Leakage equipment rapid discrimination method based on gas detection system, electronic equipment and storage medium

Abstract

The invention provides a leakage equipment rapid discrimination method based on a gas detection system, electronic equipment and a storage medium, and relates to the technical field of gas leakage positioning in the petrochemical industry, and the method specifically comprises the following steps: constructing a typical leakage scene library and an in-service point type gas detector library of a petrochemical device; and constructing a three-dimensional model of the petrochemical device, carrying out gas leakage diffusion CFD three-dimensional simulation on all scenes in the typical leakage scene library, and recording detection results of the in-service point type gas detector on each leakage scene. And establishing a leakage discrimination Bayesian network model, and learning and calculating the conditional probability table of each node in the Bayesian network by using simulation result data of a typical leakage scene. And taking alarm data and a current wind field condition as leakage evidences, and inputting the leakage evidences into the constructed leakage discrimination Bayesian network for leakage probability calculation. According to the technical scheme, the problem that in the prior art, a rapid and accurate method for judging leakage equipment is lacked is solved.

Description

Technical Field

[0001] This invention relates to the field of gas leak location technology in the petrochemical industry, specifically to a rapid leak identification method, electronic equipment, and storage medium based on a gas detection system. Background Technology

[0002] Petrochemical companies typically handle large quantities of flammable and toxic hazardous gases, and usually use point-type gas detectors to monitor for leaks in real time. When a gas leak occurs, it is essential to be able to quickly identify the equipment where the leak might occur to guide a rapid on-site repair response. However, currently, there is a lack of methods for rapid and precise identification of leak locations down to the equipment level.

[0003] Petrochemical plants are equipped with gas detection systems to ensure safe production. These systems consist of numerous point-type gas detectors, each with different response patterns to leaks in different equipment, including the number and location of alarms. During the evaluation of the effectiveness of these gas detection systems, a large amount of data on gas cloud diffusion and detector alarm patterns has been generated. By deeply analyzing this data and constructing models of leaking equipment and gas detector alarm patterns, potential leaking equipment can be quickly identified.

[0004] Therefore, this paper proposes a rapid identification method, electronic equipment, and storage medium for leaking equipment based on a gas detection system. By using Bayesian inference methods to deeply explore the potential correlation between gas detector alarms and leak locations, a rapid screening and identification model for leak locations is constructed, providing petrochemical enterprises with a means to quickly locate and repair leaking equipment after a gas leak. Summary of the Invention

[0005] The main objective of this invention is to provide a method, electronic device, and storage medium for rapid identification of leaking devices based on a gas detection system, in order to solve the problem that there is a lack of a fast and accurate method for identifying leaking devices in the prior art.

[0006] To achieve the above objectives, the present invention provides a rapid identification method for leaking devices based on a gas detection system, specifically including the following steps:

[0007] S1, based on leakage sources and wind fields, constructs a library of typical leakage scenarios for petrochemical plants and a library of in-service point gas detectors.

[0008] S2. Construct a three-dimensional model of the petrochemical plant, conduct CFD three-dimensional simulations of gas leakage diffusion for all scenarios in the typical leakage scenario library, record the detection results of in-service point gas detectors for each leakage scenario, and perform data processing.

[0009] S3. Build a Bayesian network model for leakage detection. Use simulation results data from typical leakage scenarios to learn and calculate the conditional probability table of each node in the Bayesian network.

[0010] S4 uses the statistical alarm data and current wind field conditions as evidence of leakage, and inputs them into the constructed leakage discrimination Bayesian network to calculate the leakage probability.

[0011] Furthermore, step S1 specifically includes the following steps:

[0012] S1.1, through a comprehensive assessment of hazard identification (HAZID) studies, process analysis, hazard and operability analysis (HAZOP), a database of leaks from similar units, and historical leak data from the field, potential hazardous gas leak sources in petrochemical units, i.e. leaking equipment, are identified. Combined with a petrochemical unit reliability database, the frequency of leaks from each leak source is quantitatively calculated.

[0013] S1.2 Based on local atmospheric statistics, consult the wind rose diagram data released by the meteorological bureau, reasonably divide the wind field and determine the wind direction and wind speed parameters, and calculate the probability of occurrence of various wind field conditions.

[0014] S1.3, Randomly combine the parameters in the set of leakage sources and the set of wind fields to establish a typical leakage scenario library.

[0015] S1.4 Statistically analyze the parameters of in-service point gas detectors at petrochemical plant sites, including quantity, location, target gas, response time, and alarm threshold, and construct an in-service point gas detector library.

[0016] Furthermore, step S2 specifically includes the following steps:

[0017] S2.1, Construct a 1:1 three-dimensional model of the petrochemical plant by combining the plan layout diagram and equipment diagram. Set m in-service point gas detectors as m monitoring points at the corresponding positions in the three-dimensional model. Conduct CFD three-dimensional simulation of gas leakage diffusion for all scenarios in the typical leakage scenario library to obtain the steady-state gas cloud concentration distribution data of each leakage scenario. Calculate, analyze and statistically analyze the alarm response results of each in-service point gas detector to the leakage gas cloud of each scenario in the typical leakage scenario library.

[0018] S2.2, construct a gas detection response result database for the target petrochemical plant leakage scenario library, statistically analyze the alarm response of each in-service point gas detector in the gas leakage diffusion simulation of each scenario, and perform normalization, standardization and discretization preprocessing on the result data.

[0019] Furthermore, step S3 specifically includes the following steps:

[0020] S3.1 Construct a Bayesian network model for leakage discrimination based on leakage equipment in petrochemical plants using a gas detection system, and establish the correlation between nodes. Nodes include leakage sources, wind field conditions, and all in-service point gas detectors.

[0021] S3.2, determine the conditional probability table of each node in the Bayesian network model for rapid identification of leakage equipment in petrochemical plants, perform information mining and reasoning on the gas detection and analysis results after data processing, learn and calculate the conditional probability table of each node in the Bayesian network, and improve the Bayesian network.

[0022] Furthermore, step S4 specifically includes the following steps:

[0023] S4.1 When a leak occurs, quickly compile the alarm data of the in-service point gas detectors at the scene, including the alarm detector number and alarm level.

[0024] S4.2, the statistical alarm data and the current wind field conditions are used as evidence to input into the constructed leak discrimination Bayesian network for Bayesian inference calculation, to obtain the probability of each leak source leaking and sort them according to the probability.

[0025] S4.3, based on the leakage probability ranking of each leakage source, quickly carry out on-site investigation and repair work.

[0026] Further, step S1.1 specifically involves: identifying i potential hazardous gas leak sources from the petrochemical plant, where the set of leak sources is represented as L = {l1, l2, ..., l...} i}, where l i Representing the i-th leakage source; combining the petrochemical plant reliability database, quantitatively calculate the leakage occurrence frequency f(L)={f(l1),f(l2),…,f(l... i )}.

[0027] Further, step S1.2 specifically involves: collecting local wind direction and speed meteorological conditions, and dividing the wind field into spring, summer, autumn, and winter seasons based on the dominant wind direction and average wind speed of each season. The wind field is represented as W = [w j ] (j＝1,2,3,4) The probability of each monsoon field is represented as P(w j =0.25.

[0028] Further, step S1.3 specifically involves: combining the leak source and the wind field to obtain a typical leak scenario library S = {S_n} for a petrochemical plant, containing n leak scenarios. ij |n=i×j},S ij This represents a leakage scenario where the i-th leakage source is combined with the j-th wind field.

[0029] Further, step S1.4 specifically involves: identifying a total of m in-service point-type gas detectors in the petrochemical plant, with the detector set represented as D = {d1, d2, ..., dm}. m A library of in-service point gas detectors is constructed to record the location, target gas, and alarm threshold of each detector.

[0030] Further, step S2.2 specifically involves: constructing a gas detection response result matrix C = {c} for the target petrochemical unit leakage scenario library. ijk}, c ijk The result of the k-th in-service point gas detector on the i-th leak source under the j-th wind field condition is given. If the detector does not alarm, then c ijk =0, if the detector triggers a level one alarm then c ijk =1, if the detector triggers a level 2 alarm then c ijk =2.

[0031] Further, step S3.1 specifically involves: constructing a leak discrimination Bayesian network model based on the leak detection equipment of the petrochemical plant in the gas detection system, which includes one leak source node, one wind field condition node, and k in-service point gas detector nodes; wherein the leak source node contains i leak sources, the wind field condition node contains j wind field conditions; each in-service point gas detector node contains three node states, namely no alarm, first-level alarm, and second-level alarm.

[0032] Furthermore, step S3.2 specifically includes the following steps:

[0033] S3.2.1, the probability P(l) of each leakage source in the leakage source node is calculated using formula (1). i Perform the calculation:

[0034]

[0035] S3.2.2, Conditional probability table P(d) of other in-service point gas detectors detecting nodes that have not experienced leaks. state |Θ) Based on the gas detection response result matrix C, calculate using formula (2):

[0036]

[0037] Where: d state For the three states of the in-service point gas detector node, P(Θ) represents the joint probability distribution of the leakage source node and the wind field condition node, and P(d state P(Θ|d) represents the probability of the detector node occurring in one of the three states. state ) represents the joint probability distribution of the leakage source node and the wind field condition node under the three states.

[0038] Further, step S4.2 specifically involves: inputting the statistical alarm data and the current wind field conditions as evidence into the constructed leak discrimination Bayesian network, and using formula (3) to perform Bayesian inference calculations to obtain the probability P(l) of each leak source occurring. i |Φ,Wind) and sorted by probability:

[0039]

[0040] Where Φ represents the joint distribution of all in-service point gas detector nodes, Wind represents the input wind field conditions, and P(l i Y(l) represents the probability of leakage occurring from each leakage source i. i |Φ) represents the probability of leakage from leakage source i under the condition of the joint distribution Φ of in-service point gas detector nodes, P(l i |Wind) represents the probability of leakage source i causing leakage under the input wind field conditions.

[0041] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor. When the processor executes the program, it implements a method for rapid identification of leaking devices based on a gas detection system.

[0042] The present invention also provides a non-transient computer scale storage medium on which a computer program is stored. When the computer program is executed by a processor, it implements a method for rapid identification of leaking devices based on a gas detection system.

[0043] The present invention has the following beneficial effects:

[0044] Petrochemical plants typically have gas detection systems containing numerous point gas detectors for real-time monitoring of toxic or flammable gas leaks. However, due to the influence of parameters such as leak direction, wind direction, and wind speed, it is difficult to quickly and accurately determine the leak location based on alarm response data, thus hindering emergency response and maintenance. This invention proposes a rapid leak identification method based on gas detection systems. It aims to utilize the large amount of gas detector alarm data generated from leak coverage assessments of gas detection systems, employing Bayesian inference methods to deeply explore the correlation between these alarms and the leak location, constructing a rapid leak location identification model, thereby improving the efficiency and accuracy of equipment leak monitoring and response.

[0045] The method provided by this invention introduces the concept of probability. After the discrimination model is constructed, it can quickly perform fuzzy discrimination based on the alarm results of point gas detectors and give a probability ranking of leaking devices through probability statistics to guide on-site investigation.

[0046] Unlike commonly used leak location technologies that can only reach the device or unit level, this invention identifies leaks at the device level, thus improving the accuracy of leak location. Attached Figure Description

[0047] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings:

[0048] Figure 1 A flowchart of a rapid leak detection method based on a gas detection system according to the present invention is shown.

[0049] Figure 2 The leakage discrimination Bayesian network model established in Embodiment 2 of the present invention is shown.

[0050] Figure 3 The leakage discrimination Bayesian network model corresponding to alarm data 1 in Embodiment 2 of the present invention is shown.

[0051] Figure 4 The leakage discrimination Bayesian network model corresponding to alarm data 2 in Embodiment 2 of the present invention is shown.

[0052] Figure 5 The leakage discrimination Bayesian network model established in Embodiment 3 of the present invention is shown.

[0053] Figure 6 The leakage discrimination Bayesian network model corresponding to alarm data 1 in Embodiment 3 of the present invention is shown.

[0054] Figure 7 The leakage discrimination Bayesian network model corresponding to alarm data 2 in Embodiment 3 of the present invention is shown. Detailed Implementation

[0055] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] Example 1

[0057] like Figure 1 The method for rapid identification of leaking devices based on a gas detection system, as shown, specifically includes the following steps:

[0058] S1, based on leakage sources and wind fields, constructs a library of typical leakage scenarios for petrochemical plants and a library of in-service point gas detectors.

[0059] S2. Construct a three-dimensional model of the petrochemical plant, conduct CFD three-dimensional simulations of gas leakage diffusion for all scenarios in the typical leakage scenario library, record the detection results of in-service point gas detectors for each leakage scenario, and perform data processing.

[0060] S3. Build a Bayesian network model for leakage detection. Use simulation results data from typical leakage scenarios to learn and calculate the conditional probability table of each node in the Bayesian network.

[0061] S4 uses the statistical alarm data and current wind field conditions as evidence of leakage, and inputs them into the constructed leakage discrimination Bayesian network to calculate the leakage probability.

[0062] Petrochemical plants typically have gas detection systems containing numerous point gas detectors for real-time monitoring of toxic or flammable gas leaks. However, due to the influence of parameters such as leak direction, wind direction, and wind speed, it is difficult to quickly and accurately determine the leak location based on alarm response data, thus hindering emergency response and maintenance work. This invention proposes a rapid leak identification method based on gas detection systems. It aims to utilize the large amount of gas detector alarm data generated from leak coverage assessments of gas detection systems, employing Bayesian inference methods to deeply explore the correlation between these alarms and the leak location, constructing a rapid leak location identification model, thereby improving the efficiency and accuracy of equipment leak monitoring and response.

[0063] Specifically, step S1 includes the following steps:

[0064] S1.1 Identification of Leak Sources: Through comprehensive judgment based on Hazard Identification (HAZID) studies, process analysis, Hazard and Operability Analysis (HAZOP), a leak database of similar units, and historical leak data from the field, potential hazardous gas leak sources in petrochemical units, i.e. leaking equipment, are identified. Combined with the petrochemical unit reliability database, the leakage frequency of each leak source is quantitatively calculated.

[0065] S1.2, Establishing the wind field: The wind field mainly includes two parameters: wind direction and wind speed. Based on local atmospheric statistics, consult the wind rose diagram data published by the meteorological bureau, rationally divide the wind field and determine the wind direction and wind speed parameters, and calculate the probability of occurrence of various wind field conditions.

[0066] S1.3, Construct a leakage scenario library for petrochemical plants: The leakage source set and the wind field set are independent of each other, and the parameters in the leakage source set and the wind field set are randomly combined to establish a typical leakage scenario library.

[0067] S1.4, Construct an in-service point gas detector database: Collect statistics on the parameters of in-service point gas detectors at the petrochemical plant site, including quantity, location, target gas, response time, and alarm threshold, and construct an in-service point gas detector database.

[0068] Specifically, step S2 includes the following steps:

[0069] S2.1, Construct a 1:1 three-dimensional model of the petrochemical plant by combining the plan layout diagram and equipment diagram. Set m in-service point gas detectors as m monitoring points at the corresponding positions in the three-dimensional model. Conduct CFD three-dimensional simulation of gas leakage diffusion for all scenarios in the typical leakage scenario library to obtain the steady-state gas cloud concentration distribution data of each leakage scenario. Calculate, analyze and statistically analyze the alarm response results of each in-service point gas detector to the leakage gas cloud of each scenario in the typical leakage scenario library.

[0070] S2.2, construct a gas detection response result database for the target petrochemical plant leakage scenario library, statistically analyze the alarm response of each in-service point gas detector in the gas leakage diffusion simulation of each scenario, and perform normalization, standardization and discretization preprocessing on the result data.

[0071] Specifically, step S3 includes the following steps:

[0072] S3.1 Construct a Bayesian network model for leakage discrimination based on leakage equipment in petrochemical plants using a gas detection system, and establish the correlation between nodes. Nodes include leakage sources, wind field conditions, and all in-service point gas detectors.

[0073] S3.2, determine the conditional probability table of each node in the Bayesian network model for rapid identification of leakage equipment in petrochemical plants, perform information mining and reasoning on the gas detection and analysis results after data processing, learn and calculate the conditional probability table of each node in the Bayesian network, and improve the Bayesian network.

[0074] Specifically, step S4 includes the following steps:

[0075] S4.1 When a leak occurs, quickly compile the alarm data of the in-service point gas detectors at the scene, including the alarm detector number and alarm level.

[0076] S4.2, the statistical alarm data and the current wind field conditions are used as evidence to input into the constructed leak discrimination Bayesian network for Bayesian inference calculation, to obtain the probability of each leak source leaking and sort them according to the probability.

[0077] S4.3, based on the leakage probability ranking of each leakage source, quickly carry out on-site investigation and repair work.

[0078] Specifically, step S1.1 involves identifying i potential hazardous gas leak sources from the petrochemical plant, with the set of leak sources represented as L = {l1, l2, ..., l...}. i}, where l i Representing the i-th leakage source; combining the petrochemical plant reliability database, quantitatively calculate the leakage occurrence frequency f(L)={f(l1),f(l2),…,f(l... i )}.

[0079] Specifically, step S1.2 involves collecting local wind direction and speed meteorological conditions, dividing the wind field into four seasons (spring, summer, autumn, and winter) based on the prevailing wind direction and average wind speed of each season. The wind field is represented as W = [w j ] (j＝1,2,3,4) The probability of each monsoon field is represented as P(w j =0.25.

[0080] Specifically, step S1.3 involves: combining the leak source and the wind field to obtain a typical leak scenario library S = {S_n} for petrochemical plants, containing n leak scenarios. ij |n=i×j},S ij This represents a leakage scenario where the i-th leakage source is combined with the j-th wind field.

[0081] Specifically, step S1.4 involves identifying a total of m in-service point-type gas detectors in the petrochemical plant, with the detector set represented as D = {d1, d2, ..., dm}. m A library of in-service point gas detectors is constructed to record the location, target gas, and alarm threshold of each detector.

[0082] Specifically, step S2.2 involves constructing a gas detection response result matrix C = {c} for the target petrochemical unit leakage scenario library. ijk}, c ijk The result of the k-th in-service point gas detector on the i-th leak source under the j-th wind field condition is given. If the detector does not alarm, then c ijk =0, if the detector triggers a level one alarm then c ijk =1, if the detector triggers a level 2 alarm then c ijk =2.

[0083] Specifically, step S3.1 involves constructing a leak detection Bayesian network model based on the leak detection equipment of the petrochemical plant in the gas detection system. This model includes one leak source node, one wind field condition node, and k in-service point gas detector nodes. The leak source node contains i leak sources, and the wind field condition node contains j wind field conditions. Each in-service point gas detector node has three node states: no alarm, first-level alarm, and second-level alarm.

[0084] Specifically, step S3.2 includes the following steps:

[0085] S3.2.1, the probability P(l) of each leakage source in the leakage source node is calculated using formula (1). i Perform the calculation:

[0086]

[0087] S3.2.2, Conditional probability table p(d) of other in-service point gas detectors detecting nodes that have not experienced leaks. state |Θ) Based on the gas detection response result matrix C, calculate using formula (2):

[0088]

[0089] Where: d state For the three states of the in-service point gas detector node, p(Θ) is the joint probability distribution of the leakage source node and the wind field condition node, P(d state P(Θ|d) represents the probability of the detector node occurring in one of the three states. state ) represents the joint probability distribution of the leakage source node and the wind field condition node under the three states.

[0090] Specifically, step S4.2 involves inputting the statistical alarm data and the current wind field conditions as evidence into the constructed leak discrimination Bayesian network, and using formula (3) to perform Bayesian inference calculations to obtain the probability P(l) of each leak source occurring. i |Φ,Wind) and sorted by probability:

[0091]

[0092] Where Φ represents the joint distribution of all in-service point gas detector nodes, Wind represents the input wind field conditions, and P(l i Let P(l) represent the probability of leakage occurring from each leakage source i. i |Φ) represents the probability of leakage from leakage source i under the condition of the joint distribution Φ of in-service point gas detector nodes, P(l i |Wind) represents the probability of leakage source i causing leakage under the input wind field conditions.

[0093] This invention is mainly applied to the field of gas leak location in petrochemical industry equipment.

[0094] This invention fully utilizes the three-dimensional gas diffusion simulation data obtained from the coverage assessment of point gas detectors in petrochemical plants to construct a rapid gas leak location identification model. This facilitates on-site identification of potential leak locations based on alarm information from point gas detectors, enabling timely emergency response and maintenance. This method can be widely applied to various petrochemical plants equipped with point gas detectors and has promising application prospects.

[0095] Example 2

[0096] Taking a petrochemical plant as an example, the method provided by the present invention will be explained in detail.

[0097] Step 1: Through hazard identification analysis, nine potential hazardous leak sources were identified for the petrochemical unit, denoted as L={l1,l2,…,l9}, with hydrogen being the primary hazardous gas. Based on the petrochemical unit reliability database, the leakage frequency of each leak source was quantitatively calculated and listed in Table 1.

[0098] Consulting the wind rose diagrams and other atmospheric data released by the local meteorological bureau, the location of the petrochemical plant exhibits distinct winter and summer seasons, with significant differences in wind speed and direction. Therefore, it is classified into two wind fields: winter and summer, represented as W = [w j ] (j＝1,2) The probability of each monsoon field can be expressed as P(w j = 0.5.

[0099] Table 1. List of potential leakage sources for the example device in this embodiment.

[0100]

[0101] By combining the leakage source and the wind field, a device leakage scenario library S = {S} is obtained, containing n = 9 × 2 scenarios. ij |n=i×j=18}, for example S 21 This indicates a leakage scenario where the second leakage source is combined with the summer monsoon field.

[0102] The identification device has a total of m = 12 point-type gas detectors in service, and the detector set is represented as D = {d1, d2, ..., d...} 12 The gas detector database is constructed as shown in Table 2.

[0103] Table 2. Database of In-Service Point Gas Detectors for the Example Device in This Embodiment

[0104]

[0105] Step 2: Construct a 1:1 three-dimensional model of the case device, and set the in-service point gas detectors as 12 monitoring points at the corresponding positions in the three-dimensional model.

[0106] A CFD 3D simulation of gas leakage diffusion was conducted on 18 leakage scenarios in the device leakage scenario library. Considering the influence of leakage direction, six leakage directions (+X, -X, +Y, -Y, +Z, -Z) were calculated for each leakage scenario. The concentration data of the target gas of each detector after the gas cloud stabilized were statistically analyzed. At the same time, the alarm response results of each gas detector to each scenario in the library were calculated and statistically analyzed.

[0107] A gas detection response result matrix C = {c} is constructed based on a database of leakage scenarios in a target petrochemical plant. ijk}, matrix C can be expanded into Table 3 to display its elements.

[0108] Table 3. Matrix Elements of Gas Detection Response Results

[0109]

[0110]

[0111] Step 3: Construct a Bayesian network model for rapid leak detection in the case study device. This model includes one leak source node, one wind field condition node, and twelve point gas detector nodes. The leak source node has nine node states, each representing a leak source; the wind field condition node has two node states, each representing a wind field condition; and each point gas detector node has three node states: State 0 (no alarm), State 1 (level 1 alarm), and State 2 (level 2 alarm). The Bayesian network model for rapid leak detection in the case study device is as follows: Figure 2 As shown.

[0112] The probability P(l1) of each leakage source in the leakage source node is calculated using the normalization formula (1), and the results are listed in Table 1.

[0113]

[0114] The conditional probability tables for other detector nodes are calculated based on the gas detection response result matrix C using formula (2). Taking detector d1 as an example, its conditional probability table is shown in Table 4.

[0115]

[0116] Where: d state For the three states of the in-service point gas detector node, P(Θ) represents the joint probability distribution of the leakage source node and the wind field condition node, and P(d state P(Φ|d) represents the probability of the detector node occurring in one of the three states. state ) represents the joint probability distribution of the leakage source node and the wind field condition node under the three states;

[0117] Table 4 Conditional Probability Table of Detector d1 in Case Device

[0118]

[0119] Step 4: Randomly select historical leakage alarm data from the case device as evidence and input it into the constructed Bayesian network. Combine the Bayesian inference formula to calculate the probability of leakage from each leakage source.

[0120] (1) Alarm data 1: Summer monsoon field, detectors D3, D4, D7 and D8 triggered a level 1 alarm, and the leak source was found to be L4;

[0121] (2) Alarm data 2: Winter wind field, detector D8 triggered a level 1 alarm, and detectors D7, D9 and D10 triggered level 2 alarms. The leak source was found to be L5.

[0122] The statistical alarm data and current wind field conditions are input into the constructed leak discrimination Bayesian network as evidence. Bayesian inference calculation is performed using formula (3) to obtain the probability P(l) of each leak source occurring. i |Φ,Wind) and sorted by probability:

[0123]

[0124] Where Φ represents the joint distribution of all in-service point gas detector nodes, Wind represents the input wind field conditions, and P(l i Let P(l) represent the probability of leakage occurring from each leakage source i. i |Φ) represents the probability of leakage from leakage source i under the condition of the joint distribution Φ of in-service point gas detector nodes, P(l i |Wind) represents the probability of leakage source i causing leakage under the input wind field conditions.

[0125] The following results were obtained from Bayesian network inference analysis for rapid leakage device identification of the two leakage alarm data (i.e., alarm data 1 and alarm data 2) in this embodiment:

[0126] (1) Alarm Data 1: Evidence information is calculated with the corresponding state of the node set to 100%. After inference calculation, the leakage device quickly determines the posterior probability distribution of each node in the Bayesian network as follows: Figure 3 As shown, the order of on-site investigation based on leakage probability is: L4, L7, L8, L3, L5. The actual leaking device is L4, and the calculation result is valid.

[0127] (2) Alarm Data 2: Similarly, the leakage device, after inference and calculation, quickly determines the posterior probability distribution of each node in the Bayesian network as follows: Figure 4As shown, the order of on-site investigation based on leakage probability is: L8, L5, L7, L4, L6. The actual leaking device is L5, and the calculation result is valid.

[0128] Petrochemical plants typically have gas detection systems containing numerous point gas detectors for real-time monitoring of toxic or flammable gas leaks. However, due to the influence of parameters such as leak direction, wind direction, and wind speed, it is difficult to quickly and accurately determine the leak location based on alarm response data, thus hindering emergency response and maintenance. This invention proposes a rapid leak identification method based on gas detection systems. It aims to utilize the large amount of gas detector alarm data generated from leak coverage assessments of gas detection systems, employing Bayesian inference methods to deeply explore the correlation between these alarms and the leak location, constructing a rapid leak location identification model, thereby improving the efficiency and accuracy of equipment leak monitoring and response.

[0129] This invention is mainly applied to the field of gas leak location in petrochemical industry equipment.

[0130] This invention fully utilizes the three-dimensional gas diffusion simulation data obtained from the coverage assessment of point gas detectors in petrochemical plants to construct a rapid gas leak location identification model. This facilitates on-site identification of potential leak locations based on alarm information from point gas detectors, enabling timely emergency response and maintenance. This method can be widely applied to various petrochemical plants equipped with point gas detectors and has promising application prospects.

[0131] Example 3

[0132] Taking a gas station of a certain enterprise as an example, the method provided by the present invention will be explained in detail.

[0133] Step 1: Through hazard identification analysis, seven potential hazardous leak sources were identified for the petrochemical unit, denoted as L={l1,l2,…,l7}, with methane (CH4) being the primary hazardous gas. Based on the petrochemical unit reliability database, the leakage frequency of each leak source was quantitatively calculated and listed in Table 5.

[0134] Consulting the wind rose diagrams and other atmospheric data released by the local meteorological bureau, the wind field conditions were simplified into wind fields for spring, summer, autumn, and winter, corresponding to different wind directions and speeds, expressed as W = [w j ] (j＝1,2，3,4) The probability of each monsoon field can be expressed as P(w j =0.25.

[0135] Table 5. List of potential leak sources at the case gas station

[0136]

[0137]

[0138]

[0139] By combining the leakage source and the wind field, a device leakage scenario library S = {S} is obtained, containing n = 7 × 4 scenarios. ij |n=i×j=28}, for example S 13 This indicates a leakage scenario where the first leakage source is combined with the autumn wind field.

[0140] The identification device has a total of m = 9 point-type gas detectors in service, labeled D = {d1, d2, ..., d9}, and the gas detector database is shown in Table 6.

[0141] Table 6. Database of In-Service Gas Detectors for Case Study Gas Stations

[0142] Step 2: Construct a 1:1 3D model of the gas station in the case study, and set up 9 in-service point-type gas detectors as monitoring points at the corresponding positions in the 3D model.

[0143] A CFD 3D simulation of gas leakage diffusion was conducted on 28 leakage scenarios in the device leakage scenario library. Considering the influence of leakage direction, six leakage directions (+X, -X, +Y, -Y, +Z, -Z) were calculated for each leakage scenario. The concentration data of the target gas of each detector after the gas cloud stabilized were statistically analyzed. At the same time, the alarm response results of each gas detector to each scenario in the library were calculated and statistically analyzed.

[0144] A gas detection response result matrix C = {c} is constructed based on a database of leakage scenarios in a target petrochemical plant. ijk}, matrix C can be expanded into Table 7 to display its elements.

[0145] Table 7. Elements of the Gas Detection Response Matrix

[0146] Leak source Wind field direction <![CDATA[d1]]> <![CDATA[d2]]> … <![CDATA[d9]]> <![CDATA[l1]]> spring +X 0 0 … 1 <![CDATA[l1]]> spring -X 0 2 … 1 … … … … … … … <![CDATA[l2]]> spring +X 0 0 … 0 … … … … … … … <![CDATA[l7]]> winter -Z 0 1 … 0

[0147] Step 3: Construct a Bayesian network model for rapid leak detection at the gas station in the case study. This model includes one leak source node, one wind field condition node, and nine point gas detector nodes. The leak source node has seven states, each representing a leak source; the wind field condition node has four states, each representing a wind field condition; and each point gas detector node has three states: State 0 (no alarm), State 1 (Level 1 alarm), and State 2 (Level 2 alarm). The Bayesian network model for rapid leak detection at the case study device is as follows: Figure 5 As shown.

[0148] The probability of leakage at each leakage source node is calculated using the normalization formula (1), and the results are listed below.

[0149] In Table 1;

[0150]

[0151] The conditional probability tables for other detector nodes are calculated based on the gas detection response matrix C using formula (2). Taking detector d1 as an example, its conditional probability table is shown in Table 8.

[0152]

[0153] Where: d state For the three states of the in-service point gas detector node, P(Θ) represents the joint probability distribution of the leakage source node and the wind field condition node, and P(d state P(Θ|d) represents the probability of the detector node occurring in one of the three states. state ) represents the joint probability distribution of the leakage source node and the wind field condition node under the three states.

[0154] Table 8 Conditional Probability Table of Detector d1 in Case Study: Gas Station

[0155]

[0156] Step 4: Randomly select historical leak alarm data from the gas station in the case below and input it into the constructed Bayesian network as evidence information. Combine the Bayesian inference formula to calculate the probability of each leak source causing a leak.

[0157] (1) Alarm data 1: Autumn wind field, detectors D5 and D6 triggered a level one alarm, and the leak source was found to be L1;

[0158] (2) Alarm data 2: Winter wind field, detectors D1 and D3 triggered a level 1 alarm, and detector D2 triggered a level 2 alarm. The leak source was found to be L6.

[0159] The statistical alarm data and current wind field conditions are input into the constructed leak discrimination Bayesian network as evidence. Bayesian inference calculation is performed using formula (3) to obtain the probability P(l) of each leak source occurring. i |Φ,Wind) and sorted by probability:

[0160]

[0161] Where Φ represents the joint distribution of all in-service point gas detector nodes, Wind represents the input wind field conditions, and P(l i Let P(l) represent the probability of leakage occurring from each leakage source i. i |Φ) represents the probability of leakage from leakage source i under the condition of the joint distribution Φ of in-service point gas detector nodes, P(l i |Wind) represents the probability of leakage source i causing leakage under the input wind field conditions.

[0162] Leakage device rapid discrimination Bayesian network inference analysis was performed on the two leakage alarm data (i.e., alarm data 1 and alarm data 2) in this example case, and the results are as follows:

[0163] (1) Alarm Data 1: Evidence information is calculated with the corresponding state of the node set to 100%. After inference calculation, the leakage device quickly determines the posterior probability distribution of each node in the Bayesian network as follows: Figure 6 As shown, the order of on-site investigation based on leakage probability is: L2, L1, L4, L5. The actual leaking device is L1, and the calculation result is valid.

[0164] (2) Alarm Data 2: Similarly, the leakage device, after inference and calculation, quickly determines the posterior probability distribution of each node in the Bayesian network as follows: Figure 7 As shown, the order of on-site investigation based on leakage probability is: L6, L7, L5. The actual leaking device is L6, and the calculation result is valid.

[0165] Petrochemical plants typically have gas detection systems containing numerous point gas detectors for real-time monitoring of toxic or flammable gas leaks. However, due to the influence of parameters such as leak direction, wind direction, and wind speed, it is difficult to quickly and accurately determine the leak location based on alarm response data, thus hindering emergency response and maintenance. This invention proposes a rapid leak identification method based on gas detection systems. It aims to utilize the large amount of gas detector alarm data generated from leak coverage assessments of gas detection systems, employing Bayesian inference methods to deeply explore the correlation between these alarms and the leak location, constructing a rapid leak location identification model, thereby improving the efficiency and accuracy of equipment leak monitoring and response.

[0166] This invention is mainly applied to the field of gas leak location in petrochemical industry equipment.

[0167] This invention fully utilizes the three-dimensional gas diffusion simulation data obtained from the coverage assessment of point gas detectors in petrochemical plants to construct a rapid gas leak location identification model. This facilitates on-site identification of potential leak locations based on alarm information from point gas detectors, enabling timely emergency response and maintenance. This method can be widely applied to various petrochemical plants equipped with point gas detectors and has promising application prospects.

[0168] Example 4

[0169] Step 1: Through comprehensive assessment using hazard identification (HAZID) studies, process analysis, hazard and operability analysis (HAZOP), a database of leaks from similar units, and historical leak data from the site, identify i potential hazardous gas leak sources in the petrochemical unit, denoted as L = {l1, l2, ..., l...} iBased on the petrochemical plant reliability database, the leakage occurrence frequency {f(l1),f(l2),…,f(l)} of each leakage source is quantitatively calculated. i )}.

[0170] Collect local wind direction and speed meteorological conditions, and divide the wind field into four seasons (spring, summer, autumn, and winter) based on the prevailing wind direction and average wind speed of each season, represented as W = [w j ] (j＝1,2,3,4) The probability of each monsoon field can be expressed as P(w j =0.25.

[0171] By combining the leakage source and the wind field, a device leakage scenario library S = {S} is obtained, containing n scenarios. ij |n=i×j},S ij This represents a leakage scenario where the i-th leakage source is combined with the j-th wind field.

[0172] The identification device has a total of m point-type gas detectors in service, and the detector set is represented as D = {d1, d2, ..., d...} m A database is built to record the location of each detector, the target gas, and the alarm threshold.

[0173] Step 2: Construct a 1:1 three-dimensional model of the petrochemical plant, and set up m in-service point-type gas detectors as monitoring points at the corresponding positions in the three-dimensional model.

[0174] A CFD 3D simulation of gas leakage diffusion is performed on n leakage scenarios in the device leakage scenario library. The concentration data of the target gas of each detector after the gas cloud stabilizes are statistically analyzed. At the same time, the alarm response results of each gas detector to each scenario in the library are calculated and statistically analyzed.

[0175] A gas detection response result matrix C = {c} is constructed based on a database of leakage scenarios in a target petrochemical plant. ijk}, c ijk Let c be the detection result of the k-th point gas detector for the i-th leak source under the j-th wind field condition. If the detector does not alarm, then ijk =0, if the detector triggers a level one alarm, then c ijk =1, if the detector triggers a level 2 alarm, then c ijk =2.

[0176] Step 3: Construct a Bayesian network model structure for rapid identification of leaks in petrochemical plants, which includes one leak source node, one wind field condition node, and k point gas detector nodes. The leak source node has i node states, representing each leak source; the wind field condition node has j node states, representing each wind field condition; and each point gas detector node has three node states: no alarm, first-level alarm, and second-level alarm.

[0177] The probability of leakage occurring at each leakage source node is calculated using the normalization formula (1):

[0178]

[0179] The conditional probability tables for other detector nodes are calculated based on the gas detection response matrix C using formula (2):

[0180]

[0181] Where: d state For the three states of the in-service point gas detector node, P(Θ) represents the joint probability distribution of the leakage source node and the wind field condition node, and P(d state P(Θ|d) represents the probability of the detector node occurring in one of the three states. state ) represents the joint probability distribution of the leakage source node and the wind field condition node under the three states.

[0182] Step 4: When an actual leak occurs, quickly collect statistics on the current wind field conditions and alarm data of in-service point gas detectors, including detector numbers and alarm levels. Input the statistical parameters as evidence into the constructed Bayesian network and use formula (3) to perform Bayesian inference calculations to obtain the probability of each leak source and sort them according to the probability.

[0183]

[0184] Where: Φ represents the joint distribution of all in-service point gas detector nodes, Wind represents the input wind field conditions, and P(l i Let P(l) represent the probability of leakage occurring from each leakage source i. i |Φ) represents the probability of leakage from leakage source i under the condition of the joint distribution Φ of in-service point gas detector nodes, P(l i |Wind) represents the probability of leakage source i causing leakage under the input wind field conditions.

[0185] Based on the probability ranking of each leakage source obtained from formula (3), on-site investigation and repair work can be carried out quickly.

[0186] Petrochemical plants typically have gas detection systems containing numerous point gas detectors for real-time monitoring of toxic or flammable gas leaks. However, due to the influence of parameters such as leak direction, wind direction, and wind speed, it is difficult to quickly and accurately determine the leak location based on alarm response data, thus hindering emergency response and maintenance. This invention proposes a rapid leak identification method based on gas detection systems. It aims to utilize the large amount of gas detector alarm data generated from leak coverage assessments of gas detection systems, employing Bayesian inference methods to deeply explore the correlation between these alarms and the leak location, constructing a rapid leak location identification model, thereby improving the efficiency and accuracy of equipment leak monitoring and response.

[0187] This invention is mainly applied to the field of gas leak location in petrochemical industry equipment.

[0188] This invention fully utilizes the three-dimensional gas diffusion simulation data obtained from the coverage assessment of point gas detectors in petrochemical plants to construct a rapid gas leak location identification model. This facilitates on-site identification of potential leak locations based on alarm information from point gas detectors, enabling timely emergency response and maintenance. This method can be widely applied to various petrochemical plants equipped with point gas detectors and has promising application prospects.

[0189] Example 5

[0190] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor. When the processor executes the program, it implements a method for rapid identification of leaking devices based on a gas detection system.

[0191] Petrochemical plants typically have gas detection systems containing numerous point gas detectors for real-time monitoring of toxic or flammable gas leaks. However, due to the influence of parameters such as leak direction, wind direction, and wind speed, it is difficult to quickly and accurately determine the leak location based on alarm response data, thus hindering emergency response and maintenance. This invention proposes a rapid leak identification method based on gas detection systems. It aims to utilize the large amount of gas detector alarm data generated from leak coverage assessments of gas detection systems, employing Bayesian inference methods to deeply explore the correlation between these alarms and the leak location, constructing a rapid leak location identification model, thereby improving the efficiency and accuracy of equipment leak monitoring and response.

[0192] This invention is mainly applied to the field of gas leak location in petrochemical industry equipment.

[0193] This invention fully utilizes the three-dimensional gas diffusion simulation data obtained from the coverage assessment of point gas detectors in petrochemical plants to construct a rapid gas leak location identification model. This facilitates on-site identification of potential leak locations based on alarm information from point gas detectors, enabling timely emergency response and maintenance. This method can be widely applied to various petrochemical plants equipped with point gas detectors and has promising application prospects.

[0194] Example 6

[0195] The present invention also provides a non-transient computer scale storage medium on which a computer program is stored. When the computer program is executed by a processor, it implements a method for rapid identification of leaking devices based on a gas detection system.

[0196] Petrochemical plants typically have gas detection systems containing numerous point gas detectors for real-time monitoring of toxic or flammable gas leaks. However, due to the influence of parameters such as leak direction, wind direction, and wind speed, it is difficult to quickly and accurately determine the leak location based on alarm response data, thus hindering emergency response and maintenance. This invention proposes a rapid leak identification method based on gas detection systems. It aims to utilize the large amount of gas detector alarm data generated from leak coverage assessments of gas detection systems, employing Bayesian inference methods to deeply explore the correlation between these alarms and the leak location, constructing a rapid leak location identification model, thereby improving the efficiency and accuracy of equipment leak monitoring and response.

[0197] This invention is mainly applied to the field of gas leak location in petrochemical industry equipment.

[0198] This invention fully utilizes the three-dimensional gas diffusion simulation data obtained from the coverage assessment of point gas detectors in petrochemical plants to construct a rapid gas leak location identification model. This facilitates on-site identification of potential leak locations based on alarm information from point gas detectors, enabling timely emergency response and maintenance. This method can be widely applied to various petrochemical plants equipped with point gas detectors and has promising application prospects.

[0199] Of course, the above description is not intended to limit the present invention, and the present invention is not limited to the examples given above. Any changes, modifications, additions or substitutions made by those skilled in the art within the scope of the present invention should also fall within the protection scope of the present invention.

Claims

1. A method for rapid identification of leaking equipment based on a gas detection system, characterized in that, Specifically, the steps include the following: S1, based on the leakage source and wind field, construct a typical leakage scenario library for petrochemical plants and an in-service point gas detector library; S2, construct a three-dimensional model of the petrochemical plant, conduct gas leakage diffusion CFD three-dimensional simulation of all scenarios in the typical leakage scenario library, record the detection results of in-service point gas detectors for each leakage scenario and perform data processing. S3. Build a Bayesian network model for leakage detection. Use simulation results data from typical leakage scenarios to learn and calculate the conditional probability table of each node in the Bayesian network. S4 uses the statistical alarm data and current wind field conditions as evidence of leakage, and inputs them into the constructed leakage discrimination Bayesian network to calculate the leakage probability.

2. The rapid identification method for leaking equipment based on a gas detection system according to claim 1, characterized in that, Step S1 specifically includes the following steps: S1.1, through a comprehensive judgment of hazard identification (HAZID) studies, process analysis, hazard and operability analysis (HAZOP), leakage database of similar units, and historical leakage data from the field, potential hazardous gas leakage sources in petrochemical units, i.e. leakage equipment, are identified. Combined with the petrochemical unit reliability database, the leakage frequency of each leakage source is quantitatively calculated. S1.2 Based on local atmospheric statistics, consult the wind rose diagram data released by the meteorological bureau, reasonably divide the wind field and determine the wind direction and wind speed parameters, and calculate the probability of occurrence of various wind field conditions; S1.3, Randomly combine the parameters in the set of leakage sources and the set of wind fields to establish a typical leakage scenario library; S1.4 Statistically analyze the parameters of in-service point gas detectors at petrochemical plant sites, including quantity, location, target gas, response time, and alarm threshold, and construct an in-service point gas detector library.

3. The rapid identification method for leaking equipment based on a gas detection system according to claim 1, characterized in that, Step S2 specifically includes the following steps: S2.1, Combine the plan layout diagram and equipment diagram of the petrochemical plant to construct a 1:1 three-dimensional model of the petrochemical plant. Set m in-service point gas detectors as m monitoring points at the corresponding positions of the three-dimensional model. Perform gas leakage diffusion CFD three-dimensional simulation on all scenarios in the typical leakage scenario library to obtain the steady-state gas cloud concentration distribution data of each leakage scenario. Calculate, analyze and statistically analyze the alarm response results of each in-service point gas detector to the leakage gas cloud of each scenario in the typical leakage scenario library. S2.2, construct a gas detection response result database for the target petrochemical plant leakage scenario library, statistically analyze the alarm response of each in-service point gas detector in the gas leakage diffusion simulation of each scenario, and perform normalization, standardization and discretization preprocessing on the result data.

4. The rapid identification method for leaking equipment based on a gas detection system according to claim 1, characterized in that, Step S3 specifically includes the following steps: S3.1 Construct a Bayesian network model for leakage discrimination based on leakage equipment of petrochemical plants in the gas detection system, and establish the correlation between each node. The nodes include the leakage source, wind field conditions and all in-service point gas detectors. S3.2, determine the conditional probability table of each node in the Bayesian network model for rapid identification of leakage equipment in petrochemical plants, perform information mining and reasoning on the gas detection analysis results after data processing, learn and calculate the conditional probability table of each node in the Bayesian network, and improve the Bayesian network.

5. The rapid identification method for leaking equipment based on a gas detection system according to claim 1, characterized in that, Step S4 specifically includes the following steps: S4.1 When a leak occurs, quickly compile the alarm data of the in-service point gas detectors at the scene, including the alarm detector number and alarm level; S4.2, the statistical alarm data and the current wind field conditions are used as evidence to input into the constructed leak discrimination Bayesian network for Bayesian inference calculation, to obtain the probability of each leak source leaking and sort them according to the probability. S4.3, based on the leakage probability ranking of each leakage source, quickly carry out on-site investigation and repair work.

6. The rapid identification method for leaking equipment based on a gas detection system according to claim 2, characterized in that, Step S1.1 specifically involves identifying i potential hazardous gas leak sources from the petrochemical plant, with the set of leak sources represented as L = {l1, l2, ..., l...}. i }, where l i Representing the i-th leakage source; combining the petrochemical plant reliability database, quantitatively calculate the leakage occurrence frequency f(L)={f(l1),f(l2),…,f(l... i )}.

7. The rapid identification method for leaking equipment based on a gas detection system according to claim 2, characterized in that, Step S1.2 specifically involves: collecting local wind direction and speed meteorological conditions, and dividing the wind field into spring, summer, autumn, and winter seasons based on the prevailing wind direction and average wind speed of each season. The wind field is represented as W = [w j ] (j＝1,2,3,4) The probability of each monsoon field is represented as P(w j =0.

25.

8. The rapid identification method for leaking equipment based on a gas detection system according to claim 2, characterized in that, Step S1.3 specifically involves: combining the leak source and the wind field to obtain a typical leak scenario library S = {S_n} for petrochemical plants, containing n leak scenarios. ij |n=i×j},S ij This represents a leakage scenario where the i-th leakage source is combined with the j-th wind field.

9. A rapid identification method for leaking equipment based on a gas detection system according to claim 2, characterized in that, Step S1.4 specifically involves identifying a total of m in-service point-type gas detectors in the petrochemical plant, with the detector set represented as D = {d1, d2, ..., dm}. m A library of in-service point gas detectors is constructed to record the location, target gas, and alarm threshold of each detector.

10. A rapid identification method for leaking equipment based on a gas detection system according to claim 3, characterized in that, Step S2.2 specifically involves: constructing a gas detection response result matrix C = {c} for the target petrochemical unit leakage scenario library. ijk }, c ijk The result of the k-th in-service point gas detector on the i-th leak source under the j-th wind field condition is given. If the detector does not alarm, then c ijk =0, if the detector triggers a level one alarm then c ijk =1, if the detector triggers a level 2 alarm then c ijk =2.

11. A rapid identification method for leaking equipment based on a gas detection system according to claim 4, characterized in that, Step S3.1 specifically involves: constructing a Bayesian network model for leakage discrimination based on the leakage equipment of the petrochemical plant in the gas detection system, which includes one leakage source node, one wind field condition node, and k in-service point gas detector nodes; wherein the leakage source node contains i leakage sources, the wind field condition node contains j wind field conditions; each in-service point gas detector node contains three node states, namely no alarm, first-level alarm, and second-level alarm.

12. The rapid identification method for leaking equipment based on a gas detection system according to claim 4, characterized in that, Step S3.2 specifically includes the following steps: S3.2.1, the probability P(l) of each leakage source in the leakage source node is calculated using formula (1). i Perform the calculation: S3.2.2, Conditional probability table P(d) of other in-service point gas detectors detecting nodes that have not experienced leaks. state |Θ) Based on the gas detection response result matrix C, calculate using formula (2): Where: d state For the three states of the in-service point gas detector node, P(Θ) represents the joint probability distribution of the leakage source node and the wind field condition node, and P(d state P(Θ|d) represents the probability of the detector node occurring in one of the three states. state ) represents the joint probability distribution of the leakage source node and the wind field condition node under the three states.

13. The rapid identification method for leaking equipment based on a gas detection system according to claim 5, characterized in that, Step S4.2 specifically involves inputting the statistical alarm data and current wind field conditions as evidence into the constructed leak discrimination Bayesian network, and using formula (3) to perform Bayesian inference calculations to obtain the probability P(l) of each leak source occurring. i |Φ,Wind) and sorted by probability: Where Φ represents the joint distribution of all in-service point gas detector nodes, Wind represents the input wind field conditions, and P(l i Let P(l) represent the probability of leakage occurring from each leakage source i. i |Φ) represents the probability of leakage from leakage source i under the condition of the joint distribution Φ of in-service point gas detector nodes, P(l i |Wind) represents the probability of leakage source i leaking under the input wind field conditions.

14. An electronic device comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that, When the processor executes the program, it implements the rapid identification method for leaking devices based on a gas detection system as described in any one of claims 1-13.

15. A non-transitory computer scale storage medium, wherein a computer program is stored thereon, characterized in that, When the computer program is executed by the processor, it implements the rapid identification method for leaking devices based on a gas detection system as described in any one of claims 1-13.

Images