Gas leakage real-time monitoring method and system

Abstract

The invention discloses a gas leakage real-time monitoring method and system, and the method comprises the steps: carrying out the three-dimensional modeling of a target chemical device, carrying out the fluid simulation of target gas, obtaining key data, training a traceability model, and achieving the real-time monitoring of gas leakage through the traceability model obtained through training. The monitoring requirements of different chemical devices and gas types can be met, and a flexible and comprehensive traceability solution is provided; real-time and accurate leakage source positioning can be provided, so that timely maintenance and treatment can be supported, the accident occurrence frequency is reduced, the environmental and health risks are reduced, a scientific and efficient gas leakage control means is provided for enterprises, and the environmental management level and social liability image of the enterprises are improved.

Description

Technical Field

[0001] This invention belongs to the field of gas leakage monitoring technology in chemical plants, and particularly relates to a real-time gas leakage monitoring method and system. Background Technology

[0002] Gas leaks in chemical plants are a common and potentially dangerous safety issue. Due to differences in production processes, materials used, and output products, leaked gases from chemical plants are diverse, including but not limited to various volatile organic compounds, greenhouse gases, nitrogen oxides, and sulfur oxides. Given the high frequency and sudden nature of leaks in chemical plants, gas leaks can lead to severe environmental pollution, economic losses, and even threaten human lives. Therefore, how to effectively monitor and control gas leaks has become a focus of attention for industry and academia. However, current gas leak monitoring technologies on the market have significant shortcomings and urgently need improvement.

[0003] Point detectors are among the most commonly used online gas leak detection devices. These detectors are typically installed in specific locations to form online gas leak monitoring systems, such as combustible and toxic gas detection alarm systems (GDS), which identify leaks by continuously monitoring changes in the concentration of the target gas. However, the alarm thresholds of these point detectors are usually set relatively high to ensure the effectiveness of leak detection, making it difficult to balance the timeliness of leak detection. Leaks are often only detected after they have developed to a certain extent, resulting in a loss of the "first-mover advantage" in gas leak control.

[0004] Meanwhile, most existing online gas leak monitoring systems based on point detectors lack intelligent source tracing capabilities. Once a leak occurs, the source is usually found through experience-based analysis and large-scale manual investigation, which is not only time-consuming and labor-intensive but may also lead to further escalation of the danger.

[0005] Offline gas leak monitoring technology primarily focuses on Leak Detection and Repair (LDAR) technology. However, this technology is currently mainly targeted at volatile organic compounds and has not been extended to other toxic and harmful gases. This technology requires personnel to carry portable detection equipment to conduct regular leak checks on both static and dynamic sealing points of equipment and to promptly carry out repairs. Depending on the type of sealing point, the detection interval is typically three to six months. LDAR technology can pinpoint leaks precisely, but it suffers from significant identification lag; leaks may continue to develop until the next detection cycle before being identified. Furthermore, it cannot effectively track leak sources or predict leak trends.

[0006] If gas leak monitoring technology in chemical plants can only identify whether a leak has occurred, but cannot accurately locate the leak or provide effective source tracing suggestions, it will be difficult to support timely maintenance and handling, thus reducing its value and benefits to the enterprise. Similarly, if gas leak detection technology can accurately locate the leak, but the monitoring frequency is extremely low due to cost control, technical limitations, or other reasons, its value to the enterprise will also be greatly reduced.

[0007] Therefore, there is an urgent need for a real-time monitoring technology for gas leaks in chemical plants that balances the timeliness of leak monitoring with the ability to locate leaks, providing industrial enterprises with scientific and efficient gas leak control technology. Summary of the Invention

[0008] Based on this, a method and system for real-time monitoring of gas leaks are provided to address the aforementioned technical problems.

[0009] The technical solution adopted in this invention is as follows: As a first aspect of the present invention, a method for real-time monitoring of gas leaks is provided, characterized in that it includes: A three-dimensional model of the target chemical plant is established using 3D modeling technology. Based on the typical leakage data of each leakage source of the target chemical plant collected in advance and the unique identification information assigned to each leakage source, the diffusion simulation parameters of the fluid simulation software are set, and according to the three-dimensional model, the fluid simulation software simulates the diffusion distribution of the gas plume in the three-dimensional monitoring space based on the physicochemical properties of the target gas and the diffusion simulation parameters. Different combinations of the diffusion simulation parameters are used to obtain the diffusion distribution dataset of the target gas, wherein the three-dimensional monitoring space is a three-dimensional space containing the target chemical plant. Multiple monitoring sensors are installed on the target chemical plant so that the target gas leaked from each leak source can be monitored, and meteorological sensors are installed at unobstructed heights in the three-dimensional monitoring space. Key data is extracted from the diffusion distribution dataset. The key data includes unique identification information of each leakage source, wind field data, and target gas concentration data at the location of each monitoring sensor. The wind field data includes wind direction and wind speed. The neural network model is trained, validated, and tested using the key data to form a source tracing model that takes wind field data and target gas concentration data as input data and unique identification information of each leakage source and corresponding leakage probability as the prediction result output. During monitoring, multiple target gas concentration data and wind field data are collected in real time through the multiple monitoring sensors and meteorological sensors. The multiple target gas concentration data and wind field data are input into the source tracing model to obtain the unique identification information of each leakage source and the corresponding leakage probability.

[0010] As a second aspect of the present invention, a real-time gas leak monitoring system is provided, characterized in that it includes: According to the above-mentioned first aspect of a gas leak real-time monitoring method, there are multiple monitoring sensors, meteorological sensors and a source tracing model; The wireless transmission module is used to wirelessly transmit the target gas concentration data and wind field data collected by the monitoring sensor and meteorological sensor to the real-time monitoring platform. A real-time monitoring platform is provided, which includes a source tracing module. The source tracing module is used to input target gas concentration data and wind field data from the monitoring sensor and meteorological sensor into the source tracing model to obtain unique identification information of each leakage source and the corresponding leakage probability.

[0011] This invention uses 3D modeling of the target chemical plant and fluid simulation of the target gas to obtain key data for training the source tracing model. The trained source tracing model enables real-time monitoring of gas leaks, adapting to the monitoring needs of different chemical plants and gas types, and providing a flexible and comprehensive source tracing solution. It can provide real-time and accurate leak source location, thereby supporting timely maintenance and handling, reducing the frequency of accidents, lowering environmental and health risks, and providing enterprises with a scientific and efficient means of gas leak control, improving their environmental management level and social responsibility image. Attached Figure Description

[0012] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments: Figure 1 A flowchart illustrating a real-time gas leak monitoring method provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of a real-time gas leak monitoring system provided in an embodiment of the present invention. Detailed Implementation

[0013] The embodiments of the present invention will be described below with reference to the accompanying drawings. It should be noted that the embodiments described in this specification are not exhaustive and do not represent the only embodiments of the present invention. The corresponding embodiments below are only for clearly illustrating the inventive content of this patent and are not intended to limit its implementation. For those skilled in the art, different variations and modifications can be made based on the embodiments described. Any variations or modifications that fall within the technical concept and inventive content of this invention and are obvious are also within the protection scope of this invention.

[0014] like Figure 1 As shown, an embodiment of the present invention provides a method for real-time monitoring of gas leaks, including: S101. Establish a three-dimensional model of the target chemical plant using three-dimensional modeling technology.

[0015] In this embodiment, the target chemical plant is an ethylene cracking unit.

[0016] Specifically, 3D modeling can be performed using one of the following two methods: 1. CAD software modeling Collect the piping layout plan of the ethylene cracking unit, and use CAD software such as SolidWorks, AutoCAD, and CATIA to create 3D models of the larger equipment, pipelines, and frames within the unit, which will serve as material for subsequent CFD mesh generation and leakage diffusion simulation.

[0017] 2. Scanning and Reverse Engineering Modeling Choose appropriate 3D scanning technologies based on the size, complexity, and material of the object, such as laser scanning, structured light scanning, and photogrammetry. Use suitable scanning equipment to acquire surface data of the device, forming a point cloud dataset. Perform noise reduction, redundant point removal, and missing data filling on the scanned point cloud. Use software such as Geomagic, PolyWorks, and MeshLab to convert the point cloud into a mesh. Then, use reverse engineering software to convert the mesh data into a CAD model and optimize it to ensure it is suitable for CFD simulation.

[0018] S102. Based on the typical leakage data of each leakage source of the target chemical plant collected in advance and the unique identification information assigned to each leakage source, the diffusion simulation parameters of the fluid simulation software are set, and based on the above three-dimensional model, the fluid simulation software simulates the diffusion distribution of the gas plume in the three-dimensional monitoring space based on the physicochemical properties of the target gas and the diffusion simulation parameters, and different combinations of diffusion simulation parameters are made to obtain the diffusion distribution dataset of the target gas.

[0019] In this embodiment, the location of the leak source and typical leakage conditions of the target chemical plant are determined through process analysis, various historical monitoring data, and monitoring results from other online monitoring methods. Each leak source is then manually assigned a unique identifier (such as a number), and suitable monitoring sensors are selected. The specific process is as follows: 1. Collect information such as the layout plan, process flow diagram, piping and instrumentation diagram, material balance sheet, and process operation procedures of the ethylene cracking unit. Identify the equipment, pipelines, and components within the unit that involve the target gas, establish a control list, and confirm the three-dimensional monitoring space (generally slightly larger than the space occupied by the unit).

[0020] 2. Collect various historical monitoring data, such as periodic LDAR data, corrosion monitoring data, vibration monitoring data, and daily inspection data, and match and compare them with the control list to sort out the historical leakage situation of the objects to be controlled, such as the location of the leakage source, leakage time, number of leaks, leakage concentration, leakage source strength, etc. Aggregate the leakage sources that are clustered in spatial distribution, sort out the leakage source location and typical leakage situation of the target gas in the device, and manually assign each leakage source a unique identification information.

[0021] 3. Collect monitoring results from other online monitoring methods, such as monitoring results from the plant boundary online monitoring station and GDS monitoring results, to help confirm key areas within the three-dimensional monitoring space, typical environmental background values ​​of target gases, and typical meteorological conditions at the location of the equipment.

[0022] 4. Based on the process analysis results, confirm whether the target gas is the most representative characteristic gas in the device, thereby assessing whether the monitoring sensor has any selectivity limitations. The typical environmental background value of the target gas in the device is used to confirm the monitoring sensor's range, sensitivity, resolution, and detection limit, while the typical meteorological conditions at the device location are used to confirm the monitoring sensor's stability and weather resistance.

[0023] The three-dimensional monitoring space is a three-dimensional space containing the target chemical plant. The diffusion simulation parameters include the unique identifier of the simulated leakage source, the source strength level of the simulated leakage source intensity gradient, the simulated leakage source temperature, the simulated wind direction, and the simulated wind speed. When different combinations of diffusion simulation parameters are made, since the simulated leakage source temperature remains constant, the number of diffusion simulation parameter combinations is the product of the number of simulated leakage sources, the number of source strength levels, the number of simulated wind directions, and the number of simulated wind speeds.

[0024] In this embodiment, CFD software is used for fluid simulation. The specific process is as follows: 1. Import the 3D model into ICEM CFD software, fix any possible geometric errors, confirm the mesh generation area, select the mesh type and set the mesh parameters, generate the mesh, check the mesh quality, identify mesh problems that may affect the simulation results, adjust the mesh parameters or use optimization tools to improve the mesh quality, and export the mesh after the quality meets the standards.

[0025] 2. Import the mesh into Ansys Fluent software, select a suitable fluid model, and define gas properties compatible with the target gas (in Fluent, after determining the target gas, the software automatically associates physicochemical properties such as density, viscosity, and diffusion coefficient from its own library to construct a mathematical model, which is then used to calculate the numerical solution). Select a leakage source, define boundary conditions such as fluid inlet, outlet, and wall surface, and set initial conditions such as leakage source strength, leakage source temperature, wind direction, and wind speed. Set up a steady-state solution, defining the number of iterations, convergence criteria, and time step, and perform one solution calculation. Adjust the boundary conditions and initial conditions to complete the leakage diffusion simulation of a single leakage source, and then perform the leakage diffusion simulation of a second leakage source under different boundary conditions and initial conditions. Repeat this process until the leakage diffusion simulation of all leakage sources under all combinations of boundary and initial conditions is completed.

[0026] In this embodiment, the target gas for leak monitoring is ethane, which is not the most representative characteristic gas in the ethylene cracking unit. Therefore, there are limitations in the selectivity of the monitoring sensors.

[0027] S103. Install multiple monitoring sensors on the target chemical plant so that the target gas leaked from each leak source can be monitored, and install meteorological sensors at unobstructed heights in the three-dimensional monitoring space.

[0028] Specifically, in order to minimize the number of sensors and reduce costs, while also ensuring the scientific and rational deployment, the specific process for determining the required number and layout of monitoring sensors is as follows: 1. Based on the performance indicators of the monitoring sensor (effective monitoring distance under different leakage source strengths of the target gas) and the leakage source strength in typical leakage data, determine the monitoring range (monitoring radius or coverage area / space) of a single monitoring sensor. 2. Compare the monitoring range of a single monitoring sensor with the size of the three-dimensional monitoring space to preliminarily determine the required number of monitoring sensors; 3. Based on the required installation space size of the monitoring sensor (the typical required installation space size of the monitoring sensor is approximately 30×30×30cm), extract the coordinates of all possible locations where the monitoring sensor can be installed from the 3D model; 4. Extract the target gas concentration data at each location coordinate from the diffusion distribution dataset, and comprehensively evaluate the sum of the target gas concentration data at each location coordinate, the peak value, the number of high concentration data exceeding the concentration threshold, and the installation convenience score to obtain the priority score for each location coordinate. Taking location point a as an example, assuming 10 simulations are performed, the summation of the target gas concentration data is the sum of the concentration data at point a in the 10 simulations. The peak value refers to the highest concentration value at location point a in the 10 simulations.

[0029] 5. Aggregate the coordinates of location points whose priority scores exceed the score threshold and are clustered in the spatial distribution (e.g., consider points that are no more than a preset distance apart as clustered points in the spatial distribution, aggregate these points into one point, and use the coordinates of the point with the highest priority score as the coordinates of the aggregated point). If the number of location points after aggregation is greater than the number of monitoring sensors initially determined, increase the number of sensors according to the difference. For example, if the number of location points after aggregation is 22 and the number of monitoring sensors initially determined is 20, then add 2 sensors. If the number of location points after aggregation is less than the number of monitoring sensors initially determined, then deploy the extra sensors to the downwind area of ​​the three-dimensional monitoring space. For example, if the number of location points after aggregation is 16 and the number of monitoring sensors initially determined is 20, then deploy the extra 4 sensors to the downwind area. The downwind area of ​​the three-dimensional monitoring space can be determined based on the prevailing wind direction of the target chemical plant's location throughout the year.

[0030] 6. Calculate the current downwind area's sensor deployment density. If the preset target density is not reached, further increase the number of sensors in the downwind area to achieve the preset target density.

[0031] When adding sensors to the downwind area, all locations in the downwind area can be found from the locations that have never exceeded the above score threshold. Then, the corresponding number of locations can be selected. For example, in step 5, there are 30 points that have not exceeded the score threshold, of which 8 points are located in the downwind area. Then, the 4 points with the highest priority score are selected from the 8 points (assuming that 4 sensors need to be added in the downwind area).

[0032] S104. Extract key data from the diffusion distribution dataset. Key data includes unique identifiers for each leak source, wind field data, target gas concentration data and leak source strength at the location of each monitoring sensor, and wind field data including wind direction and wind speed.

[0033] S105. Train, validate, and test the neural network model using key data to form a source tracing model that takes wind field data and target gas concentration data as input data and unique identification information of each leakage source, as well as the corresponding leakage probability and leakage source strength, as the prediction results output.

[0034] Specifically, a deep neural network (DNN) model can be used to perform forward simulation, which involves constructing a functional relationship between the unique identifier of the leak source, the leak source strength and wind field data, and the target gas concentration data at the locations of each monitoring sensor. Through the back propagation process of the neural network, such as loss calculation, gradient calculation, and weight update, this functional relationship is optimized, ultimately forming a source tracing model that is compatible with the target chemical plant, target gas, and the deployment of monitoring sensors.

[0035] S106. During monitoring, multiple target gas concentration data and wind field data are collected in real time through multiple monitoring sensors and meteorological sensors. The multiple target gas concentration data and wind field data are input into the source tracing model to obtain the unique identification information of each leakage source and the corresponding leakage probability and leakage source strength.

[0036] To ensure the validity of the output results, the input data of the source tracing model can be preprocessed, such as by smoothing and noise reduction. Then, based on the input data filtering conditions, the input data of the source tracing model is filtered. Input data that does not meet the requirements, such as concentration data that is too low or has too small a change range, or data with insufficient completeness, is not allowed to be sent to the source tracing model or the source tracing model is not enabled. Only input data that meets the filtering conditions is sent to the source tracing model, ensuring that the source tracing model is enabled and the output results correspond to a real leak of the target gas as much as possible. Finally, after the source tracing model outputs the results, the output results that meet the result threshold are output externally. Assuming that the sum of the leakage probabilities of the two leak sources with the highest leakage probabilities in descending order in the output results is greater than the result threshold of 75%, the leakage probability and leakage source strength of these two leak sources are output. Finally, an alarm can be triggered for the corresponding leak source.

[0037] Based on the same concept, such as Figure 2 As shown, this embodiment of the invention also provides a real-time gas leak monitoring system, including multiple monitoring sensors 11, a meteorological sensor 12, a wireless transmission module 13, and a real-time monitoring platform 14.

[0038] The arrangement and functions of the multiple monitoring sensors 11 and meteorological sensors 12 are described above. Both are connected to the wireless transmission module 13, which is connected to the real-time monitoring platform 14.

[0039] The selection of the monitoring sensor 11 is determined based on a comprehensive evaluation of factors such as selectivity, sensitivity, resolution, detection limit, response time, stability, weather resistance, range, power consumption, and cost, and is highly compatible with the target chemical plant, target gas, and actual monitoring needs.

[0040] Considering that ethane is not the most representative characteristic gas in an ethylene cracking unit, the monitoring sensor 11 needs to have a certain degree of selectivity. If other online monitoring methods show that the typical environmental background value of ethane in the ethylene cracking unit is tens to hundreds of ppb, the detection limit of the monitoring sensor 11 needs to be the same as or lower, and the corresponding range of the monitoring sensor 11 will be limited to around tens of ppm. For monitoring timeliness considerations, the response time of the monitoring sensor 11 is generally on the order of seconds.

[0041] Also based on considerations of monitoring timeliness, the data acquisition frequency of meteorological sensor 12 is generally on the order of seconds.

[0042] The wireless transmission module 13 is used to wirelessly transmit the target gas concentration data and wind field data collected by the monitoring sensor 11 and the meteorological sensor 12 to the real-time monitoring platform 14.

[0043] The real-time monitoring platform 14 is equipped with a source tracing module and a source tracing model. The formation of the source tracing model is described above. The source tracing module is used to input the target gas concentration data and wind field data from the monitoring sensor 11 and the meteorological sensor 12 into the source tracing model to obtain the unique identification information of each leakage source and the corresponding leakage probability and leakage source strength.

[0044] As can be seen from the above, this invention uses three-dimensional modeling of the target chemical plant and fluid simulation of the target gas to obtain key data for training the source tracing model. The trained source tracing model enables real-time monitoring of gas leaks, which can adapt to the monitoring needs of different chemical plants and gas types, providing a flexible and comprehensive source tracing solution. It can provide real-time and accurate leak source location and leak intensity prediction, thereby supporting timely maintenance and handling, reducing the frequency of accidents, lowering environmental and health risks, providing enterprises with a scientific and efficient means of gas leak control, and improving the enterprise's environmental management level and social responsibility image.

[0045] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for real-time monitoring of gas leaks, characterized in that, include: A three-dimensional model of the target chemical plant is established using 3D modeling technology. Based on the typical leakage data of each leakage source of the target chemical plant collected in advance and the unique identification information assigned to each leakage source, the diffusion simulation parameters of the fluid simulation software are set, and according to the three-dimensional model, the fluid simulation software simulates the diffusion distribution of the gas plume in the three-dimensional monitoring space based on the physicochemical properties of the target gas and the diffusion simulation parameters. Different combinations of the diffusion simulation parameters are used to obtain the diffusion distribution dataset of the target gas, wherein the three-dimensional monitoring space is a three-dimensional space containing the target chemical plant. Multiple monitoring sensors are installed on the target chemical plant so that the target gas leaked from each leak source can be monitored, and meteorological sensors are installed at unobstructed heights in the three-dimensional monitoring space. Key data is extracted from the diffusion distribution dataset. The key data includes unique identification information of each leakage source, wind field data, and target gas concentration data at the location of each monitoring sensor. The wind field data includes wind direction and wind speed. The neural network model is trained, validated, and tested using the key data to form a source tracing model that takes wind field data and target gas concentration data as input data and unique identification information of each leakage source and corresponding leakage probability as the prediction result output. During monitoring, multiple target gas concentration data and wind field data are collected in real time through the multiple monitoring sensors and meteorological sensors. The multiple target gas concentration data and wind field data are input into the source tracing model to obtain the unique identification information of each leakage source and the corresponding leakage probability.

2. The method for real-time monitoring of gas leakage according to claim 1, characterized in that, The diffusion simulation parameters include the unique identifier of the simulated leak source, the source strength level of the simulated leak source intensity gradient, the simulated leak source temperature, the simulated wind direction, and the simulated wind speed. The number of combinations is the product of the number of simulated leak sources, the number of source strength levels, the number of simulated wind directions, and the number of simulated wind speeds. The simulated leak source temperature remains constant.

3. The method for real-time monitoring of gas leakage according to claim 2, characterized in that, The key data also includes the source strength of each leakage source, and the source tracing model also outputs the leakage source strength corresponding to the unique identification information of the leakage source.

4. The method for real-time monitoring of gas leaks according to claim 2, characterized in that, The provision of multiple monitoring sensors on the target chemical plant further includes: The monitoring range of a single monitoring sensor is determined based on the performance indicators of the monitoring sensor and the leakage source strength in the typical leakage data. By comparing the monitoring range of a single monitoring sensor with the size of the three-dimensional monitoring space, the required number of monitoring sensors can be preliminarily determined. Based on the required installation space of the monitoring sensor, extract the coordinates of all possible locations where the monitoring sensor can be installed from the 3D model; The target gas concentration data at each location point coordinate is extracted from the diffusion distribution dataset. The priority score of each location point coordinate is obtained by comprehensively evaluating the sum of the target gas concentration data at each location point coordinate, the peak value, the number of high concentration data exceeding the concentration threshold, and the installation convenience score. The coordinates of location points whose priority scores exceed the score threshold and are clustered in the spatial distribution are aggregated. If the number of location points after aggregation is greater than the number of monitoring sensors initially determined, the number of sensors is increased according to the difference. If the number of location points after aggregation is less than the number of monitoring sensors initially determined, the extra sensors are deployed to the downwind area of ​​the three-dimensional monitoring space. Calculate the current downwind area's sensor deployment density. If the preset target density is not reached, further increase the number of sensors in the downwind area to achieve the preset target density.

5. The method for real-time monitoring of gas leaks according to claim 1, characterized in that, The step of inputting the multiple target gas concentration data and wind field data into the source tracing model to obtain the unique identifier information of each leakage source and the corresponding leakage probability further includes: The input data of the source tracing model is preprocessed; The input data of the tracing model is filtered according to the input data filtering conditions, and only the input data that meets the filtering conditions is sent to the tracing model. After the source tracing model outputs its results, it will output the results that meet the result threshold.

6. The method for real-time monitoring of gas leaks according to claim 1, characterized in that, The target chemical plant is an ethylene cracking unit, and the target gas is ethane.

7. A real-time gas leak monitoring system, characterized in that, include: The gas leak real-time monitoring method according to any one of claims 1-6 includes multiple monitoring sensors, meteorological sensors, and a source tracing model; The wireless transmission module is used to wirelessly transmit the target gas concentration data and wind field data collected by the monitoring sensor and meteorological sensor to the real-time monitoring platform. A real-time monitoring platform is provided, which includes a source tracing module. The source tracing module is used to input target gas concentration data and wind field data from the monitoring sensor and meteorological sensor into the source tracing model to obtain unique identification information of each leakage source and the corresponding leakage probability.

Images