leakage probability of the quadrilateral region numbered 9.

[0060] In an alternative embodiment, prior to step S20213, the method further includes:

[0061] During the aforementioned historical period, a target CFD model is established. The target CFD model is a three-dimensional CFD model of a predetermined space and a three-dimensional CFD model of the hydrogen leak detection system. The predetermined space is the space between the predetermined area and the top area. The top area has the same shape as the predetermined area and is located directly above the predetermined area and parallel to the predetermined area. The distance between the top area and the predetermined area is the third predetermined distance.

[0062] Specifically, in some implementations, the designated space is a parking garage for hydrogen fuel cell trucks, and the designated area is the ground area of ​​the parking garage for hydrogen fuel cell trucks.

[0063] Multiple hydrogen leak scenarios are set up, and the hydrogen leak scenarios include at least: the three-dimensional coordinates of the point where the hydrogen leak occurs in the predetermined coordinate system, the rate of hydrogen leak, and the leakage probability of all the quadrilateral regions.

[0064] Specifically, the hydrogen leak scenario simulates the diffusion behavior of hydrogen under various leak conditions. The hydrogen leak scenario also includes: the initial pressure value of hydrogen, space obstacles, temperature, and humidity, such as... Figure 5 As shown, the target CFD model is a three-dimensional CFD model of a hydrogen fuel cell vehicle parking garage and a hydrogen leak detection system. The hydrogen leak scenario simulation shows hydrogen leaks occurring at leak point 1 and leak point 2.

[0065] In the aforementioned target CFD model, each of the aforementioned hydrogen leakage scenarios is run, and multiple hydrogen data sets are acquired. One hydrogen data set corresponds to one of the aforementioned hydrogen leakage scenarios. One hydrogen data set includes multiple sets of hydrogen data. One set of hydrogen data includes: hydrogen pressure value and hydrogen concentration value. One set of hydrogen data corresponds to one of the aforementioned hydrogen sensors. One set of hydrogen data is collected by the corresponding aforementioned hydrogen sensor.

[0066] Specifically, such as Figure 5 As shown, the target CFD model is a three-dimensional CFD model of a hydrogen fuel cell vehicle parking garage and a hydrogen leak detection system. Using computational fluid dynamics software, various hydrogen leak scenarios are run in the target CFD model to record the changes in hydrogen concentration over time for each hydrogen sensor under different hydrogen leak scenarios, thus obtaining multiple hydrogen data sets.

[0067] Specifically, for a hydrogen leak scenario, a hydrogen data set is obtained, and the GCN model is trained using this hydrogen data set to obtain a hydrogen leak detection model.

[0068] Specifically, this application trains the GCN model based on various hydrogen data sets obtained from multiple hydrogen leakage scenarios, which improves the adaptability of the hydrogen leakage detection model to environmental changes, enhances the accuracy of the hydrogen leakage detection model, and ensures the stability and reliability of the hydrogen leakage detection model.

[0069] Step S2022: Determine whether each of the above-mentioned leakage probabilities meets a predetermined condition, wherein the predetermined condition is that the above-mentioned leakage probability is greater than a predetermined probability.

[0070] Step S2023: If at least one of the above-mentioned leakage probabilities meets the above-mentioned predetermined conditions, it is determined that a hydrogen leak has occurred in the above-mentioned predetermined area.

[0071] Specifically, if at least one leakage probability is greater than a predetermined probability, a hydrogen leak occurs in the quadrilateral region to which the leakage probability belongs, thereby determining that a hydrogen leak has occurred in the predetermined region.

[0072] In an alternative embodiment, after step S2023, the method further includes:

[0073] Obtain the target region number, where the target region is the quadrilateral region whose leakage probability satisfies the predetermined condition.

[0074] Based on the target area number, a prompt message is generated. The prompt message indicates that a hydrogen leak has occurred in the target area. The prompt message includes the target area number.

[0075] Specifically, a prompt message is generated based on the target area number, that is, based on the number of the quadrilateral area where the hydrogen leak occurred, a prompt message is generated to remind staff of the specific location of the hydrogen leak.

[0076] Through the above embodiments, hydrogen sensors are uniformly arranged directly above the predetermined area, and hydrogen sensors are also arranged at the edge directly above the predetermined area to enhance the monitoring of hydrogen leakage at the edge of the predetermined area. This achieves comprehensive coverage of hydrogen leakage monitoring for the entire predetermined area. Based on the data from all groups of sensors, it is determined whether a hydrogen leak has occurred in the predetermined area. This shortens the reaction time for hydrogen leaks and improves the timeliness of hydrogen leak detection. This solves the problem that traditional hydrogen leak detection methods in the prior art usually rely on data from a single or a few sensors, resulting in low timeliness of hydrogen leak detection.

[0077] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0078] This application also provides a hydrogen leak detection device. It should be noted that the hydrogen leak detection device of this application can be used to execute the hydrogen leak detection method provided in this application. This device is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0079] The hydrogen leak detection device provided in the embodiments of this application will be described below.

[0080] Figure 6 This is a structural block diagram of a hydrogen leak detection device according to an embodiment of this application. Figure 6 As shown, the device includes:

[0081] The first acquisition unit 10 is used to acquire multiple sets of sensor data in the current time period. Each set of sensor data includes: the three-dimensional coordinates of the hydrogen sensor in the predetermined coordinate system, the pressure value of hydrogen, and the concentration value of hydrogen. Each set of sensor data corresponds to one hydrogen sensor. The pressure value of hydrogen and the concentration value of hydrogen in each set of sensor data are acquired by the corresponding hydrogen sensor.

[0082] The determining unit 20 is used to determine whether a hydrogen leak has occurred in the predetermined area based on the sensor data of all groups.

[0083] In one alternative embodiment, the aforementioned determining unit includes:

[0084] The first determining subunit is used to determine multiple leakage probabilities based on the sensor data of all groups, wherein each leakage probability corresponds one-to-one with the quadrilateral region, and the leakage probability is the probability of hydrogen leakage occurring in the corresponding quadrilateral region.

[0085] In one alternative embodiment, the first determined subunit includes:

[0086] The processing module is used to perform normalization processing on each group of the above-mentioned sensor data to obtain multiple groups of normalized sensor data, and each group of the above-mentioned normalized sensor data corresponds to a group of the above-mentioned sensor data.

[0087] Specifically, the three-dimensional coordinates of the hydrogen sensor in the predetermined coordinate system, the hydrogen pressure value, and the hydrogen concentration value in each set of sensor data are normalized to eliminate the influence of dimensions and unify the data scale. The normalization process can be min-maximum normalization or hydrogen leak detection score normalization.

[0088] The construction module is used to construct a graph structure, which consists of multiple nodes and multiple edges. Each node corresponds one-to-one with a hydrogen sensor, and each node represents the corresponding hydrogen sensor. Each node includes a node feature vector, which includes a set of normalized sensor data corresponding to the node. Any two nodes are connected by an edge, and each edge includes an edge feature vector, which includes the direction vectors of the two nodes connected by the edge and the distance between the two nodes connected by the edge.

[0089] Specifically, the graph structure is as follows: Figure 3 As shown, a circle represents a node, a line segment represents an edge, and the node feature vector is N. i =[(x i ,y i ,z i ),F i ], where x i Let y be the x-coordinate of the i-th hydrogen detection sensor in the predetermined coordinate system. i Let z be the ordinate of the i-th hydrogen detection sensor in the predetermined coordinate system. i Let F be the vertical coordinate of the i-th hydrogen detection sensor in the predetermined coordinate system. i =[p i (tT),...,p i (t); c i (tT),...,c i [(t)], where T is the length of the current time period, and c i (tT) represents the hydrogen concentration value collected by the i-th hydrogen detection sensor at time tT, c i (t) represents the hydrogen concentration value collected by the i-th hydrogen detection sensor at time t, p i (tT) represents the hydrogen pressure value collected by the i-th hydrogen detection sensor at time tT, p i (t) represents the hydrogen pressure value collected by the i-th hydrogen detection sensor at time t. The connection between nodes is determined by the distance and direction vector between hydrogen sensors. All nodes in this application are interconnected, and the edge feature vector is e. ij =(v ij ,||v ij ||), v ij =(x i -x j ,y i -y j ,z i -z j ) represents the direction vector from the i-th node to the j-th node, d ij =||vij || represents the distance between the i-th node and the j-th node, x j Let y be the x-coordinate of the j-th hydrogen detection sensor in the predetermined coordinate system. j Let z be the ordinate of the j-th hydrogen detection sensor in the predetermined coordinate system. j Let be the vertical coordinate of the j-th hydrogen detection sensor in the predetermined coordinate system.

[0090] The input module is used to input the above graph structure into the hydrogen leak detection model to obtain multiple leak probabilities. The leak detection model is obtained by training the GCN model with multiple sets of training data. Each set of training data includes the above graph structure and the above leak probabilities obtained in historical time periods.

[0091] Specifically, by utilizing a graph structure, the spatial relationships between hydrogen sensors (i.e., the nodes and edges of the graph structure) are explicitly utilized, allowing the GCN model to improve the accuracy of localization through spatial interactions. In this way, even if a hydrogen sensor in a quadrilateral region does not directly detect a hydrogen leak, the GCN model can infer the possibility of a hydrogen leak from the data collected by hydrogen sensors in neighboring quadrilateral regions. This solves the problem that traditional hydrogen leak detection methods in the prior art ignore the spatial connections between sensors, cannot effectively utilize the spatial layout information of the sensor network, and limit the accuracy of hydrogen leak localization.

[0092] In one alternative embodiment, the input module includes:

[0093] The first input submodule is used to input the above graph structure into the input layer to obtain an initial node feature matrix, which is composed of all the above node feature vectors.

[0094] The second input submodule is used to input the initial node feature matrix into the graph convolutional network to obtain the final node feature matrix. The graph convolutional network consists of N sequentially connected graph convolutional layers. The final node feature matrix is ​​the node feature matrix output by the Nth graph convolutional layer, and the (l+1)th graph convolutional layer is: l is an integer, and -1 < l < N, H (0) Let H be the initial node feature matrix mentioned above. (l +1) H is the node feature matrix output by the (l+1)th convolutional layer of the above graph. (l) Let W be the node feature matrix output by the l-th convolutional layer of the above graph, where σ is the ReLU activation function, and W is the node feature matrix. (l) It is the weight matrix of the l-th convolutional layer of the above graph. A is the adjacency matrix, A ij Let A be an element in A, and A ijLet 1 represent the communication connection between the i-th node and the j-th node, and A ij A value of 0 indicates that the i-th node and the j-th node are not connected. D is a degree matrix, and the degree matrix is ​​a diagonal matrix. ii The number of nodes that are connected to communicate with the i-th node. It is e ij The weight matrix, e ij The edge feature vector is the edge that connects the i-th node and the j-th node, and N(i) is the set of nodes that are communicatively connected to the i-th node.

[0095] Specifically, in this application, each hydrogen sensor is connected to all other hydrogen sensors, as shown in Figures 1(a), 1(b), and 1(c). Eighteen hydrogen sensors are set up. At this time, the adjacency matrix is ​​an 18*18 matrix, and all elements except the diagonal are 1.

[0096] The third input submodule is used to input the final node feature matrix into the fully connected layer to obtain the classification feature vector, wherein the fully connected layer is: Z = Pooling(H (N) W fc +b fc Where Pooling() is global average pooling, W fc Let b be the weight matrix of the fully connected layer mentioned above. fc H is the bias matrix of the fully connected layer mentioned above. (N) Z is the node feature matrix output by the Nth graph convolutional layer mentioned above, and Z is the classification feature vector mentioned above.

[0097] The fourth input submodule is used to input the aforementioned classification feature vector into the output layer to obtain a probability vector. This probability vector consists of all the aforementioned leakage probabilities. The output layer is composed of a Softmax activation function, which is... in, Let Z be the probability vector mentioned above. i Let be the i-th element in the above classification feature vector, and M be the total number of the above quadrilateral regions.

[0098] Specifically, such as Figure 4As shown in Figures 1(a), 1(b), and 1(c), the hydrogen leak detection model consists of an input layer, a first graph convolutional layer, a second graph convolutional layer, a fully connected layer, and an output layer. With 18 hydrogen sensors, the initial node feature matrix output by the input layer consists of 18 node feature vectors, with a dimension of 18×(3+2T). The output layer outputs the initial node feature matrix to the graph convolutional network. The node feature matrix output by the first graph convolutional layer has a dimension of 18×64, and the node feature matrix output by the second graph convolutional layer has a dimension of 18×32, resulting in a final node feature matrix of 18×32. The graph convolutional network outputs the final node feature matrix to the fully connected layer. The classification feature vector output by the fully connected layer has a dimension of 1×9. The fully connected layer then outputs the classification feature vector to the output layer to obtain a probability vector, also with a dimension of 1×9. Let be the leakage probability of quadrilateral region numbered 1. Let be the leakage probability of quadrilateral region number 2. This represents the leakage probability of the quadrilateral region numbered 9.

[0099] In one alternative embodiment, the above-mentioned apparatus further includes:

[0100] The establishment unit is used to establish a target CFD model during the aforementioned historical period. The target CFD model is a three-dimensional CFD model of a predetermined space and a three-dimensional CFD model of the hydrogen leak detection system. The predetermined space is the space between the predetermined area and the top area. The top area has the same shape as the predetermined area and is located directly above the predetermined area and parallel to the predetermined area. The distance between the top area and the predetermined area is the third predetermined distance.

[0101] Specifically, in some implementations, the designated space is a parking garage for hydrogen fuel cell trucks, and the designated area is the ground area of ​​the parking garage for hydrogen fuel cell trucks.

[0102] The setting unit is used to set multiple hydrogen leakage scenarios, which include at least: the three-dimensional coordinates of the point where the hydrogen leakage occurs in the predetermined coordinate system, the rate of hydrogen leakage, and the leakage probability of all the quadrilateral regions.

[0103] Specifically, the hydrogen leak scenario simulates the diffusion behavior of hydrogen under various leak conditions. The hydrogen leak scenario also includes: the initial pressure value of hydrogen, space obstacles, temperature, and humidity, such as... Figure 5 As shown, the target CFD model is a three-dimensional CFD model of a hydrogen fuel cell vehicle parking garage and a hydrogen leak detection system. The hydrogen leak scenario simulation shows hydrogen leaks occurring at leak point 1 and leak point 2.

[0104] The running unit is used to run each of the above-mentioned hydrogen leakage scenarios in the above-mentioned target CFD model and acquire multiple hydrogen data sets. One hydrogen data set corresponds to one of the above-mentioned hydrogen leakage scenarios. One hydrogen data set includes multiple sets of hydrogen data. One set of hydrogen data includes: hydrogen pressure value and hydrogen concentration value. One set of hydrogen data corresponds to one of the above-mentioned hydrogen sensors. One set of hydrogen data is collected by the corresponding hydrogen sensor.

[0105] Specifically, such as Figure 5 As shown, the target CFD model is a three-dimensional CFD model of a hydrogen fuel cell vehicle parking garage and a hydrogen leak detection system. Using computational fluid dynamics software, various hydrogen leak scenarios are run in the target CFD model to record the changes in hydrogen concentration over time for each hydrogen sensor under different hydrogen leak scenarios, thus obtaining multiple hydrogen data sets.

[0106] Specifically, for a hydrogen leak scenario, a hydrogen data set is obtained, and the GCN model is trained using this hydrogen data set to obtain a hydrogen leak detection model.

[0107] Specifically, this application trains the GCN model based on various hydrogen data sets obtained from multiple hydrogen leakage scenarios, which improves the adaptability of the hydrogen leakage detection model to environmental changes, enhances the accuracy of the hydrogen leakage detection model, and ensures the stability and reliability of the hydrogen leakage detection model.

[0108] The second determining subunit is used to determine whether each of the above-mentioned leakage probabilities meets a predetermined condition, wherein the predetermined condition is that the above-mentioned leakage probability is greater than a predetermined probability.

[0109] The third determining subunit is used to determine that a hydrogen leak has occurred in the predetermined area if at least one of the above-mentioned leakage probabilities meets the above-mentioned predetermined conditions.

[0110] Specifically, if at least one leakage probability is greater than a predetermined probability, a hydrogen leak occurs in the quadrilateral region to which the leakage probability belongs, thereby determining that a hydrogen leak has occurred in the predetermined region.

[0111] In one alternative embodiment, the above-mentioned apparatus further includes:

[0112] The second acquisition unit is used to acquire the number of the target area, wherein the target area is the quadrilateral area whose leakage probability satisfies the predetermined condition.

[0113] The generation unit is used to generate a prompt message based on the number of the target area. The prompt message is used to indicate that a hydrogen leak has occurred in the target area. The prompt message includes the number of the target area.

[0114] Specifically, a prompt message is generated based on the target area number, that is, based on the number of the quadrilateral area where the hydrogen leak occurred, a prompt message is generated to remind staff of the specific location of the hydrogen leak.

[0115] Through the above embodiments, hydrogen sensors are uniformly arranged directly above the predetermined area, and hydrogen sensors are also arranged at the edge directly above the predetermined area to enhance the monitoring of hydrogen leakage at the edge of the predetermined area. This achieves comprehensive coverage of hydrogen leakage monitoring for the entire predetermined area. Based on the data from all groups of sensors, it is determined whether a hydrogen leak has occurred in the predetermined area. This shortens the reaction time for hydrogen leaks and improves the timeliness of hydrogen leak detection. This solves the problem that traditional hydrogen leak detection methods in the prior art usually rely on data from a single or a few sensors, resulting in low timeliness of hydrogen leak detection.

[0116] The aforementioned hydrogen leak detection device includes a processor and a memory. The first acquisition unit and the determination unit, among others, are stored as program units in the memory. The processor executes these program units stored in the memory to achieve the corresponding functions. All of the above modules reside in the same processor; alternatively, the modules may be located in different processors in any combination.

[0117] The processor contains a kernel, which retrieves the corresponding program units from memory. One or more kernels can be configured, and adjusting kernel parameters can address the problem of low detection timeliness in traditional hydrogen leak detection methods that typically rely on data from a single or a few sensors.

[0118] The memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0119] This invention provides a computer-readable storage medium including a stored program, wherein the program, when running, controls the device containing the computer-readable storage medium to perform the hydrogen leak detection method.

[0120] Specifically, hydrogen leak detection methods include:

[0121] Step S201: In the current time period, acquire multiple sets of sensor data. Each set of sensor data includes: the three-dimensional coordinates of the hydrogen sensor in the predetermined coordinate system, the pressure value of hydrogen, and the concentration value of hydrogen. Each set of sensor data corresponds to one hydrogen sensor. The pressure value of hydrogen and the concentration value of hydrogen in each set of sensor data are acquired by the corresponding hydrogen sensor.

[0122] Step S202: Based on the sensor data from all groups, determine whether a hydrogen leak has occurred in the predetermined area.

[0123] This invention provides a processor for running a program, wherein the program executes the hydrogen leak detection method described above.

[0124] Specifically, hydrogen leak detection methods include:

[0125] Step S201: In the current time period, acquire multiple sets of sensor data. Each set of sensor data includes: the three-dimensional coordinates of the hydrogen sensor in the predetermined coordinate system, the pressure value of hydrogen, and the concentration value of hydrogen. Each set of sensor data corresponds to one hydrogen sensor. The pressure value of hydrogen and the concentration value of hydrogen in each set of sensor data are acquired by the corresponding hydrogen sensor.

[0126] Step S202: Based on the sensor data from all groups, determine whether a hydrogen leak has occurred in the predetermined area.

[0127] This invention provides a device including a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it performs at least the following steps:

[0128] Step S201: In the current time period, acquire multiple sets of sensor data. Each set of sensor data includes: the three-dimensional coordinates of the hydrogen sensor in the predetermined coordinate system, the pressure value of hydrogen, and the concentration value of hydrogen. Each set of sensor data corresponds to one hydrogen sensor. The pressure value of hydrogen and the concentration value of hydrogen in each set of sensor data are acquired by the corresponding hydrogen sensor.

[0129] Step S202: Based on the sensor data from all groups, determine whether a hydrogen leak has occurred in the predetermined area.

[0130] This application also provides a computer program product, which, when executed on a data processing device, is suitable for executing an initialization program having at least the following method steps:

[0131] Step S201: In the current time period, acquire multiple sets of sensor data. Each set of sensor data includes: the three-dimensional coordinates of the hydrogen sensor in the predetermined coordinate system, the pressure value of hydrogen, and the concentration value of hydrogen. Each set of sensor data corresponds to one hydrogen sensor. The pressure value of hydrogen and the concentration value of hydrogen in each set of sensor data are acquired by the corresponding hydrogen sensor.

[0132] Step S202: Based on the sensor data from all groups, determine whether a hydrogen leak has occurred in the predetermined area.

[0133] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. They can be implemented using computer-executable program code, and thus can be stored in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those described herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0134] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0135] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in one or more flowchart illustrations and / or one or more block diagrams.

[0136] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means that implement the functions specified in one or more flowcharts and / or one or more block diagrams.

[0137] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, such that the instructions, which execute on the computer or other programmable apparatus, provide steps for implementing the functions specified in one or more flowcharts and / or one or more block diagrams.

[0138] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0139] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0140] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0141] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0142] As can be seen from the above description, the embodiments of this application achieve the following technical effects:

[0143] 1) In the hydrogen leak detection method of this application, hydrogen sensors are evenly arranged directly above a predetermined area, and hydrogen sensors are arranged at the edge directly above the predetermined area to enhance the monitoring of hydrogen leaks at the edge of the predetermined area, thereby achieving comprehensive coverage of hydrogen leak monitoring of the entire predetermined area. Based on the sensor data of all groups, it is determined whether a hydrogen leak has occurred in the predetermined area, which can shorten the reaction time of hydrogen leak detection and improve the timeliness of hydrogen leak detection. This solves the problem that the traditional hydrogen leak detection methods in the prior art usually rely on the data of a single or a few sensors, resulting in low timeliness of hydrogen leak detection.

[0144] 2) In the hydrogen leak detection system of this application, hydrogen sensors are evenly arranged directly above the predetermined area, and hydrogen sensors are arranged at the edge directly above the predetermined area to enhance the monitoring of hydrogen leaks at the edge of the predetermined area, thereby achieving comprehensive coverage of hydrogen leak monitoring of the entire predetermined area. Based on the data of all groups of sensors, it is determined whether a hydrogen leak has occurred in the predetermined area, which can shorten the reaction time of hydrogen leak detection and improve the timeliness of hydrogen leak detection. This solves the problem that the traditional hydrogen leak detection methods in the prior art usually rely on the data of a single or a few sensors, resulting in low timeliness of hydrogen leak detection.

[0145] 3) In the hydrogen leak detection device of this application, hydrogen sensors are evenly arranged directly above the predetermined area, and hydrogen sensors are arranged at the edge directly above the predetermined area to enhance the monitoring of hydrogen leaks at the edge of the predetermined area, thereby achieving comprehensive coverage of hydrogen leak monitoring of the entire predetermined area. Based on the data of all groups of sensors, it is determined whether a hydrogen leak has occurred in the predetermined area, which can shorten the reaction time of hydrogen leaks and improve the timeliness of hydrogen leak detection. This solves the problem that the traditional hydrogen leak detection methods in the prior art usually rely on the data of a single or a few sensors, resulting in low timeliness of hydrogen leak detection.

[0146] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for detecting hydrogen leaks, characterized in that, A hydrogen leak detection system includes: multiple hydrogen sensors; a predetermined area composed of nine identical quadrilateral regions arranged in a 3x3 grid, where the quadrilaterals are squares and the quadrilateral regions are numbered sequentially in a serpentine order; a hydrogen sensor is positioned at a first predetermined distance directly above each vertex of the central quadrilateral region; a hydrogen sensor is positioned at a second predetermined distance directly above each first target point; the first target point is the intersection of any two quadrilateral regions excluding the central quadrilateral region, and the first target point is located on the outer contour of the predetermined area; a hydrogen sensor is positioned at the second predetermined distance directly above the center point of the central quadrilateral region; a hydrogen sensor is positioned at a third predetermined distance directly above each second target point; the second target point is the center point of a quadrilateral region whose only two sides intersect with other quadrilateral regions; a hydrogen sensor is positioned at the third predetermined distance directly above the center point of the central quadrilateral region; the second predetermined distance is greater than the first predetermined distance, and the third predetermined distance is greater than the second predetermined distance; the method includes: During the current time period, multiple sets of sensor data are acquired. Each set of sensor data includes: the three-dimensional coordinates of the hydrogen sensor in a predetermined coordinate system, the pressure value of hydrogen, and the concentration value of hydrogen. Each set of sensor data corresponds to one hydrogen sensor. The pressure value and concentration value of hydrogen in each set of sensor data are collected by the corresponding hydrogen sensor. Based on the sensor data from all groups, determine whether a hydrogen leak has occurred in the designated area. Determining whether hydrogen leakage has occurred in the predetermined area based on the sensor data from all groups includes: determining multiple leakage probabilities based on the sensor data from all groups, wherein each leakage probability corresponds one-to-one with a quadrilateral area, and the leakage probability is the probability that hydrogen leakage has occurred in the corresponding quadrilateral area; determining whether each leakage probability satisfies a predetermined condition, wherein the predetermined condition is that the leakage probability is greater than a predetermined probability; and determining that hydrogen leakage has occurred in the predetermined area if at least one leakage probability satisfies the predetermined condition. Based on the sensor data from all groups, multiple leakage probabilities are determined, including: normalizing the sensor data from each group to obtain multiple sets of normalized sensor data, where each set of normalized sensor data corresponds to a set of sensor data; constructing a graph structure consisting of multiple nodes and multiple edges, where each node corresponds one-to-one with a hydrogen sensor, and each node represents the corresponding hydrogen sensor; each node includes a node feature vector, which includes the set of normalized sensor data corresponding to the node; any two nodes are connected by an edge, which includes an edge feature vector, which includes the direction vectors of the two nodes connected by the edge and the distance between the two nodes connected by the edge; inputting the graph structure into a hydrogen leakage detection model to obtain multiple leakage probabilities, where the leakage detection model is obtained by training a GCN model using multiple sets of training data, each set of training data including the graph structure and the leakage probabilities acquired during historical periods.

2. The method according to claim 1, characterized in that, After determining that a hydrogen leak has occurred in the predetermined area, the method further includes: Obtain the number of the target region, where the target region is the quadrilateral region whose leakage probability satisfies the predetermined condition; Based on the target area number, a prompt message is generated. The prompt message is used to indicate that a hydrogen leak has occurred in the target area. The prompt message includes: the target area number.

3. The method according to claim 1, characterized in that, Before inputting the graph structure into the hydrogen leak detection model to obtain multiple leak probabilities, the method further includes: During the historical period, a target CFD model is established. The target CFD model is a three-dimensional CFD model of a predetermined space and a three-dimensional CFD model of the hydrogen leak detection system. The predetermined space is the space between the predetermined area and the top area. The top area has the same shape as the predetermined area, and the top area is located directly above the predetermined area and parallel to the predetermined area. The distance between the top area and the predetermined area is the third predetermined distance. Multiple hydrogen leak scenarios are set up, and the hydrogen leak scenarios include at least: the three-dimensional coordinates of the point where the hydrogen leak occurs in the predetermined coordinate system, the rate of hydrogen leak, and the leak probability of all the quadrilateral regions; In the target CFD model, each hydrogen leak scenario is run, and multiple hydrogen data sets are acquired. Each hydrogen data set corresponds to one hydrogen leak scenario. Each hydrogen data set includes multiple sets of hydrogen data. Each set of hydrogen data includes: hydrogen pressure value and hydrogen concentration value. Each set of hydrogen data corresponds to one hydrogen sensor. Each set of hydrogen data is collected by the corresponding hydrogen sensor.

4. The method according to claim 1, characterized in that, Inputting the graph structure into the hydrogen leak detection model yields multiple leak probabilities, including: The graph structure is input into the input layer to obtain an initial node feature matrix, which is composed of all the node feature vectors. The initial node feature matrix is ​​input into a graph convolutional network to obtain the final node feature matrix, wherein the graph convolutional network consists of sequentially connected nodes. The graph is composed of convolutional layers, and the final node feature matrix is ​​the first convolutional layer. The node feature matrix output by the graph convolutional layer, the first... The graph convolutional layer is: , It is an integer, and , The initial node feature matrix, For the first The node feature matrix output by the graph convolutional layer, For the first The node feature matrix output by the graph convolutional layer, It is the ReLU activation function. It is the first The weight matrix of the graph convolutional layer , It is an adjacency matrix. for The elements in, and 1 represents the first The node and the first The nodes are connected in communication, and 0 represents the first The node and the first The nodes in question are not connected in communication. Let be a degree matrix, and the degree matrix be a diagonal matrix. In order to be with the first The number of nodes that the node is communicatively connected to. yes The weight matrix, To connect the first The node and the first The edge feature vector contained in the edges of the nodes. In order to be with the first The set of nodes that are communicatively connected to the node; The final node feature matrix is ​​input into a fully connected layer to obtain a classification feature vector, wherein the fully connected layer is: ,in, For global average pooling, Here is the weight matrix of the fully connected layer. Here is the bias matrix of the fully connected layer. For the first The node feature matrix output by the graph convolutional layer, The classification feature vector; The classification feature vector is input into the output layer to obtain a probability vector, which is composed of all the leakage probabilities. The output layer consists of a Softmax activation function. ,in, Let be the probability vector. The first in the classification feature vector One element, This represents the total number of quadrilateral regions.

5. A hydrogen leak detection device, characterized in that, A hydrogen leak detection system includes: multiple hydrogen sensors; a predetermined area composed of nine identical quadrilateral regions arranged in a 3x3 grid, where the quadrilaterals are squares and the quadrilateral regions are numbered sequentially in a serpentine order; a hydrogen sensor is positioned at a first predetermined distance directly above each vertex of the central quadrilateral region; a hydrogen sensor is positioned at a second predetermined distance directly above each first target point; the first target point is the intersection of any two quadrilateral regions excluding the central quadrilateral region, and the first target point is located on the outer contour of the predetermined area; a hydrogen sensor is positioned at the second predetermined distance directly above the center point of the central quadrilateral region; a hydrogen sensor is positioned at a third predetermined distance directly above each second target point; the second target point is the center point of a quadrilateral region whose only two sides intersect with other quadrilateral regions; a hydrogen sensor is positioned at the third predetermined distance directly above the center point of the central quadrilateral region; the second predetermined distance is greater than the first predetermined distance, and the third predetermined distance is greater than the second predetermined distance. The device includes: The first acquisition unit is used to acquire multiple sets of sensor data in the current time period. Each set of sensor data includes: the three-dimensional coordinates of the hydrogen sensor in a predetermined coordinate system, the pressure value of hydrogen, and the concentration value of hydrogen. Each set of sensor data corresponds to one hydrogen sensor. The pressure value and concentration value of hydrogen in each set of sensor data are acquired by the corresponding hydrogen sensor. A determining unit is configured to determine, based on the sensor data from all groups, whether a hydrogen leak has occurred in the predetermined area; The determining unit includes: a first determining subunit, configured to determine multiple leakage probabilities based on the sensor data of all groups, wherein each leakage probability corresponds one-to-one with a quadrilateral region, and the leakage probability is the probability of hydrogen leakage occurring in the corresponding quadrilateral region; a second determining subunit, configured to determine whether each leakage probability satisfies a predetermined condition, wherein the predetermined condition is that the leakage probability is greater than a predetermined probability; and a third determining subunit, configured to determine that hydrogen leakage has occurred in the predetermined region if at least one leakage probability satisfies the predetermined condition. The first determining subunit includes: a processing module, used to perform normalization processing on each group of sensor data to obtain multiple groups of normalized sensor data, where each group of normalized sensor data corresponds to a group of sensor data; a construction module, used to construct a graph structure, the graph structure consisting of multiple nodes and multiple edges, each node corresponding one-to-one with a hydrogen sensor, and each node representing the corresponding hydrogen sensor, each node including: a node feature vector, the node feature vector including: a group of normalized sensor data corresponding to the node, any two nodes are connected by the edge, the edge including: an edge feature vector, the edge feature vector including: the direction vectors of the two nodes connected by the edge, and the distance between the two nodes connected by the edge; and an input module, used to input the graph structure into a hydrogen leak detection model to obtain multiple leak probabilities, the leak detection model being obtained by training a GCN model using multiple sets of training data, each set of training data including: the graph structure and the leak probabilities acquired in historical time periods.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device containing the computer-readable storage medium to perform the hydrogen leak detection method according to any one of claims 1 to 4.

7. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the hydrogen leak detection method according to any one of claims 1 to 4.

Images