Hydrogen-doped gas storage and handling facility leak detection methods, systems, and apparatuses

Abstract

The present application belongs to the technical field of hydrogen leakage detection, and particularly relates to a hydrogen-doped gas storage and transportation facility leakage detection method, system and device, aiming to solve the problem of low efficiency, easy misjudgment or missed detection of the existing leakage detection method. The present application comprises: applying a hydrogen-sensitive color developing gel to a to-be-detected area, collecting gel image information of the to-be-detected area, comparing the gel image with a database image through a pipeline leakage detection neural network, obtaining a leakage source type or a superimposed leakage source type of pipeline leakage, accurately judging the leakage source position through CFD forward and inverse comparison, and simultaneously obtaining the degree of pipeline leakage. The present application solves the problem of easy misjudgment or missed detection of the traditional adhesive tape method, can accurately identify the leakage source type, leakage degree and accurate position of the pipeline, improves the detection accuracy, and the detection method is simple.

Description

Technical Field

[0001] This invention belongs to the field of hydrogen leak detection technology, specifically relating to a method, system, and equipment for accurate leak detection in hydrogen-blended gas storage and transportation facilities. Background Technology

[0002] The development and utilization of new energy sources has become an important direction for future development. Hydrogen energy, as a renewable energy source and a 21st-century energy source, has seen increasing research interest in this area. In numerous scientific research and practical applications, the preparation, storage, and transportation of hydrogen energy have received widespread attention. However, due to the high energy flux density and high explosion limits (4%-75%) of hydrogen, its large-scale promotion and application have been hindered for a long time. To efficiently promote the large-scale application of hydrogen, safety is a crucial issue that needs to be addressed.

[0003] Safety is paramount during the storage and transportation of hydrogen. Hydrogen atoms are extremely small, easily penetrating iron atoms within steel containers to react with carbon, leading to container failure. Similarly, flange connections and pipe welds are unavoidable at stations and between pipeline sections, and these high-risk locations are prone to leaks ranging from micro-leaks to full-blown ruptures. Therefore, early detection and precise sealing of micro-leak sources are crucial for improving the safety of hydrogen energy utilization.

[0004] The conventional method for locating the leak source mainly consists of the following steps: (1) The station operates normally until the detection threshold of the hydrogen-sensitive alarm is reached, and an alarm signal is issued; (2) The station pipeline is shut down to investigate risks; (3) Patrol workers use handheld hydrogen-sensitive detectors to search for leak sources along the line; (4) Construction is carried out to seal and repair the pipeline section; (5) Production is resumed.

[0005] Because the location of leaks cannot be predicted immediately, risk assessment time will increase significantly, potentially leading to substantial economic losses. During manual pipeline inspections, the simplest and quickest method is to use soapy water or foaming water for initial leak location detection, primarily relying on the visual appearance of bubbles in the soapy water to indicate the approximate location of the leak; or by wrapping leak detection tape around the leak and observing a discoloration of a specific area as a leak marker. However, these two methods cannot provide the morphology and location of micro-leaks, failing to offer sufficient information for subsequent maintenance and protection. This can lead to missed detections or failed risk assessments, resulting in subsequent pipeline ruptures or production accidents. Therefore, developing a simple method that accurately and visually displays the location of leaks, especially micro-leaks, and can describe the leak type in relatively detailed terms is extremely necessary in actual production processes. Summary of the Invention

[0006] To address the aforementioned problems in existing technologies, namely the low efficiency and susceptibility to false or false detections in current leak detection methods, this invention proposes a precise method for detecting micro-leakage in hydrogen-doped gas storage and transportation facilities. This method is applicable to the detection of storage and transportation facilities, tanks, valves, and other related fields, and includes:

[0007] Step S100: Apply the hydrogen-sensitive colorimetric gel to the area to be tested;

[0008] Step S200: Continuously acquire gel image information of the area to be detected over a continuous time period; the gel image information includes an outer surface view image and a cross-sectional view image;

[0009] Step S300: Based on the gel image information, obtain leakage information through a pipeline leakage detection neural network.

[0010] In some preferred embodiments, the hydrogen-sensitive colorimetric gel comprises a first detection gel with a water content of 65%-85% and a second detection gel with a water content of 65%-40%.

[0011] In some preferred embodiments, step S100 specifically involves the following steps: when the station leak detection alarm sounds and the alarm is discontinuous, the leak-prone area is designated as the area to be detected, and hydrogen-sensitive color-developing gel is applied to the area to be detected. Specifically, a second detection gel is made into a strip and wrapped around the flange; a first detection gel is applied to the pipe weld or the stress concentration position of the pipe body, left to stand for a preset waiting time, and after the gel is removed, the leak location is marked. If a leak is present, a cloud-like discoloration area will appear on the outer surface of the gel, and a columnar discoloration area can be seen when viewed from the side. Different leak sources result in different gel discoloration column shapes.

[0012] In some preferred embodiments, the leakage information includes the type of leakage source, the extent of leakage, and the precise location of leakage.

[0013] In some preferred embodiments, the leakage source types include point leakage sources, point leakage source superposition, strip leakage sources, strip leakage source superposition, and point and strip leakage source superposition.

[0014] In some preferred embodiments, when the leakage source type is a point leakage source, the color change shape of the outer surface view image of the gel image information is O-shaped; when the leakage source type is a strip leakage source, the color change shape of the outer surface view image of the gel image information is a segmented I-shaped; when the leakage source type is two points, the color change shape of the outer surface view image of the gel image information is Φ-shaped.

[0015] In some preferred embodiments, when the leakage source type is a superposition of point leakage sources, as the leakage point location continues to increase and diffuses to the outer plane of the gel, a square-like color-changing area appears in the outer surface view image;

[0016] When the leakage source type is a superposition of strip-shaped leakage sources, the superposition occurs in the middle of the strip-shaped discoloration area, the degree of discoloration increases dramatically, and a diamond-shaped discoloration area appears; in particular, when the strip-shaped leakage sources are superimposed vertically, three jet columns will form in the discoloration area; when three strip-shaped leakage sources are superimposed, four jet columns will appear in the discoloration area.

[0017] In some preferred embodiments, when the leakage source type is a combination of point and strip leakage sources, parabolic, key-shaped and rocket-shaped color-changing areas appear respectively.

[0018] In some preferred embodiments, the training method for the pipeline leak detection neural network is as follows:

[0019] Step A100: Based on the set leak point location A(x1, y1, z1), the diffusion law and cloud map of hydrogen or hydrogen-doped gas in the hydrogel under different leak conditions are obtained by CFD positive time simulation. Simulated images of point leak source, point leak source superposition, strip leak source, superimposed leak source and point and strip leak source superposition are generated by COMSOL Multiphysics forward simulation as the initial training set images.

[0020] Step A200: Label the simulated image with the corresponding leak type, leak level, and precise leak location;

[0021] Step A300: Input the training set images of each time period into the pipeline leak detection neural network, and output the training set leak type, training set leak degree and training set leak estimation location A″;

[0022] Step A400: Compare the training set leakage type, training set leakage degree, and training set leakage estimation position A″ with the label, calculate the loss function, and use the stochastic gradient descent algorithm until the loss function is lower than a preset threshold to obtain the trained pipeline leakage detection neural network.

[0023] In some preferred embodiments, the training method of the pipeline leak detection neural network further includes a step of inverse time inversion to improve accuracy, specifically:

[0024] Step A500: Data slices are made on the surface view image and cross-sectional view image at a uniform time interval. The inverted leak point location A′(x2, y2, z2) is obtained by CFD inverse time inversion. The inverted leak point location is compared with the leak point location A(x1, y1, z1) extrapolated in forward time to preliminarily determine the leak source location difference and obtain the judgment factor ξ.

[0025] Step A600: The judgment factor is ξ = |AA′|. If |AA′| > ε, then the steps A100-A500 are repeated continuously and the network parameters are adjusted. The estimated leakage location A″A″ in the training set is compared with A and A′. When the judgment factor |AA′| + |A′A″| + |AA″| < ε, it indicates that the calculated accurate leakage location A″ in the training set is the accurate location, and the trained pipeline leakage detection neural network is obtained.

[0026] In another aspect, the present invention provides a leak detection system for hydrogen-doped gas storage and transportation facilities, characterized in that the system comprises: a dynamic gel image information acquisition module and a leak identification module;

[0027] The gel image information dynamic acquisition module is configured to take a picture every Δt time interval according to actual needs, and to collect gel image information of the area to be detected as a reference for dynamic image slicing.

[0028] The leakage identification module is configured to obtain leakage information based on the gel image information and through a pipeline leakage detection neural network.

[0029] A third aspect of the present invention provides an electronic device comprising: at least one processor; and a memory communicatively connected to at least one of the processors; wherein the memory stores instructions executable by the processor for implementing the above-described method for detecting leaks in hydrogen-doped gas storage and transportation facilities.

[0030] In a fourth aspect, the present invention provides a computer-readable storage medium storing computer instructions for execution by a computer to implement the above-described method for detecting leaks in hydrogen-doped gas storage and transportation facilities.

[0031] The beneficial effects of this invention are:

[0032] (1) The method for detecting leaks in hydrogen-doped gas storage and transportation facilities proposed in this invention solves the problem of easy false detection or missed detection in the traditional tape method. It can accurately identify the type of leak source, the degree of leakage and the precise location of the leak, thus improving the accuracy of detection. Attached Figure Description

[0033] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0034] Figure 1 This is a schematic flowchart of the method for detecting leaks in hydrogen-doped gas storage and transportation facilities in an embodiment of the present invention;

[0035] Figure 2This is a schematic diagram of the first stage of applying gel to detect micro-leakage in pipelines or storage and transportation facilities in an embodiment of the present invention;

[0036] Figure 3 This is a schematic diagram of the second stage of applying gel to detect micro-leakage in pipelines or storage and transportation facilities in an embodiment of the present invention;

[0037] Figure 4 The simulation results of point leakage sources, point leakage source superposition, and strip leakage sources obtained by using COMSOL Multiphysics software in the embodiments of the present invention are as follows:

[0038] Figure 5 These are diffusion simulation diagrams of different leakage source types in embodiments of the present invention;

[0039] Figure 6 This is a simulation diagram of the diffusion effect of different leakage source types superimposed in an embodiment of the present invention;

[0040] Figure 7 This is a cloud map and effect diagram of the detection of a leak at a certain location in an embodiment of the present invention. Detailed Implementation

[0041] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0042] 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.

[0043] This invention proposes a method for detecting leaks in hydrogen-doped gas storage and transportation facilities, comprising:

[0044] Step S100: Apply the hydrogen-sensitive colorimetric gel to the area to be tested;

[0045] Step S200: Collect gel image information of the area to be detected over a continuous time period; the gel image information includes an outer surface view image and a cross-sectional view image;

[0046] Step S300: Based on the gel image information, obtain leakage information through a pipeline leakage detection neural network.

[0047] This method solves the problem of false detection or missed detection that is common in traditional tape methods, and can accurately identify the type, degree and precise location of pipeline leaks.

[0048] To more clearly explain the leakage detection method for hydrogen-doped gas storage and transportation facilities of the present invention, the following is in conjunction with... Figure 1 The steps in the embodiments of the present invention will be described in detail below.

[0049] In the field of leak detection, directly using image recognition to identify and determine the source of a leak is mostly focused on pinpointing a certain area of ​​the leak. However, this method of pinpointing the area may miss the crucial indicator of the leak's morphological characteristics, posing a certain risk to subsequent leak sealing or treatment. This invention provides a novel solution to this problem.

[0050] Precisely locating leak sources using images displayed by leak gels requires two preparations: first, a computational fluid dynamics (CFD) model library; and second, an image recognition neural network algorithm and a corresponding database. To accurately locate the leak source and identify its type, a verification subroutine based on CFD forward and inverse modeling is incorporated into the image recognition neural network algorithm.

[0051] The method for detecting leaks in hydrogen-doped gas storage and transportation facilities according to the first embodiment of the present invention includes:

[0052] Step S100: Apply hydrogen-sensitive colorimetric gel to the area to be detected; the hydrogen-sensitive colorimetric gel can be pre-wrapped on the surface of the storage and transportation facility that needs to be monitored in real time to observe the gel image information at each location.

[0053] In this embodiment, the hydrogen-sensitive colorimetric gel includes a first detection gel with a water content of 65%-85% and a second detection gel with a water content of 65%-40%.

[0054] In this embodiment, step S100 specifically involves the following steps: When the station leak detection alarm sounds and the alarm is discontinuous, the leak-prone area is designated as the area to be detected. Hydrogen-sensitive color-developing gel is applied to the area to be detected. Specifically, a second detection gel is made into a strip and wrapped around the flange; a first detection gel is applied to the pipe weld or stress concentration point of the pipe body; after a preset waiting time, the gel is removed, and the leak location is marked. If a leak exists, a radial point source color-changing area will appear in the gel. The leak-prone area includes hazardous point sources and hazardous surface sources. Hazardous point sources include concealed leak sources and stress concentration areas, specifically including pipe openings, flange connections, and weld locations. Hazardous surface sources refer to leaks occurring on the surface of tanks with small radii of curvature or large-diameter pipe sections. In this case, applying strip-shaped gel is insufficient for large-area detection; therefore, a thin gel is used for surface coating.

[0055] The physical characteristics of micro-leakage in stress concentration areas such as valves and welds include the following: (1) The pressure is not high, and it is basically a small hole, strip or band leakage at normal pressure; (2) The leakage occurs at normal temperature and pressure, and there will be no obvious temperature rise or drop; (3) The micro-leakage occurs in a buried environment, and the leakage process is similar to the diffusion process of natural gas in a porous medium.

[0056] Therefore, micro-leakage is difficult to detect in actual production processes. When using the conventional direct detection tape method for initial inspection of leak locations, it is easy to make false or missed detections.

[0057] Detecting micro-leaks in pipelines or storage facilities by applying gel, such as Figure 2 and Figure 3 As shown, the process is divided into two stages. In the first stage, when the gel is coated on the surface, air bubbles will appear between the tube wall and the gel layer. Since the leak is near the air to be detected, the entrained air bubbles will contain hydrogen gas. Then, the unsaturated hydrogel will gradually absorb moisture from the air until it reaches saturation. In the second stage, under the action of surface tension, the gel will quickly fill the area to be detected in a short time. During this process, a large amount of residual gas will be squeezed out. Because the micro-leakage process occurs slowly, hydrogen molecules will gradually penetrate into the gel, forming obvious gas columns, which can relatively accurately locate the leak source and greatly reduce missed or false detections.

[0058] Step S200: Continuously acquire gel image information of the area to be detected over a continuous time period; the gel image information includes an outer surface view image and a cross-sectional view image;

[0059] Step S300: Based on the gel image information, obtain leakage information through a pipeline leakage detection neural network.

[0060] In this embodiment, the leakage information includes the type of leakage source, the extent of leakage, and the precise location of leakage.

[0061] In this embodiment, the leakage source types include point leakage sources, point leakage source superposition, strip leakage sources, strip leakage source superposition, and point and strip leakage source superposition.

[0062] Regarding the fluid flow and heat / mass transfer problems involved in this invention, assuming a micro-leakage pressure of 100 Pa, simulations were performed using COMSOL Multiphysics software. The simulation results for point leak sources, superimposed point leak sources, and strip-shaped leak sources are as follows: Figure 4 As shown, in Figure 4As can be seen, within a certain release time, a gas column appears inside the gel, its distribution showing a diffusion path from point source to surface source. Within this region, H2, under the catalysis of nano-Pt particles, can react with WO3, TiO2, and other particles doped in a certain proportion, resulting in a cone-shaped discoloration area. Subsequently, by determining the center of the ellipse or circle on the outer side of the gel, the leak source location can be found relatively accurately. Similarly, when multiple elliptical and circular discoloration areas appear on the surface, we can locate multiple leak locations through positioning. Circular leak sources are mostly caused by corrosion perforation, while strip-shaped leak sources are mainly caused by hydrogen-induced cracking and pipe section settlement leading to micro-misalignment at the flange connection. Therefore, when simulating the component graphic library, three main categories were considered: the number and relative positions of different point leak sources, the relative positions of points and strip-shaped leak sources, and the relative positions of strip-shaped leak sources. The above three scenarios represent corrosion leakage, corrosion and hydrogen-induced cracking, and hydrogen-induced cracking and flange misalignment, respectively.

[0063] In this embodiment, when the leakage source type is a point leakage source, the outer surface view image of the gel image information is type O; when the leakage source type is a strip leakage source, the outer surface view image of the gel image information is a segmented type I; when the leakage source type is two points, the outer surface view image of the gel image information is type Φ. Diffusion simulation for different leakage source types is as follows: Figure 5 As shown. By Figure 5 As can be seen, when multiple types of leak holes appear at different locations, the pattern reflected on the outside of the detection gel should be a superposition of the aforementioned patterns. Therefore, we can relatively accurately locate and describe the type and location of the leak holes.

[0064] In this embodiment, when the leakage source type is a superposition of point leakage sources, as the location of the leakage point continues to increase and diffuses to the outer plane of the gel, a square-like color-changing area appears in the outer surface view image;

[0065] When the leakage source type is a superposition of strip-shaped leakage sources, the superposition occurs in the middle of the strip-shaped discoloration area, the degree of discoloration increases dramatically, and a diamond-shaped discoloration area appears; in particular, when the strip-shaped leakage sources are superimposed vertically, three jet columns will form in the discoloration area; when three strip-shaped leakage sources are superimposed, four jet columns will appear in the discoloration area.

[0066] In this embodiment, when point and strip-shaped leak sources are superimposed, parabolic, key-shaped, and rocket-shaped color-changing areas appear respectively. We can roughly determine the leak location, number of leak points, and type of leak source by superimposing the leak sources. This leak detection technology and identification method can help to quickly and directly detect leaks and minimize the possibility of pipeline leak explosion accidents.

[0067] Diffusion simulation effects of different leakage source types, such as Figure 6 As shown.

[0068] Examples of defining and determining leakage types using this invention include Figure 7 As shown, in Figure 7 In this study, the collected gel image information was processed and compared with point leak sources, strip leak sources, and diffusion cloud maps formed by superposition in the database to identify the leak conditions. It was determined that the leak was caused by four point leak sources and two strip leak sources. It can be concluded that the leak was caused by the superposition of hydrogen-induced cracking and modified perforation, and that the large number of leak points within a single plane made it prone to dangerous leaks, requiring prompt action.

[0069] In this embodiment, the training method for the pipeline leak detection neural network is as follows:

[0070] Step A100: Based on the set leak point location A(x1, y1, z1), the diffusion law and cloud map of hydrogen or hydrogen-doped gas in the hydrogel under different leak conditions are obtained by CFD positive time simulation. Simulated images of point leak source, point leak source superposition, strip leak source, superimposed leak source and point and strip leak source superposition are generated by COMSOL Multiphysics forward simulation as the initial training set images.

[0071] Step A200: Label the simulated image with the corresponding leak type, leak level, and precise leak location;

[0072] Step A300: Input the training set images of each time period into the pipeline leak detection neural network, and output the training set leak type, training set leak degree and training set leak estimation location A″;

[0073] Step A400: Compare the training set leakage type, training set leakage degree, and training set leakage estimation position A″ with the label, calculate the loss function, and use the stochastic gradient descent algorithm until the loss function is lower than a preset threshold to obtain the trained pipeline leakage detection neural network.

[0074] In this embodiment, the method further includes a step of inverse time inversion to improve accuracy. Specifically, step A500 involves slicing the surface view image and cross-sectional view image over a uniform time interval, obtaining the inverted leak point location A′(x2, y2, z2) through CFD inverse time inversion, comparing the inverted leak point location with the leak point location A(x1, y1, z1) derived in forward time inversion to preliminarily determine the leak source location difference, and obtaining the judgment factor ξ.

[0075] In step A600, the judgment factor is v = |AA′|. If |AA′| > ε, then the steps A100-A500 are repeated continuously and the network parameters are adjusted. The estimated leakage location A″A″ in the training set is compared with A and A′. When the judgment factor |AA′| + |A′A″| + |AA″| < ε, it indicates that the calculated accurate leakage location A″ in the training set is the accurate location, and the trained pipeline leakage detection neural network is obtained.

[0076] Another aspect of the present invention proposes a leak detection system for hydrogen-blended gas storage and transportation facilities, comprising functional modules described in detail below:

[0077] The system includes: a dynamic gel image information acquisition module and a leakage identification module;

[0078] The gel image information dynamic acquisition module is configured to take a picture every Δt time interval according to actual needs, and to collect gel image information of the area to be detected as a reference for dynamic image slicing.

[0079] The leakage identification module is configured to obtain leakage information based on the gel image information through a pipeline leakage detection neural network.

[0080] The CFD forward and inverse subroutines, as one of the inputs to the neural network algorithm, are used to accurately determine the location of the leak source.

[0081] An electronic device according to a third embodiment of the present invention includes: at least one processor; and a memory communicatively connected to at least one of the processors; wherein the memory stores instructions executable by the processor, the instructions being executed by the processor to implement the above-described method for detecting leaks in hydrogen-doped gas storage and transportation facilities.

[0082] A fourth embodiment of the present invention provides a computer-readable storage medium storing computer instructions, which are executed by the computer to implement the above-described method for detecting leaks in hydrogen-doped gas storage and transportation facilities.

[0083] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process and related descriptions of the storage device and processing device described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0084] The terms “first”, “second”, etc., are used to distinguish similar objects, not to describe or indicate a specific order or sequence.

[0085] The term "comprising" or any other similar term is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus / device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent in such process, method, article, or apparatus / device.

[0086] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A method for detecting leaks in hydrogen-blended gas storage and transportation facilities, characterized in that, The method includes: Step S100: Apply the hydrogen-sensitive colorimetric gel to the area to be tested; Step S200: Continuously acquire gel image information of the area to be detected over a continuous time period; the gel image information includes an outer surface view image and a cross-sectional view image; Step S300: Based on the gel image information, obtain leakage information through a pipeline leakage detection neural network; The training method for the pipeline leakage detection neural network is as follows: Step A100: Based on the set leak point location A(x1,y1,z1), the diffusion law and cloud map of hydrogen or hydrogen-doped gas in hydrogel under different leak conditions are obtained by CFD positive time simulation. Simulated images of point leak source, point leak source superposition, strip leak source, superimposed leak source and point and strip leak source superposition are generated by COMSOL Multiphysics forward simulation as the initial training set images. Step A200: Label the simulated image with the corresponding leak source type, leak level, and precise leak location; Step A300: Input the training set images of each time period into the pipeline leak detection neural network, and output the training set leak type, training set leak degree and training set leak estimation location A″; Step A400: Compare the training set leakage type, training set leakage degree, and training set leakage estimation position A″ with the label, calculate the loss function, and use the stochastic gradient descent algorithm until the loss function is lower than a preset threshold to obtain the trained pipeline leakage detection neural network.

2. The method for detecting leaks in hydrogen-blended gas storage and transportation facilities according to claim 1, characterized in that, The hydrogen-sensitive colorimetric gel includes a first detection gel with a water content of 65%-85% and a second detection gel with a water content of 65%-40%.

3. The method for detecting leaks in hydrogen-blended gas storage and transportation facilities as described in claim 2, characterized in that, Step S100 specifically involves the following steps: When the station leak detection alarm sounds and the alarm is discontinuous, the leak-prone area is designated as the area to be detected. Hydrogen-sensitive color-developing gel is applied to the area to be detected. Specifically, the second detection gel is made into a strip and wrapped around the flange; the first detection gel is applied to the pipe weld or the stress concentration position of the pipe body, left to stand for a preset waiting time, and after the gel is removed, the leak location is marked. If a leak is found, a radial point source color-changing area will appear in the gel.

4. The method for detecting leaks in hydrogen-blended gas storage and transportation facilities according to claim 1, characterized in that, The leakage information includes the type of leakage source, the extent of the leakage, and the precise location of the leakage.

5. The method for detecting leaks in hydrogen-blended gas storage and transportation facilities according to claim 4, characterized in that, The types of leakage sources include point leakage sources, superimposed point leakage sources, strip leakage sources, superimposed strip leakage sources, and superimposed point and strip leakage sources.

6. The method for detecting leaks in hydrogen-blended gas storage and transportation facilities according to claim 5, characterized in that, When the leakage source type is a point leakage source, the color change shape of the outer surface view image of the gel image information is O-shaped; when the leakage source type is a strip leakage source, the color change shape of the outer surface view image of the gel image information is a segmented I-shaped; when the leakage source type is two points, the color change shape of the outer surface view image of the gel image information is Φ-shaped.

7. The method for detecting leaks in hydrogen-blended gas storage and transportation facilities according to claim 5, characterized in that, When the leakage source type is a superposition of point leakage sources, as the location of the leakage point continues to increase and diffuses to the outer plane of the gel, a square-like discolored area appears in the external surface view image; When the leakage source type is a superimposed strip-shaped leakage source, the superposition occurs in the middle of the strip-shaped discoloration area, the degree of discoloration increases dramatically, and a diamond-shaped discoloration area appears.

8. The method for detecting leaks in hydrogen-blended gas storage and transportation facilities according to claim 5, characterized in that, When point and strip-shaped leakage sources are superimposed, parabolic, key-shaped, or rocket-shaped color-changing areas appear respectively.

9. The method for detecting leaks in hydrogen-blended gas storage and transportation facilities according to claim 1, characterized in that, The training method for the pipeline leak detection neural network further includes a step of inverse time reversal to improve accuracy, specifically: Step A500: Data slices are made on the surface view image and cross-sectional view image at a uniform time interval. The inverted leak point location A′(x2,y2,z2) is obtained by CFD inverse time inversion. The inverted leak point location is compared with the leak point location A(x1,y1,z1) extrapolated in forward time to preliminarily determine the leak source location difference and obtain the judgment factor ξ. Step A600: The judgment factor is ξ = |AA'|. If |AA'| > ε, then the steps A100-A500 are repeated continuously and the network parameters are adjusted. The estimated leakage location A″ in the training set is compared with A and A'. When the judgment factor |AA'| + |A'A"| + |AA"| < ε, it indicates that the calculated estimated leakage location A″ in the training set is an accurate location, and the trained pipeline leakage detection neural network is obtained.

10. A leak detection system for hydrogen-blended gas storage and transportation facilities, characterized in that, The system includes: a dynamic gel image information acquisition module and a leakage identification module; The gel image information dynamic acquisition module is configured to take a picture every Δt time interval according to actual needs, and continuously acquire gel image information of the area to be detected over a continuous time period as a reference for dynamic image slicing. The leakage identification module is configured to obtain leakage information based on the gel image information through a pipeline leakage detection neural network. The training method for the pipeline leak detection neural network is as follows: Step A100: Based on the set leak point location A(x1,y1,z1), the diffusion law and cloud map of hydrogen or hydrogen-doped gas in hydrogel under different leak conditions are obtained by CFD positive time simulation. Simulated images of point leak source, point leak source superposition, strip leak source, superimposed leak source and point and strip leak source superposition are generated by COMSOL Multiphysics forward simulation as the initial training set images. Step A200: Label the simulated image with the corresponding leak source type, leak level, and precise leak location; Step A300: Input the training set images of each time period into the pipeline leak detection neural network, and output the training set leak type, training set leak degree and training set leak estimation location A″; Step A400: Compare the training set leakage type, training set leakage degree, and training set leakage estimation position A″ with the label, calculate the loss function, and use the stochastic gradient descent algorithm until the loss function is lower than a preset threshold to obtain the trained pipeline leakage detection neural network.

Images