Hydrogen micro-leakage grading quantification method and system for hydrogen storage environment

Abstract

The application discloses a hydrogen micro-leakage grading quantification method and system for a hydrogen storage environment, and relates to the technical field of gas detection.The method comprises the following steps: deploying a hydrogen leakage detection adhesive tape and a fixed thermal imager on the target hydrogen storage environment, obtaining hydrogen-sensitive adhesive tape leakage image data and key hydrogen storage area temperature data; using an HSV color space to extract image brightness to obtain leakage brightness change data and establish a hydrogen leakage degree quantification model; after time series analysis and risk level conversion of the model, combined with temperature data, weighted quantification analysis is carried out, the hydrogen micro-leakage grading quantification result is determined, and early warning decision response is carried out.The application solves the technical problems that in the existing hydrogen storage environment, hydrogen micro-leakage is difficult to be accurately captured, and the degree and risk thereof cannot be quantified, and achieves the technical effects of accurate identification, quantification and time series risk grading of hydrogen micro-leakage, improvement of hydrogen storage environment safety prevention and control capability, and adaptation to different monitoring scenes.

Description

Technical Field

[0001] This invention relates to the field of gas detection technology, and in particular to a method and system for the classification and quantification of hydrogen micro-leakage in hydrogen storage environments. Background Technology

[0002] Physical hydrogen storage is a core component of the hydrogen energy industry chain, and its safe and stable operation is crucial for the large-scale application of hydrogen energy. Monitoring micro-leakage in physical hydrogen storage is key to its safety and control. Current technologies for detecting leaks in physical hydrogen storage largely rely on traditional gas sensors and pressure monitoring, which can be effective in stable hydrogen storage environments. However, with the diversification of physical hydrogen storage scenarios, traditional methods struggle to accurately capture micro-leakage signals and quantify the degree and timing of leaks in complex environments. This results in incomplete and inaccurate data, failing to meet the needs for precise assessment and effective management of physical hydrogen storage environments. Summary of the Invention

[0003] This application provides a method and system for the classification and quantification of hydrogen micro-leakage in hydrogen storage environments, which solves the technical problem that hydrogen micro-leakage in existing hydrogen storage environments is difficult to accurately capture and its degree and risk cannot be quantified.

[0004] The first aspect of this application provides a method for grading and quantifying hydrogen micro-leakage in hydrogen storage environments. The method includes: deploying hydrogen leak detection tape and a fixed thermal imager in a target hydrogen storage environment; acquiring hydrogen leak detection tape leak image data and key hydrogen storage area temperature data using the hydrogen leak detection tape and the fixed thermal imager; extracting brightness values ​​from the hydrogen leak detection tape leak image data using the HSV color space to obtain hydrogen leak brightness variation data; establishing a hydrogen leak severity quantification model based on the hydrogen leak brightness variation data; performing time series analysis and risk level conversion on the hydrogen leak severity quantification model to obtain a hydrogen leak sequence risk level; combining the key hydrogen storage area temperature data to perform weighted quantification analysis of the hydrogen leak sequence risk level to determine the hydrogen micro-leakage grading and quantification result; and using the hydrogen micro-leakage grading and quantification result for early warning decision response.

[0005] A second aspect of this application provides a hydrogen micro-leakage classification and quantification system for hydrogen storage environments. The system includes: an image temperature data acquisition module for deploying hydrogen leak detection tape and a fixed thermal imager in a target hydrogen storage environment, acquiring hydrogen leak detection tape leak image data and key hydrogen storage area temperature data through the hydrogen leak detection tape and the fixed thermal imager; a hydrogen leak degree quantification model construction module for extracting brightness values ​​from the hydrogen leak detection tape leak image data using the HSV color space to obtain hydrogen leak brightness change data, and establishing a hydrogen leak degree quantification model based on the hydrogen leak brightness change data; a hydrogen leak sequence risk level acquisition module for performing time series analysis and risk level conversion on the hydrogen leak degree quantification model to obtain a hydrogen leak sequence risk level; and an early warning decision response execution module for performing weighted quantification analysis of the hydrogen leak sequence risk level based on the key hydrogen storage area temperature data to determine the hydrogen micro-leakage classification and quantification result, and performing an early warning decision response based on the hydrogen micro-leakage classification and quantification result.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: This application deploys hydrogen leak detection tape and a fixed thermal imager in the target hydrogen storage environment to collect image data of hydrogen leaks from the detection tape and temperature data of key hydrogen storage areas. After processing including color space conversion, filtering and noise reduction, edge segmentation, and brightness extraction, data on the brightness changes of the hydrogen leak is obtained. A quantitative model of the leak degree is established, and time-series analysis and risk level conversion are performed. Combined with temperature data for weighted verification, this accurately classifies and quantifies the situation of micro-leaks in hydrogen storage environments. This makes the monitoring results of micro-leaks in hydrogen storage environments more accurate and reliable, supports second-level early warning and emergency decision-making, and achieves the technical effects of accurate identification, quantification, and time-series risk classification of micro-leaks, improving the safety control capabilities of hydrogen storage environments and adapting to different monitoring scenarios. Attached Figure Description

[0007] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0008] Figure 1 This is a schematic flowchart of a hydrogen micro-leakage classification and quantification method for hydrogen storage environments provided in an embodiment of this application.

[0009] Figure 2 This is a schematic diagram of a hydrogen micro-leakage classification and quantification system for hydrogen storage environments provided in this application embodiment.

[0010] Figure labeling: Module 1: Image temperature data acquisition; Module 2: Hydrogen leakage degree quantification model construction; Module 3: Hydrogen leakage sequence risk level acquisition; Module 4: Early warning decision response execution. Detailed Implementation

[0011] This application provides a method and system for the classification and quantification of hydrogen micro-leakage in hydrogen storage environments, which solves the technical problem that hydrogen micro-leakage in existing hydrogen storage environments is difficult to accurately capture and its degree and risk cannot be quantified.

[0012] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0013] It should be noted that the terms "first," "second," etc., in the specification 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 so that the embodiments of this application described herein can be implemented in orders other than those illustrated or 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 server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules not explicitly listed or inherent to such processes, methods, products, or devices.

[0014] Example 1, as Figure 1 As shown, a method for grading and quantifying hydrogen microleakage in hydrogen storage environments is described, wherein the method includes: Deploy hydrogen leak detection tape and a fixed thermal imager in the target hydrogen storage environment to acquire hydrogen leak image data and temperature data of key hydrogen storage areas through the hydrogen leak detection tape and the fixed thermal imager.

[0015] In this embodiment, the hydrogen leak detection tape is a specialized leak detection tape that is highly sensitive, easy to stick to hydrogen-related components such as pipes and valves, and undergoes an irreversible color change upon contact with hydrogen, thus visually indicating the location of hydrogen leaks. It is widely used in hydrogen refueling stations, hydrogen storage tanks, hydrogen energy plants, and other hydrogen-related facilities.

[0016] Specifically, the process begins by acquiring information on the distribution of hydrogen storage areas and the attribute information of hydrogen storage equipment within the target hydrogen storage environment. Next, based on the attribute information of the hydrogen storage equipment, areas with hydrogen leakage risk are identified, forming a set of potential leakage points. Based on this set of points, surface treatment and deployment marking of hydrogen leakage detection tape are then performed. Finally, based on the hydrogen storage area distribution information, temperature monitoring coverage analysis is conducted to determine a set of locations for deploying thermal imagers. Fixed thermal imagers are then deployed according to this set of locations. This step will be explained in detail later.

[0017] Next, hydrogen leak detection tape has been deployed in the target hydrogen storage environment according to the identified risk points for hydrogen leaks. When a hydrogen leak occurs in the target hydrogen storage environment, the tape will visually change color or brightness upon contact with hydrogen. The deployed fixed thermal imager will continuously capture images of the areas where the hydrogen leak detection tape is deployed, based on its image acquisition function. By capturing the visual changes in the tape caused by the hydrogen leak, it generates and outputs hydrogen leak detection tape image data. In addition, the fixed thermal imager also has a temperature monitoring function. Its deployment location, after coverage analysis, can cover key hydrogen storage areas. These key hydrogen storage areas are those where the core equipment is located, with clearly defined hydrogen storage area distribution information. During the acquisition of tape images, the ambient temperature of the key hydrogen storage areas is simultaneously monitored in real time, converting the temperature signal within the key hydrogen storage areas into a data signal, and then generating and outputting the temperature data for the key hydrogen storage areas.

[0018] The brightness values ​​of the hydrogen-sensitive tape leakage image data were extracted using the HSV color space to obtain hydrogen leakage brightness change data. Based on the hydrogen leakage brightness change data, a quantitative model of the degree of hydrogen leakage was established.

[0019] In this embodiment of the application, the HSV color space is a color model that defines and describes color according to three dimensions: hue, saturation, and value. It is often used in scenarios such as image color analysis and feature extraction, for example, to accurately extract brightness-related data from an image to reflect changes in the target.

[0020] Optionally, the hydrogen-sensitive tape leak image data is first read and converted to RGB color space to obtain an RGB color image of the hydrogen-sensitive tape leak. Then, a Gaussian filter is used to filter and denoise this RGB color image to obtain a usable RGB color image of the hydrogen-sensitive tape leak. Subsequently, edge detection and background segmentation are performed on the usable RGB color image of the hydrogen-sensitive tape leak to separate the foreground region of the hydrogen-sensitive tape leak. Finally, the foreground region is converted to HSV color space, and brightness values ​​are extracted to obtain the final hydrogen leak brightness variation data.

[0021] Next, a sample set of hydrogen-sensitive tape leak images was collected, and the sample set was used to train a hydrogen leakage assessment model. Then, the hydrogen leakage brightness change data was input into the assessment model, and the leakage assessment output the hydrogen leakage degree change level. The leakage degree change level was quantitatively fitted according to the time series, and finally a quantitative model of hydrogen leakage degree was established.

[0022] The hydrogen leakage degree quantification model is subjected to time series analysis and risk level conversion to obtain the hydrogen leakage sequence risk level.

[0023] In one embodiment of this application, a time series trend analysis is first performed on the hydrogen leakage level quantification model to generate a hydrogen leakage level fluctuation trend chart. Then, based on the safety management requirements of the hydrogen storage environment and the needs of hydrogen storage applications, a hydrogen leakage risk classification system is constructed. Finally, based on this hydrogen leakage risk classification system, the hydrogen leakage level fluctuation trend chart is transformed into a risk level, ultimately obtaining the hydrogen leakage sequence risk level.

[0024] By combining the temperature data of the key hydrogen storage area, a weighted quantitative analysis of the risk level of the hydrogen leakage sequence is performed to determine the grading and quantification results of hydrogen micro-leakage, and early warning decision-making response is made based on the grading and quantification results of hydrogen micro-leakage.

[0025] Specifically, the temperature data of key hydrogen storage areas is first aligned with the risk levels of hydrogen leakage sequences over time to obtain key hydrogen storage area sequence temperature data. Then, based on this key hydrogen storage area sequence temperature data, a verification weighted quantitative analysis is conducted on the risk levels of hydrogen leakage sequences to ultimately determine the quantitative results of hydrogen micro-leakage classification. This step will be explained in detail later.

[0026] Next, a system corresponding to early warning levels and response measures will be established. Referring to the requirements for hydrogen storage environment safety management and emergency response standards, different response measures will be assigned to the risk levels of the hydrogen micro-leakage classification and quantification results. For example, low risk corresponds to continuous data monitoring, medium risk to on-site personnel inspection, high risk to activation of local ventilation equipment, and extremely high risk to emergency shutdown and evacuation of personnel. Then, the hydrogen micro-leakage classification and quantification results will be acquired in real time through a data acquisition system. The risk level of the quantification results at each time point will be determined to confirm whether the early warning trigger conditions have been met. If so, the response process will commence.

[0027] Subsequently, based on the determined warning level, instructions are automatically sent to the corresponding execution system. For example, when a medium-risk warning is triggered, an inspection notice is sent to the on-site management terminal; when an extremely high-risk warning is triggered, a shutdown signal is sent to the hydrogen storage equipment control system, and an evacuation order is sent to the emergency command system. Finally, the subsequent hydrogen micro-leakage classification and quantification results are continuously monitored to determine whether the warning response measures are effective. If the quantification results show that the risk level has decreased, the current measures are maintained; if the risk level has not decreased or has increased, the response measures are upgraded until the risk level drops to a safe range.

[0028] Furthermore, the method provided in this application embodiment includes: The process involves acquiring hydrogen storage area distribution information and hydrogen storage equipment attribute information for the target hydrogen storage environment; identifying hydrogen leakage risk areas based on the hydrogen storage equipment attribute information to obtain a set of hydrogen storage equipment leakage risk points; surface-treating and marking the hydrogen leakage detection tape on the target hydrogen storage environment according to the set of hydrogen storage equipment leakage risk points; performing temperature monitoring coverage analysis based on the hydrogen storage area distribution information to determine the set of thermal imager deployment locations, and deploying fixed thermal imagers on the target hydrogen storage environment according to the set of thermal imager deployment locations.

[0029] Specifically, those skilled in the art first conduct on-site mapping of the target hydrogen storage environment, recording the spatial layout of hydrogen storage facilities such as hydrogen storage tanks, hydrogen pipelines, and valve assemblies to form hydrogen storage area distribution information. Next, they extract attribute information of the hydrogen storage equipment, such as material parameters, design pressure, service life, sealing structure type, and historical maintenance records, from the hydrogen storage equipment management system to ensure that subsequent risk identification and equipment deployment have complete data support.

[0030] Next, based on the attribute information of hydrogen storage equipment, a list of typical leakage risk locations for hydrogen storage equipment commonly used in the industry is retrieved. This list clearly identifies the high-frequency leakage locations and corresponding attribute association conditions for different types of hydrogen storage equipment such as hydrogen storage tanks, hydrogen pipelines, and shut-off valves. Then, the attribute information of each device in the target hydrogen storage environment, such as equipment type, service life, sealing structure form, and historical maintenance records, is matched with the association conditions in the list. For example, hydrogen pipelines with a service life exceeding 60% of their design life are matched with the risk item of pipeline welding in the list, and valves with seal replacements in the past year in the historical maintenance records are matched with the risk item of valve sealing rings in the list. Subsequently, only the successfully matched equipment locations are selected and these locations are directly integrated to form a set of leakage risk points for hydrogen storage equipment.

[0031] Next, the surface of the hydrogen storage equipment is treated, specifically by solvent cleaning and sanding. First, solvents such as industrial ethanol are used to clean the surface of the hydrogen storage equipment at risk points, removing oil, dust, and other impurities. Then, fine-grit sandpaper is used to lightly sand the surface of the hydrogen storage equipment to improve the adhesion between the tape and the equipment surface. After completing the surface treatment of the hydrogen storage equipment, hydrogen leak detection tape is precisely affixed to each risk point, and physical markings are used for deployment and marking, such as attaching fluorescent markers around the tape to facilitate accurate image capture by the thermal imager.

[0032] Subsequently, the target hydrogen storage area was divided into several spatial grids according to a preset size, and the coordinates and coverage of each grid were defined. The preset size for grid division needs to be determined based on the distribution information of hydrogen storage areas and the attribute information of hydrogen storage equipment in the target hydrogen storage environment. The core basis includes three aspects: First, the physical size characteristics of hydrogen storage equipment, such as the diameter of hydrogen storage tanks, the laying spacing of hydrogen pipelines, and the installation size of valve groups, to ensure that a single grid can completely cover at least one key equipment unit or equipment prone to leakage, avoiding the fragmentation of key equipment areas and insufficient monitoring accuracy due to excessively large grids, or the increase of data processing redundancy due to excessively small grids; Second, the monitoring radius parameters of fixed thermal imagers. The preset size needs to match the effective monitoring range of thermal imagers, and the maximum side length of a single grid should not exceed 1 / 2 of the monitoring radius of the thermal imager, to ensure that the thermal imager can clearly capture the tape images and temperature data within the grid at potential deployment points; Third, the overall spatial layout of the hydrogen storage area. If the equipment in the hydrogen storage area is dense, the preset size needs to be reduced accordingly to distinguish the monitoring areas of different equipment. If the equipment in the area is scattered, the size can be appropriately increased to balance monitoring accuracy and deployment efficiency. Finally, a unified preset size is determined by comprehensively considering the above three aspects, and the spatial grid is divided accordingly.

[0033] Then, using the equipment locations in the hydrogen storage area distribution information as the core reference, potential deployment points are initially screened within the target hydrogen storage area. These potential deployment points must avoid the hydrogen storage equipment itself and other obstructions, such as support columns and pipeline supports. Priority is given to higher platforms, walls, or dedicated support structures within the area, forming a list of potential deployment points. Next, for each potential deployment point in this list, a theoretical monitoring circle is established with that point as the center and the thermal imager's monitoring radius as the radius. This theoretical monitoring circle is then overlaid with the pre-divided spatial grid to determine if each spatial grid falls completely within the theoretical monitoring circle. For grids falling within the theoretical monitoring circle, further verification is conducted to check whether there are any hydrogen storage devices or other obstructions between the grid and the potential deployment point. If there are no obstructions and the line connecting the grid's center point and the potential deployment point does not pass through any obstructions, the grid is determined to be a valid coverage grid; otherwise, it is determined to be a blind zone grid. Finally, the number of effective coverage grids for each potential deployment point is counted, and potential deployment points that can make all grids in the key hydrogen storage area effective coverage grids are selected. These points are integrated to form a set of thermal imager deployment locations, where the key hydrogen storage area is the area where the core equipment is located, based on the clear distribution information of the hydrogen storage area.

[0034] Finally, according to the deployment location set of the thermal imager, the fixed thermal imager was installed in place using a bracket to ensure that it can not only collect temperature data of key hydrogen storage areas, but also clearly capture images of hydrogen leak detection tape.

[0035] Furthermore, the method provided in this application embodiment includes: The hydrogen-sensitive tape leakage image data is read and converted to RGB color space to obtain a hydrogen-sensitive tape leakage RGB color image; a Gaussian filter is used to filter and denoise the hydrogen-sensitive tape leakage RGB color image to obtain a usable hydrogen-sensitive tape leakage RGB color image; edge detection and background segmentation are performed on the usable hydrogen-sensitive tape leakage RGB color image to obtain the hydrogen-sensitive tape leakage foreground region; the HSV color space is used to convert the color space of the hydrogen-sensitive tape leakage foreground region and extract the brightness value to obtain the hydrogen leakage brightness change data.

[0036] Optionally, first retrieve the hydrogen-sensitive tape leakage image data acquired by the fixed thermal imager, and use RGB color space conversion to convert the original image data format from the default grayscale format of the thermal imager to RGB color mode. By calculating the intensity value of each pixel in the three color channels of red (R), green (G), and blue (B), the final RGB color image of the hydrogen-sensitive tape leakage is obtained.

[0037] Next, when using a Gaussian filter to filter and denoise the RGB color image of the hydrogen-sensitive tape leak, random noise will be present in the image due to factors such as light fluctuations and equipment vibrations in the hydrogen storage environment. The Gaussian filter is based on the weighted average filtering principle. A 3×3 or 5×5 filtering window is constructed with each pixel in the image as the center. The weight value of each pixel in the window is calculated according to the Gaussian function. The weight decreases as the distance from the center pixel increases. Then, the intensity value of the RGB channel of each pixel in the window is multiplied by the corresponding weight value and summed to obtain the intensity value of the RGB channel of the center pixel after denoising. The denoising process is completed by traversing the entire image to obtain a usable RGB color image of the hydrogen-sensitive tape leak.

[0038] Next, the Canny edge detection algorithm is used to extract the edges of the hydrogen leak detection tape. First, the RGB color image of the usable hydrogen leak detection tape is converted into a grayscale image. By calculating the grayscale gradient of the grayscale image, non-maximum suppression is used to eliminate non-maximum pixels in the edge response. Strong edges and weakly connected edges are filtered through double thresholding, ultimately forming a clear edge contour of the hydrogen leak detection tape. Then, based on the edge contour, a threshold segmentation algorithm is applied. A segmentation threshold is set according to the difference in grayscale values ​​between the foreground hydrogen leak detection tape and the background hydrogen storage equipment and target hydrogen storage environment. Areas with grayscale values ​​below the threshold are marked as background and removed, while areas with grayscale values ​​above the threshold are retained, thus obtaining the foreground area of ​​the hydrogen leak detection tape.

[0039] Finally, the HSV color space was used to perform color space conversion and extract brightness values ​​for the foreground area of ​​the hydrogen-sensitive tape leak. To accurately separate brightness information, the RGB color mode of the foreground area was converted to the HSV color mode, where HSV corresponds to hue, saturation, and lightness, respectively. This conversion was performed by calculating the hue, saturation, and lightness parameters of each pixel. Subsequently, the lightness V channel values ​​of all pixels in the foreground area were read individually in HSV mode. The magnitude of the lightness value directly reflects the brightness of the tape. These lightness values ​​were organized into a data set according to time series or pixel position, ultimately yielding the brightness variation data of the hydrogen leak.

[0040] Furthermore, the method provided in this application embodiment includes: A sample set of hydrogen-sensitive tape leak images is collected, and the leak level assessment model is trained on the sample set to generate a hydrogen leak level assessment model. The hydrogen leak level assessment model is used to assess the hydrogen leak brightness change data and output the hydrogen leak level change level. The hydrogen leak level change level is then quantitatively fitted according to the time series to establish the hydrogen leak level quantization model.

[0041] Specifically, firstly, the brightness values ​​of the hydrogen-sensitive tape leakage image sample set are extracted using the HSV color space to obtain a hydrogen-sensitive tape leakage brightness sample set. Next, the leakage degree of this brightness sample set is graded and labeled to obtain a hydrogen-sensitive tape leakage degree sample set. Finally, a deep neural network is used to train the hydrogen-sensitive tape leakage degree sample set for leakage degree assessment, ultimately generating a hydrogen leakage degree assessment model.

[0042] Next, the hydrogen leakage level assessment and grade output are performed. The generated hydrogen leakage level assessment model is deployed as a real-time inference service using the TensorFlowServing model deployment tool, ensuring the model can efficiently receive input data and output results. For the hydrogen leakage brightness change data obtained in the previous steps, preprocessing is performed using the Min-Max normalization method. This method calculates (brightness data value - minimum brightness data value) / (maximum brightness data value - minimum brightness data value) to normalize the brightness data to the 0-1 range, eliminating the interference of differences in brightness value ranges on the model assessment. After preprocessing, the normalized brightness data is input as an array into the deployed hydrogen leakage level assessment model inference service. The model performs a forward propagation operation to analyze the leakage level probability of the input brightness data, outputting the probability values ​​corresponding to Level 0 (no leakage), Level 1 (minor leakage), Level 2 (light leakage), Level 3 (moderate leakage), and Level 4 (severe leakage). Finally, the level with the highest probability is selected as the current hydrogen leakage level change grade, completing the leakage assessment and grade output.

[0043] Next, the scikit-learn library in Python was used to perform time-series quantitative fitting of the hydrogen leakage level change. First, historical data was organized, using second-level timestamps as the time axis, and the hydrogen leakage level change corresponding to different collection times was arranged in chronological order, forming a two-dimensional "time-level" time-series dataset. Then, a second-order polynomial fitting method was chosen. The time features were converted into second-order polynomial features using the PolynomialFeatures module in the scikit-learn library. Next, the LinearRegression module was used to train linear regression on the "polynomial time feature-leakage level" dataset, and the fitting equation was obtained, as shown below. Let y be the leakage level, t be the time, and a, b, and c be the fitting coefficients. This fitting equation is the hydrogen leakage level quantification model, which can predict the hydrogen leakage level at a corresponding moment based on the input time parameter, and reflect the change of leakage level over time through the trend of the equation.

[0044] By implementing real-time model inference, ensuring data consistency through Min-Max standardization, and constructing time series correlations through polynomial fitting, discrete hydrogen leakage brightness data is ultimately transformed into a quantitative model of hydrogen leakage degree that can describe the change of leakage degree over time. This provides accurate quantitative basis for subsequent leakage risk time series analysis and early warning decision-making.

[0045] Furthermore, the method provided in this application embodiment includes: The brightness values ​​of the hydrogen-sensitive tape leakage image sample set are extracted using the HSV color space to obtain a hydrogen-sensitive tape leakage brightness sample set; the leakage degree of the hydrogen-sensitive tape leakage brightness sample set is classified and labeled to obtain a hydrogen-sensitive tape leakage degree sample set; a deep neural network is used to train the hydrogen-sensitive tape leakage degree sample set for leakage degree assessment to generate the hydrogen leakage degree assessment model.

[0046] Specifically, the process begins by reading each image of hydrogen-sensitive tape leakage from the sample set. These images are typically stored in RGB format. The `cvtColor` function from the OpenCV image processing library is then used to convert the RGB images to HSV color space format. The conversion follows industry-standard color space conversion formulas, decomposing the image's color information into three independent channels: hue, saturation, and lightness. Since the lightness (V) channel value directly corresponds to the image's brightness information, only the lightness (V) channel values ​​of all pixels in each image are extracted. These values ​​are then organized into a two-dimensional data matrix according to the image number and pixel position. Rows represent image samples, and columns represent the brightness values ​​of the corresponding pixels, ultimately forming a lightness sample set for hydrogen-sensitive tape leakage.

[0047] Next, the leakage degree of the hydrogen-sensitive tape leakage brightness sample set was graded and labeled. First, historical hydrogen leakage experimental data was collected, and the brightness value range of the hydrogen-sensitive tape under different leakage amounts was statistically analyzed. For example, a brightness value in the range of 0-50 corresponds to no leakage (Level 0), 51-100 corresponds to minor leakage (Level 1), 101-150 corresponds to slight leakage (Level 2), 151-200 corresponds to moderate leakage (Level 3), and 201-255 corresponds to severe leakage (Level 4), forming a clear grading threshold standard. Then, 3-5 experts in the field of hydrogen storage safety were invited to independently label each sample in the brightness sample set according to this threshold standard. If there were differences in the experts' labeling results, for example, two experts labeled a certain image sample as Level 1 and three experts labeled it as Level 2, the labeling result of the majority of experts was taken as the final level. Then, a corresponding leakage degree level label was added to each brightness sample, for example, represented by numbers 0-4. After processing, the hydrogen-sensitive tape leakage degree sample set was obtained.

[0048] Finally, a deep neural network was used to train the model. The lightweight convolutional neural network MobileNet was chosen as the base model due to its low parameter count, high training efficiency, and suitability for assessing leakage levels in image-based scenarios with small sample sizes. During model construction, the input layer was set to a 224×224×1 single-channel feature map, corresponding to the grayscale image format of the brightness samples, converting the brightness data matrix to this image size. The hidden layer contains six convolutional blocks, each consisting of a 3×3 convolutional layer, a batch normalization layer, a ReLU activation function, and three max-pooling layers with a stride of 2 to reduce the feature map dimension. The output layer has five neurons, corresponding to the 0-4 leakage levels mentioned above, and uses the Softmax activation function to output the probability values ​​for each level.

[0049] During training, the sample set of hydrogen-sensitive tape leakage levels was divided into training and validation sets in a 7:3 ratio. The error between the model's predicted level and the true label was calculated using the cross-entropy loss function. The model parameters were iteratively updated using an Adam optimizer with a learning rate of 0.001. After each iteration, the model accuracy was evaluated using the validation set. Training was stopped when the validation set accuracy stabilized above 95% for five consecutive iterations, and the final model file was saved, thus generating the hydrogen leakage level assessment model. The input to this model is the brightness data of the hydrogen-sensitive tape, i.e., a 224×224×1 single-channel feature map. The output is the probability value of each of the five leakage level grades (0-4). The leakage level with the highest probability is taken as the final leakage level assessment result.

[0050] By accurately extracting brightness features using the HSV color space, combining expert annotation with thresholding to achieve sample grading, and efficiently training the MobileNet neural network, a hydrogen leakage assessment model that can stably input brightness data and output accurate leakage level is finally constructed, providing a reliable assessment tool for subsequent quantification of hydrogen leakage levels.

[0051] Furthermore, the method provided in this application embodiment includes: A time series trend analysis is performed on the hydrogen leakage degree quantification model to generate a hydrogen leakage level fluctuation trend map; a hydrogen leakage risk classification system is constructed based on hydrogen storage environment safety management and hydrogen storage application requirements; and the hydrogen leakage level fluctuation trend map is converted into a risk level based on the hydrogen leakage risk classification system to obtain the hydrogen leakage sequence risk level.

[0052] Specifically, the process begins with a time-series trend analysis of the hydrogen leakage level quantification model. Leakage level data at continuous time points is extracted from the model, using a second-level data collection interval as an example. This yields the leakage level every 10 seconds over the past 24 hours, forming a raw data sequence of "timestamp-leakage level". Subsequently, a three-point simple moving average method is used to smooth the raw data sequence. This involves calculating the average of each data point with its preceding and following data points, replacing the original leakage level with the average of the previous data point, thus eliminating the interference of short-term random fluctuations on the trend. After processing, a trend chart is plotted, with the horizontal axis representing the timestamp accurate to the minute and the vertical axis representing the smoothed leakage level. This line chart visually presents the fluctuation pattern of the leakage level over time, generating a trend chart of hydrogen leakage level fluctuations.

[0053] Next, combining industry safety standards, such as the "Safety Requirements for Hydrogen Storage Systems," and historical leakage accident case data, the classification rules are determined. First, the number of risk levels is clearly defined, typically divided into four levels: low risk, medium risk, high risk, and extremely high risk. Then, based on the impact of the leakage level on the hydrogen storage equipment and the priority of emergency response, the leakage level range and duration threshold corresponding to each risk level are set: for example, low risk corresponds to leakage level 0-1 with a duration of <24 hours; medium risk corresponds to leakage level 1-2 with a duration of 24-72 hours; high risk corresponds to leakage level 2-3 with a duration of >72 hours; and extremely high risk corresponds to leakage level 3-4 without considering duration. Finally, the risk levels, corresponding leakage level ranges, duration thresholds, and recommended emergency measures, such as monitoring only for low risk and immediate shutdown for extremely high risk, are compiled into a structured table, forming a complete hydrogen leakage risk classification system.

[0054] Finally, the smoothed data of the hydrogen leak level fluctuation trend chart was exported as a "Time-Smoothed Leakage Level" data format, and a "Risk Level" column was added. Then, conditional judgment logic was written to match the smoothed leakage level at each time point with the leakage level range and duration threshold in the hydrogen leak risk classification system. For example, if the smoothed leakage level at a certain time point is level 2 and this level has lasted for 36 hours, then it is considered a medium risk match. After completing the risk level matching for all time points, the "Time-Risk Level" data was arranged in chronological order to obtain the dynamically changing hydrogen leak sequence risk level over time.

[0055] By analyzing time-series trends, constructing a risk grading system by combining industry standards and case studies, and completing the level conversion through data matching, the output of the hydrogen leakage degree quantification model is finally transformed into a time-series hydrogen leakage sequence risk level, providing a clear risk time-series basis for subsequent weighted analysis and emergency early warning decision-making based on temperature data.

[0056] Furthermore, the method provided in this application embodiment includes: The temperature data of the key hydrogen storage area is aligned with the risk level of the hydrogen leakage sequence by time series to obtain the temperature data of the key hydrogen storage area sequence; the risk level of the hydrogen leakage sequence is verified and weighted by combining the temperature data of the key hydrogen storage area sequence to determine the quantification result of the hydrogen micro-leakage classification.

[0057] In one embodiment, the timestamp information of the key hydrogen storage area temperature data and the hydrogen leak sequence risk level is first extracted. The key hydrogen storage area temperature data is generated by a fixed thermal imager at preset acquisition intervals, and each temperature data point corresponds to a unique acquisition timestamp. The hydrogen leak sequence risk level is generated by the time series analysis in the aforementioned steps, and each risk level also corresponds to a unique timestamp. Next, a unified time granularity is determined, such as using 1 second as the smallest time unit, and the timestamps of the two types of data are standardized. If there is an inconsistency in timestamp precision, such as temperature data being accurate to milliseconds and risk level data being accurate to seconds, the millisecond-level timestamp of the temperature data is rounded down to the second level to ensure that the time dimension units of the two are consistent.

[0058] Then, iterate through all timestamps of the hydrogen leak sequence risk levels, filtering out the temperature values ​​corresponding to the same timestamps from the temperature data of key hydrogen storage areas. If a certain risk level timestamp does not have a direct match in the temperature data, take the two adjacent temperature data points before and after that timestamp and complete the corresponding temperature data through linear interpolation, that is, calculate the temperature value of the missing time point based on the time difference and temperature difference between the two adjacent points. Finally, organize the matched or completed temperature data, along with the corresponding timestamps and hydrogen leak sequence risk levels, into structured data to form key hydrogen storage area sequence temperature data, ensuring that each hydrogen leak sequence risk level has synchronized temperature data to support subsequent verification and weighted quantification.

[0059] Next, based on historical hydrogen storage leakage temperature datasets, a regional temperature-hydrogen leakage level model is constructed. Then, this model is used to analyze the leakage level and convert the risk level of key hydrogen storage area temperature sequences, obtaining the risk level of the hydrogen sequence in the key areas. Finally, based on the risk level of the hydrogen sequence in the key areas, the risk level of the hydrogen leakage sequence is validated and weighted, ultimately determining the quantification result of the hydrogen micro-leakage classification. This step will be explained in detail later.

[0060] Furthermore, the method provided in this application embodiment includes: Based on historical hydrogen storage leakage temperature datasets, a regional temperature-hydrogen leakage degree model is constructed. This model is then used to analyze the leakage degree and convert the risk level of the key hydrogen storage region's sequential temperature data, obtaining the risk level of the key region's hydrogen sequence. Based on the risk level of the key region's hydrogen sequence, the risk level of the hydrogen leakage sequence is verified and weighted to determine the quantification result of the hydrogen micro-leakage classification.

[0061] Optionally, a historical hydrogen storage leakage temperature dataset is first collected. This dataset contains temperature values ​​of key hydrogen storage areas at different time points in the past hydrogen storage environment, along with corresponding data on the degree of hydrogen leakage detected manually or recorded by equipment at those time points. The dataset is then processed using Z-score normalization. By calculating (individual temperature value - dataset mean temperature) / dataset temperature standard deviation, the temperature data is transformed into standardized data with a mean of 0 and a standard deviation of 1, thus eliminating the influence of differences in temperature measurement ranges between different monitoring devices. Simultaneously, box plots are used to identify outliers. Temperature values ​​and leakage degree values ​​exceeding 1.5 times the interquartile range are identified as outliers and removed to obtain a clean historical analysis dataset.

[0062] Furthermore, the interquartile range (IQR) in the above steps is an indicator used to describe the dispersion of data. It is calculated by arranging the data in ascending order and then subtracting the value at the 25th percentile (Q1) from the value at the 75th percentile (the third quartile). In the scenario of processing historical hydrogen storage leakage temperature datasets, by calculating the IQR of both temperature and leakage severity values, the range for outlier determination can be established: the lower limit is Q1 minus 1.5 times the IQR, and the upper limit is Q3 plus 1.5 times the IQR. When a temperature or leakage severity value exceeds this range, it is considered an outlier and is subsequently removed to obtain a clean historical analysis dataset.

[0063] Next, the cleaned historical analysis dataset is divided into feature variables and target variables. The feature variables are standardized temperature values, and the target variables are the corresponding hydrogen leakage levels. These are split into training and test sets in an 8:2 ratio. A linear regression model is constructed using the `LinearRegression` module from Python's `scikit-learn` library, similar to the previous steps. The model takes the feature variables from the training set as input and the target variables as output, and fits the linear relationship between temperature and leakage levels using the least squares method. After training, the model performance is validated using test set data, and the coefficient of determination between the model's predicted values ​​and the actual leakage levels on the test set is calculated. ,when The model was determined to fit the requirements, and the linear relationship equation was finally identified as the regional temperature-hydrogen leakage degree model.

[0064] Next, the sequential temperature data of the key hydrogen storage area is processed. First, the same Z-score standardization parameters as in the previous steps—namely, the mean and standard deviation of the historical dataset—are used to standardize the sequential temperature data, ensuring the data format is consistent with the input requirements of the regional temperature-hydrogen leakage level model. The standardized sequential temperature data is then input into the regional temperature-hydrogen leakage level model point by point in chronological order. The model calculates the predicted hydrogen leakage level for each time point using a pre-defined linear equation. Referring to the hydrogen leakage risk classification system constructed in the previous steps, this system establishes a correspondence between leakage level and risk level based on the safety requirements of the hydrogen storage environment. For example, leakage level 0-1 corresponds to low risk, 1-2 to medium risk, 2-3 to high risk, and 3-4 to extremely high risk. The predicted leakage level for each time point is matched with the intervals in the risk classification system to determine the risk level corresponding to each time point. These risk levels are then organized chronologically to form the hydrogen sequence risk level for the key area.

[0065] Finally, the consistency of the hydrogen leak sequence risk level is verified based on the hydrogen sequence risk level in the key area, and the consistency result of the hydrogen leak sequence is obtained. Subsequently, according to the consistency result of the hydrogen leak sequence, the risk level of the hydrogen sequence in the key area and the risk level of the hydrogen leak sequence are weighted and quantified to finally determine the quantification result of the hydrogen micro-leakage classification. This step will be explained in detail in the following content.

[0066] By standardizing the data format using Z-score, constructing a correlation model between temperature and leakage degree using linear regression algorithm, and completing the conversion from leakage degree to risk level using a pre-set risk classification system, an accurate temperature-dimensional risk reference is provided for verifying and weighting the risk level of hydrogen leakage sequence based on the risk level of hydrogen sequence in key areas. This ensures that the subsequent hydrogen micro-leakage classification and quantification results are more consistent with the actual safety status of the hydrogen storage environment.

[0067] Furthermore, the method provided in this application embodiment includes: Based on the risk level of the hydrogen sequence in the key area, the consistency of the risk level of the hydrogen leakage sequence is verified to obtain the consistency result of the hydrogen leakage sequence; according to the consistency result of the hydrogen leakage sequence, the risk level of the hydrogen sequence in the key area and the risk level of the hydrogen leakage sequence are weighted and quantified to determine the grading and quantification result of the hydrogen micro-leakage.

[0068] In one embodiment, consistency verification is first performed. Both the risk levels of the hydrogen gas sequence and the risk levels of the hydrogen leak sequence are risk data arranged chronologically, and verification is achieved using a point-by-point comparison method. First, a comparison standard is set: the same risk level indicates complete consistency; a difference of one level indicates partial consistency; and a difference of two levels or more indicates inconsistency. Next, the risk levels of all corresponding time points in both sequences are traversed, and the number of time points with complete consistency, partial consistency, and inconsistency is counted. The proportion of completely consistent time points to the total number of time points is calculated, and this proportion represents the consistency result of the hydrogen leak sequence.

[0069] Then, weighted quantification calculations are performed, and weighting rules are determined based on the consistency results of the hydrogen leak sequence. If the complete consistency rate is ≥80%, the weights of the risk levels of both the key area hydrogen sequence and the hydrogen leak sequence are set to 0.5; if 50% ≤ complete consistency rate < 80%, the weight of the risk level of the key area hydrogen sequence is set to 0.6, and the risk level of the hydrogen leak sequence is set to 0.4; if the complete consistency rate is <50%, the weight of the risk level of the key area hydrogen sequence is set to 0.7, and the risk level of the hydrogen leak sequence is set to 0.3. For each time point, the two risk levels are multiplied by their corresponding weights and summed to obtain the weighted risk value for that time point. All weighted risk values ​​are then organized by time to determine the quantification result of the hydrogen micro-leakage classification.

[0070] In summary, the hydrogen micro-leakage classification and quantification method for hydrogen storage environments provided in this application has the following technical effects: This application deploys hydrogen leak detection tape and a fixed thermal imager in the target hydrogen storage environment to acquire images of hydrogen leaks and temperature data of key areas. Brightness is enhanced using the HSV color space, a leak severity quantification model is constructed, and time-series analysis and temperature-weighted verification are combined to determine the grading and quantification results of hydrogen micro-leaks. This achieves accurate identification and second-level early warning of leaks in the hydrogen storage environment, enhancing safety control capabilities. The application demonstrates the technical effectiveness of accurately identifying, quantifying, and grading time-series risks of hydrogen micro-leaks, improving the safety control capabilities of the hydrogen storage environment, and adapting to different monitoring scenarios.

[0071] Example 2, as Figure 2 As shown, based on the same inventive concept as in Embodiment 1 above, this application provides a hydrogen micro-leakage classification and quantification system for hydrogen storage environments, the system comprising: Image temperature data acquisition module 1 is used to deploy hydrogen leak detection tape and a fixed thermal imager in the target hydrogen storage environment, and to acquire hydrogen leak image data and temperature data of key hydrogen storage areas through the hydrogen leak detection tape and the fixed thermal imager.

[0072] The hydrogen leakage degree quantification model construction module 2 is used to extract the brightness value of the hydrogen-sensitive tape leakage image data using the HSV color space to obtain hydrogen leakage brightness change data, and establish a hydrogen leakage degree quantification model based on the hydrogen leakage brightness change data.

[0073] The hydrogen leakage sequence risk level acquisition module 3 is used to perform time series analysis and risk level conversion on the hydrogen leakage degree quantification model to obtain the hydrogen leakage sequence risk level.

[0074] The early warning decision response execution module 4 is used to perform weighted quantitative analysis of the risk level of the hydrogen leakage sequence by combining the temperature data of the key hydrogen storage area, determine the hydrogen micro-leakage classification and quantification result, and perform early warning decision response based on the hydrogen micro-leakage classification and quantification result.

[0075] Furthermore, the image temperature data acquisition module 1 is used to perform the following steps: The process involves acquiring hydrogen storage area distribution information and hydrogen storage equipment attribute information for the target hydrogen storage environment; identifying hydrogen leakage risk areas based on the hydrogen storage equipment attribute information to obtain a set of hydrogen storage equipment leakage risk points; surface-treating and marking the hydrogen leakage detection tape on the target hydrogen storage environment according to the set of hydrogen storage equipment leakage risk points; performing temperature monitoring coverage analysis based on the hydrogen storage area distribution information to determine the set of thermal imager deployment locations, and deploying fixed thermal imagers on the target hydrogen storage environment according to the set of thermal imager deployment locations.

[0076] Furthermore, the hydrogen leakage degree quantification model construction module 2 is used to perform the following steps: The hydrogen-sensitive tape leakage image data is read and converted to RGB color space to obtain a hydrogen-sensitive tape leakage RGB color image; a Gaussian filter is used to filter and denoise the hydrogen-sensitive tape leakage RGB color image to obtain a usable hydrogen-sensitive tape leakage RGB color image; edge detection and background segmentation are performed on the usable hydrogen-sensitive tape leakage RGB color image to obtain the hydrogen-sensitive tape leakage foreground region; the HSV color space is used to convert the color space of the hydrogen-sensitive tape leakage foreground region and extract the brightness value to obtain the hydrogen leakage brightness change data.

[0077] Furthermore, the hydrogen leakage degree quantification model construction module 2 is used to perform the following steps: A sample set of hydrogen-sensitive tape leak images is collected, and the leak level assessment model is trained on the sample set to generate a hydrogen leak level assessment model. The hydrogen leak level assessment model is used to assess the hydrogen leak brightness change data and output the hydrogen leak level change level. The hydrogen leak level change level is then quantitatively fitted according to the time series to establish the hydrogen leak level quantization model.

[0078] Furthermore, the hydrogen leakage degree quantification model construction module 2 is used to perform the following steps: The brightness values ​​of the hydrogen-sensitive tape leakage image sample set are extracted using the HSV color space to obtain a hydrogen-sensitive tape leakage brightness sample set; the leakage degree of the hydrogen-sensitive tape leakage brightness sample set is classified and labeled to obtain a hydrogen-sensitive tape leakage degree sample set; a deep neural network is used to train the hydrogen-sensitive tape leakage degree sample set for leakage degree assessment to generate the hydrogen leakage degree assessment model.

[0079] Furthermore, the hydrogen leakage sequence risk level acquisition module 3 is used to perform the following steps: A time series trend analysis is performed on the hydrogen leakage degree quantification model to generate a hydrogen leakage level fluctuation trend map; a hydrogen leakage risk classification system is constructed based on hydrogen storage environment safety management and hydrogen storage application requirements; and the hydrogen leakage level fluctuation trend map is converted into a risk level based on the hydrogen leakage risk classification system to obtain the hydrogen leakage sequence risk level.

[0080] Furthermore, the early warning decision response execution module 4 is used to perform the following steps: The temperature data of the key hydrogen storage area is aligned with the risk level of the hydrogen leakage sequence by time series to obtain the temperature data of the key hydrogen storage area sequence; the risk level of the hydrogen leakage sequence is verified and weighted by combining the temperature data of the key hydrogen storage area sequence to determine the quantification result of the hydrogen micro-leakage classification.

[0081] Furthermore, the early warning decision response execution module 4 is used to perform the following steps: Based on historical hydrogen storage leakage temperature datasets, a regional temperature-hydrogen leakage degree model is constructed. This model is then used to analyze the leakage degree and convert the risk level of the key hydrogen storage region's sequential temperature data, obtaining the risk level of the key region's hydrogen sequence. Based on the risk level of the key region's hydrogen sequence, the risk level of the hydrogen leakage sequence is verified and weighted to determine the quantification result of the hydrogen micro-leakage classification.

[0082] Furthermore, the early warning decision response execution module 4 is used to perform the following steps: Based on the risk level of the hydrogen sequence in the key area, the consistency of the risk level of the hydrogen leakage sequence is verified to obtain the consistency result of the hydrogen leakage sequence; according to the consistency result of the hydrogen leakage sequence, the risk level of the hydrogen sequence in the key area and the risk level of the hydrogen leakage sequence are weighted and quantified to determine the grading and quantification result of the hydrogen micro-leakage.

[0083] The hydrogen micro-leakage classification and quantification system for hydrogen storage environments provided in this embodiment of the invention can execute the hydrogen micro-leakage classification and quantification method for hydrogen storage environments provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0084] Although this application makes various references to certain modules in the system according to the embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy distinction between each other and are not used to limit the scope of protection of this invention.

[0085] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application. In some cases, the actions or steps described in this application can be performed in a different order than that shown in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A method for the classification and quantification of hydrogen micro-leakage in hydrogen storage environments, characterized in that, The method includes: Deploy hydrogen leak detection tape and a fixed thermal imager in the target hydrogen storage environment, and acquire hydrogen leak image data and temperature data of key hydrogen storage areas through the hydrogen leak detection tape and the fixed thermal imager. The brightness values ​​of the hydrogen-sensitive tape leakage image data are extracted using the HSV color space to obtain hydrogen leakage brightness change data. Based on the hydrogen leakage brightness change data, a quantitative model of the degree of hydrogen leakage is established. Time series analysis and risk level conversion were performed on the hydrogen leakage degree quantification model to obtain the hydrogen leakage sequence risk level. By combining the temperature data of the key hydrogen storage area, a weighted quantitative analysis of the risk level of the hydrogen leakage sequence is performed to determine the grading and quantification results of hydrogen micro-leakage, and early warning decision-making response is made based on the grading and quantification results of hydrogen micro-leakage. Establish a quantitative model for the degree of hydrogen leakage, including: Collect a sample set of hydrogen-sensitive tape leak images, train the leak degree assessment model on the sample set of hydrogen-sensitive tape leak images, and generate a hydrogen leak degree assessment model. The hydrogen leakage level assessment model is used to assess the hydrogen leakage brightness change data and output the hydrogen leakage level change level. The hydrogen leakage level change is quantitatively fitted according to the time series to establish a quantitative model of the hydrogen leakage level; A hydrogen leakage assessment model was developed, including: The brightness values ​​of the hydrogen-sensitive tape leakage image sample set were extracted using the HSV color space to obtain the hydrogen-sensitive tape leakage brightness sample set. The leakage intensity sample set of the hydrogen-sensitive tape is classified and labeled according to the leakage intensity level to obtain the hydrogen-sensitive tape leakage intensity sample set; The hydrogen leakage assessment model is generated by training a deep neural network on the sample set of hydrogen-sensitive tape leakage levels to assess the leakage level. The risk levels of the hydrogen leak sequence were obtained, including: A time series trend analysis was performed on the hydrogen leakage level quantification model to generate a hydrogen leakage level fluctuation trend chart. Based on the requirements of hydrogen storage environment safety management and hydrogen storage application, a hydrogen leakage risk classification system shall be constructed. Based on the hydrogen leakage risk classification system, the risk level of the hydrogen leakage level fluctuation trend chart is converted into a risk level to obtain the hydrogen leakage sequence risk level.

2. The method for grading and quantifying hydrogen micro-leakage in hydrogen storage environments as described in claim 1, characterized in that, Deploying hydrogen leak detection tape and fixed thermal imagers in the target hydrogen storage environment includes: Obtain information on the distribution of hydrogen storage areas and the attribute information of hydrogen storage equipment in the target hydrogen storage environment; Based on the attribute information of the hydrogen storage equipment, hydrogen leakage risk areas are identified to obtain a set of hydrogen storage equipment leakage risk points. According to the set of leakage risk points of the hydrogen storage equipment, the hydrogen leakage detection tape is surface-treated and deployed and marked on the target hydrogen storage environment; Based on the hydrogen storage area distribution information, a temperature monitoring coverage analysis is performed to determine the set of thermal imager deployment locations, and fixed thermal imagers are deployed on the target hydrogen storage environment according to the set of thermal imager deployment locations.

3. The method for grading and quantifying hydrogen micro-leakage in hydrogen storage environments as described in claim 1, characterized in that, Data on the brightness changes during hydrogen leakage were obtained, including: The hydrogen-sensitive tape leakage image data is read and converted to RGB color space to obtain an RGB color image of the hydrogen-sensitive tape leakage. A Gaussian filter was used to filter and denoise the hydrogen-sensitive tape leakage RGB color image to obtain a usable hydrogen-sensitive tape leakage RGB color image. Edge detection and background segmentation are performed on the RGB color image of the hydrogen-sensitive tape leak to obtain the foreground region of the hydrogen-sensitive tape leak. The HSV color space was used to perform color space conversion and brightness value extraction on the foreground area of ​​the hydrogen-sensitive tape leakage to obtain the brightness change data of the hydrogen leakage.

4. The method for grading and quantifying hydrogen micro-leakage in hydrogen storage environments as described in claim 1, characterized in that, Determine the grading and quantification results of hydrogen micro-leakage, including: The temperature data of the key hydrogen storage area is aligned with the risk level of the hydrogen leakage sequence over time to obtain the temperature data sequence of the key hydrogen storage area. By combining the temperature data of the key hydrogen storage area sequence, the risk level of the hydrogen leakage sequence is verified and weighted to determine the quantification result of the hydrogen micro-leakage classification.

5. The method for grading and quantifying hydrogen micro-leakage in hydrogen storage environments as described in claim 4, characterized in that, By combining the temperature data of the key hydrogen storage area sequence, the risk level of the hydrogen leakage sequence is verified and weighted to determine the quantification result of the hydrogen micro-leakage classification, including: Based on historical hydrogen storage leakage temperature dataset, a regional temperature-hydrogen leakage degree model is constructed. The leakage degree analysis and risk level conversion of the key hydrogen storage area sequence temperature data are performed by the regional temperature-hydrogen leakage degree model to obtain the risk level of the key area hydrogen sequence. Based on the risk level of the hydrogen sequence in the key area, the risk level of the hydrogen leakage sequence is verified and weighted to determine the quantification result of the hydrogen micro-leakage classification.

6. The method for grading and quantifying hydrogen micro-leakage in hydrogen storage environments as described in claim 5, characterized in that, Based on the risk level of the hydrogen sequence in the key area, the risk level of the hydrogen leakage sequence is verified and weighted to determine the quantification result of the hydrogen micro-leakage classification, including: Based on the risk level of the hydrogen sequence in the key area, the consistency of the risk level of the hydrogen leakage sequence is verified, and the consistency result of the hydrogen leakage sequence is obtained. The risk levels of the hydrogen sequence in the key area and the risk levels of the hydrogen leakage sequence are weighted and quantified according to the consistency results of the hydrogen leakage sequence to determine the grading and quantification results of the hydrogen micro-leakage.

7. A hydrogen micro-leakage classification and quantification system for hydrogen storage environments, characterized in that, For implementing the hydrogen micro-leakage classification and quantification method for hydrogen storage environments according to any one of claims 1-6, the system comprises: The image temperature data acquisition module is used to deploy hydrogen leak detection tape and a fixed thermal imager in the target hydrogen storage environment, and to acquire hydrogen leak image data and key hydrogen storage area temperature data through the hydrogen leak detection tape and the fixed thermal imager. The hydrogen leakage degree quantification model construction module is used to extract the brightness value of the hydrogen-sensitive tape leakage image data using the HSV color space to obtain hydrogen leakage brightness change data, and to establish a hydrogen leakage degree quantification model based on the hydrogen leakage brightness change data. The hydrogen leakage sequence risk level acquisition module is used to perform time series analysis and risk level conversion on the hydrogen leakage degree quantification model to obtain the hydrogen leakage sequence risk level. The early warning decision response execution module is used to perform weighted quantitative analysis of the risk level of the hydrogen leakage sequence by combining the temperature data of the key hydrogen storage area, determine the graded quantitative result of hydrogen micro-leakage, and perform early warning decision response based on the graded quantitative result of hydrogen micro-leakage.

Images