CN117034740B - Method and system for positioning combustible gas leakage source and predicting leakage rate in tunnel - Google Patents

Method and system for positioning combustible gas leakage source and predicting leakage rate in tunnel Download PDF

Info

Publication number
CN117034740B
CN117034740B CN202310838809.0A CN202310838809A CN117034740B CN 117034740 B CN117034740 B CN 117034740B CN 202310838809 A CN202310838809 A CN 202310838809A CN 117034740 B CN117034740 B CN 117034740B
Authority
CN
China
Prior art keywords
leakage
data
tunnel
time sequence
combustible gas
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202310838809.0A
Other languages
Chinese (zh)
Other versions
CN117034740A (en
Inventor
吕山
黄小美
阳东
彭世尼
漆琦
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Chongqing University
Original Assignee
Chongqing University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Chongqing University filed Critical Chongqing University
Priority to CN202310838809.0A priority Critical patent/CN117034740B/en
Publication of CN117034740A publication Critical patent/CN117034740A/en
Application granted granted Critical
Publication of CN117034740B publication Critical patent/CN117034740B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M3/00—Investigating fluid-tightness of structures
    • G01M3/02—Investigating fluid-tightness of structures by using fluid or vacuum
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M3/00—Investigating fluid-tightness of structures
    • G01M3/02—Investigating fluid-tightness of structures by using fluid or vacuum
    • G01M3/04—Investigating fluid-tightness of structures by using fluid or vacuum by detecting the presence of fluid at the leakage point
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
    • G01N33/0004—Gaseous mixtures, e.g. polluted air
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
    • G01N33/0004—Gaseous mixtures, e.g. polluted air
    • G01N33/0009—General constructional details of gas analysers, e.g. portable test equipment
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
    • G01N33/22—Fuels; Explosives
    • G01N33/225—Gaseous fuels, e.g. natural gas
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01P—MEASURING LINEAR OR ANGULAR SPEED, ACCELERATION, DECELERATION, OR SHOCK; INDICATING PRESENCE, ABSENCE, OR DIRECTION, OF MOVEMENT
    • G01P5/00—Measuring speed of fluids, e.g. of air stream; Measuring speed of bodies relative to fluids, e.g. of ship, of aircraft
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/0464—Convolutional networks [CNN, ConvNet]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/10—Services
    • G06Q50/26—Government or public services

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Chemical & Material Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Business, Economics & Management (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Computational Linguistics (AREA)
  • Mathematical Physics (AREA)
  • Computing Systems (AREA)
  • Molecular Biology (AREA)
  • Data Mining & Analysis (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Economics (AREA)
  • Medicinal Chemistry (AREA)
  • Food Science & Technology (AREA)
  • Pathology (AREA)
  • Human Resources & Organizations (AREA)
  • Immunology (AREA)
  • Strategic Management (AREA)
  • Biochemistry (AREA)
  • Analytical Chemistry (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Marketing (AREA)
  • Development Economics (AREA)
  • Combustion & Propulsion (AREA)
  • Game Theory and Decision Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Computer Hardware Design (AREA)
  • Chemical Kinetics & Catalysis (AREA)
  • General Chemical & Material Sciences (AREA)
  • Oil, Petroleum & Natural Gas (AREA)

Abstract

The invention discloses a method and a system for positioning a combustible gas leakage source and predicting leakage rate in a tunnel, which are used for obtaining combustible gas concentration time sequence data and wind speed time sequence data and recording leakage source coordinates, sensor coordinates, gas leakage starting time and leakage direction in an experiment; simulating to obtain time sequence concentration field and time sequence wind field data; extracting combustible gas concentration time sequence and wind speed time sequence data of the positions of the sensors from time sequence concentration field and time sequence wind field data, and recording leakage source coordinates, coordinates of the sensors, gas leakage starting time and leakage direction; establishing a deep learning prediction model through the data; inputting the actual tunnel combustible gas concentration time sequence and wind speed time sequence data into a deep learning prediction model to obtain the prediction results of the combustible gas leakage source positioning, the leakage rate, the leakage starting time and the leakage direction. The invention can realize rapid and accurate prediction and provide guiding basis for fire-fighting operation, emergency rescue and accident ventilation in highway tunnels and urban pipe galleries.

Description

Method and system for positioning combustible gas leakage source and predicting leakage rate in tunnel
Technical Field
The invention relates to a method for positioning a combustible gas leakage source and predicting the leakage rate in a tunnel, and also relates to a system using the method for positioning the combustible gas leakage source and predicting the leakage rate in the tunnel.
Background
At present, highway tunnels and urban pipe galleries are quite common in cities, and both bring great convenience to citizens' life and economic development. However, the accident of the highway tunnel and the occurrence of leakage of the gas pipeline in the pipeline corridor occur, and because the highway tunnel and the urban pipeline corridor have the characteristic of limited space, the gas is difficult to dilute rapidly, so that the gas is easy to cause serious economic property loss and casualties due to leakage of the gas in the facilities. The gas leakage alarm device has the advantages that the gas leakage alarm device can quickly alarm gas leakage in the highway tunnel and the urban pipe gallery, quickly and accurately position the gas leakage points, determine the gas leakage rate and the starting time of the gas leakage, and can provide powerful guidance for emergency response of accidents, personnel evacuation and execution of fire-fighting work.
The existing inversion modes of leakage sources in highway tunnels and urban pipe galleries mainly comprise the following steps:
1. the movable robot with the gas concentration sensor is arranged in the tunnel (pipe gallery), the moving direction of the robot is controlled through an algorithm according to the gas concentration change detected by the sensor, and finally the robot is guided to reach the leakage source, so that the positioning of the leakage source is completed.
2. By arranging a certain number of gas concentration sensors in a tunnel (pipe gallery) and transmitting time sequence gas concentration data acquired by the sensors into a computer, a program algorithm carried in the computer judges the position and the intensity of a leakage source according to the input data.
The program algorithm is mainly a method for searching a leakage source based on a forward leakage diffusion model, a method based on function regression fitting or a method based on a positioning optimization algorithm and the like, and the defects of the methods are as follows: ⑴ The prediction accuracy of the leakage source position and the leakage rate is poor; ⑵ Prediction of the leak start time cannot be considered; ⑶ The system cost is high.
Disclosure of Invention
The first object of the invention is to provide a method for positioning and predicting the leakage rate of a combustible gas leakage source in a tunnel, which can quickly and accurately obtain the position coordinates of the combustible gas leakage source, the leakage rate value, the gas leakage starting time and the leakage direction.
The first object of the invention is achieved by the following technical measures: the method for positioning and predicting the leakage rate of the combustible gas leakage source in the tunnel is characterized by comprising the following steps of:
S1, arranging a plurality of gas concentration sensors and air velocity sensors in an experimental tunnel, acquiring combustible gas concentration time sequence data at the positions of the gas concentration sensors and wind velocity time sequence data at the positions of the wind velocity sensors under different leakage diffusion conditions and tunnel environment conditions, establishing a coordinate system in the experimental tunnel, and recording the coordinates of a leakage source, the coordinates of the sensors, the gas leakage starting time and the leakage direction in the experiment, wherein the data form a data set;
S2, constructing a tunnel simulation model, arranging a plurality of gas concentration sensors and wind speed sensors in a simulation tunnel, simulating to obtain time sequence concentration field data and time sequence wind field data in the tunnel under different leakage diffusion conditions and tunnel environment conditions according to set key leakage parameters and tunnel environment parameters, establishing a coordinate system in the simulation tunnel, extracting the time sequence data and wind speed time sequence data of the combustible gas concentration at the coordinate positions of each sensor from the time sequence concentration field data and the time sequence wind field data according to the set coordinates of the gas concentration sensors and the wind speed sensors in the simulation tunnel, and recording the coordinates of a leakage source, the coordinates of each sensor, the starting time of gas leakage and the leakage direction, wherein the data form a data set;
s3, processing the data set into a training set, a verification set and a test set for training a prediction model, and training, verifying and testing the prediction model to obtain a deep learning prediction model;
S4, arranging a plurality of gas concentration sensors and air speed sensors in a tunnel in which the leakage of the combustible gas occurs to be predicted, inputting the acquired time sequence data of the concentration of the combustible gas and the acquired time sequence data of the air speed into a deep learning prediction model, and obtaining prediction results of positioning, leakage rate, leakage starting time and leakage direction of a plurality of groups of combustible gas leakage sources;
S5, synthesizing all groups of prediction results to obtain final prediction results of the positioning of the combustible gas leakage source, the leakage rate, the leakage starting time and the leakage direction.
The invention constructs a deep learning prediction model, which takes wind speed time sequence data, combustible gas concentration time sequence data and position coordinate information of corresponding sensors at a plurality of sensor setting positions as input and takes leakage source positions, leakage source strong leakage rate (leakage intensity and leakage rate), leakage starting time and leakage direction as output. The collected data are input into a computer carrying a deep learning prediction model, and when the gas concentration in the tunnel is detected to exceed the alarm lower limit value, the prediction of the leakage source position, the leakage flow rate and the leakage intensity is started to be executed, so that the rapid and accurate prediction can be realized. The prediction result of the invention can provide guiding basis for fire-fighting operation, emergency rescue and accident ventilation in highway tunnels and urban pipe galleries, and can be widely applied to the safe operation field of highway tunnels and urban pipe galleries.
In the step S2, one sensor is selected from a plurality of sensors and is used as a reference sensor, the relative coordinates between the coordinates of the leakage source and the reference sensor are calculated, and the positional relationship between the reference sensor and the leakage source is used as the basis for positioning the leakage source in the prediction result.
In the step S2, the gas concentration sensor and the air velocity sensor are translated for several times along the length direction of the simulated tunnel, so as to obtain the wind speed time sequence data and the combustible gas concentration time sequence data acquired by the current coordinate sensor after each translation, and the relative coordinates between the leakage source and the reference sensor.
In the simulated field data, the method for recording the time sequence data of the speed and the concentration recorded by the sensor after the translation and the position relation between the reference sensor and the leakage source after the translation by translating the preset sensor position coordinates is realized, so that the data enhancement is realized, and the data obtained by the sensor when the leakage source and the sensor group (the reference sensor) have different relative positions is obtained. The method can avoid performing a large amount of simulation by taking the position of the leakage source as a variable in order to obtain sensor data of the leakage source at different positions in the tunnel in a numerical simulation stage.
In the steps S1 and S2, a gas concentration sensor with concentration higher than a preset gas concentration alarm lower limit and the moment at the moment are detected for the first time in wind speed time sequence data and combustible gas concentration time sequence data acquired by the sensor, and the difference value between the moment and the gas leakage starting moment is the leakage starting time predicted by the model.
According to the method, after gas leakage occurs, an o-group sensor group is determined according to a first sensor I which is detected in a tunnel and exceeds a concentration alarm lower limit Clow value, wind speed data and concentration data which are monitored by the o-group sensor group at the same time are input into a deep learning prediction model in an o-group mode, o-group prediction results are obtained, and prediction of the final leakage source position, leakage rate and leakage starting time of the model is determined by combining the o-group prediction results; the dividing mode of the o group sensor is as follows: the gas concentration sensor l is included, and the relative coordinates of all the sensors meet the relative coordinate relation of the sensor group required by the deep learning model, namely the requirement in the step S2.
In the step S4, the sensors in the tunnel are deployed according to the relative coordinates and the distances of the sensors set for simulating the tunnel in the step S2, so that the sensors deployed in the tunnel meet the relative coordinates of n groups of sensors which are continuous, and the relative coordinates and the distances of the n groups of sensors set for the model in the step S2; or the relative coordinates of the n groups of sensors after m sensor deployment positions are spaced apart satisfy the relative coordinates and the distances of the n groups of sensors set for the model in step S2.
In the step S3, the data set is the data set in the step S1 or/and the data set in the step S2.
In the steps S1 and S2, a coordinate system is built by taking the elevation of the tunnel ground as a reference, and the origin of coordinates of the coordinate system is a combustible gas leakage source on the ground or above the ground.
The deep learning prediction model is composed of a multi-layer convolutional neural network, a full-connection network, a long-term and short-term memory network and a full-connection network, wind speed time sequence data, combustible gas concentration time sequence data and coordinates of each sensor after a data set is processed are used as input layers, and a leakage source position, a gas leakage flow rate, a leakage starting time and a leakage direction corresponding to a sample of the input layers are used as output layers to construct the deep learning prediction model.
The second object of the present invention is to provide a system using the method for positioning the leakage source and predicting the leakage rate of the combustible gas in the tunnel.
The second object of the present invention is achieved by the following technical measures: a system for using the method for positioning and predicting leakage rate of combustible gas leakage sources in tunnels, comprising:
The data acquisition module is used for acquiring the concentration time sequence data of the combustible gas, the wind speed time sequence data and the coordinate information of each sensor in real time;
The prediction module is used for inputting the data into a deep learning prediction model to obtain prediction results of the positioning of the combustible gas leakage source, the leakage rate, the leakage starting time and the leakage direction;
And the output module is used for outputting the prediction result.
The data acquisition module comprises a data experiment unit and a data simulation unit, wherein the data experiment unit comprises a plurality of gas concentration sensors, a plurality of wind speed sensors and a host, the host consists of a central processor, a coordinate system construction module, an alarm module and a display module, wherein the coordinate system construction module, the alarm module and the display module are respectively connected with the central processor, and the gas concentration sensors and the wind speed sensors are respectively connected with the host; the data simulation unit is used for obtaining a data set by simulating a calculation fluid simulation method according to the simulated tunnel geometric parameters and the simulated accident working condition parameters.
Compared with the prior art, the invention has the following remarkable effects:
⑴ The invention constructs a deep learning prediction model, which takes wind speed time sequence data, combustible gas concentration time sequence data and position coordinate information of corresponding sensors at a plurality of sensor setting positions as input and takes leakage source positions, leakage source strong leakage rate (leakage intensity and leakage rate), leakage starting time and leakage direction as output. The collected data are input into a computer carrying a preset deep learning prediction model, and when the gas concentration in the tunnel is detected to exceed the alarm lower limit value, the prediction of the leakage source position, the leakage flow rate, the leakage intensity and the leakage direction is started to be performed, so that the rapid and accurate prediction can be realized.
⑵ According to the invention, a training data set is provided for the tunnel by performing experiments in advance or adopting a computational fluid dynamics mode, and translational sampling is performed through sensor coordinates in field data obtained in an analog simulation model, so that data enhancement is realized, excessive numerical simulation processes are avoided, the use of computational resources is reduced, and meanwhile, rapid deployment of the model can be realized.
⑶ The prediction result of the invention can provide guiding basis for fire-fighting operation, emergency rescue and accident ventilation in highway tunnels and urban pipe galleries, and application occasions include but are not limited to highway tunnels, urban pipe galleries, underground pipe ditches and long and narrow spaces.
⑷ The gas applicable to the invention is not limited to fuel gas, but is also applicable to other toxic and harmful gases.
Drawings
The invention will now be described in further detail with reference to the drawings and to specific examples.
FIG. 1 is a block flow diagram of an embodiment of the present invention;
FIG. 2 is a block diagram of a two-dimensional simulated tunnel in accordance with an embodiment of the invention;
FIG. 3 is a schematic diagram of a two-dimensional simulation result of gas leakage diffusion in a tunnel according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of the distribution of model training sensors within a tunnel in accordance with an embodiment of the present invention;
FIG. 5 is a schematic diagram of a model training gas concentration sensor data acquisition process according to an embodiment of the present invention;
FIG. 6 is a schematic diagram of a model training wind speed sensor data acquisition process according to an embodiment of the present invention;
FIG. 7 is a schematic diagram of a deep learning prediction model according to an embodiment of the present invention;
FIG. 8 is a schematic diagram of sensor deployment within a tunnel under test in accordance with an embodiment of the present invention;
FIG. 9 is a block diagram of a data acquisition and prediction process in accordance with an embodiment of the present invention.
Detailed Description
The present invention will be described in detail with reference to the following examples and the accompanying drawings to help those skilled in the art to better understand the inventive concept of the present invention, but the scope of the claims of the present invention is not limited to the following examples, and it is within the scope of the present invention to those skilled in the art that all other examples obtained without making creative efforts are included in the scope of the present invention.
The invention discloses a method for positioning a combustible gas leakage source and predicting the leakage rate in a tunnel, which comprises the following steps:
S1, arranging a plurality of gas concentration sensors and air velocity sensors in an experimental tunnel, acquiring combustible gas concentration time sequence data at the positions of the gas concentration sensors and wind velocity time sequence data at the positions of the wind velocity sensors under different leakage diffusion conditions and tunnel environment conditions, establishing a coordinate system in the experimental tunnel, and recording the coordinates of a leakage source, the coordinates of the sensors, the gas leakage starting time and the leakage direction in the experiment, wherein the data form a data set;
The specific process is as follows:
1.1 Through an equal-proportion tunnel experiment or a shrinkage-size tunnel experiment, key leakage parameters (leakage rate, leakage point position and leakage direction) and tunnel environment parameters (wind fields in the tunnel) are controlled, and time sequence data of combustible gas concentration and wind speed at deployment positions of the sensor groups under different leakage diffusion conditions and tunnel environment conditions are acquired by deploying the gas concentration sensor and the wind speed sensor group in the experimental tunnel. Constructing a coordinate system in the tunnel, recording coordinates of the leakage points and the sensors in the experiment, and recording the moment when the fuel gas starts to leak;
Specific:
1.1.1 Constructing an x-y-z three-dimensional coordinate system in the tunnel by taking the elevation of the bottom of the tunnel as a reference, selecting the length direction of the tunnel as an x axis, the width direction as a y axis and the height direction as a z axis, selecting the central position of the tunnel along the width direction to be positioned on the ground, and positioning the central position and the leakage point on the same position in the length direction, wherein the positions are defined as coordinate origins (xs=0, ys=0, zs=0);
1.1.2 Setting and controlling different combustible gas leakage flow rates vgas according to the actual conditions of the gas leakage accidents in the tunnel;
1.1.3 Setting and controlling different combustible gas leakage directions;
1.1.4 Setting and controlling different tunnel inlet wind speeds vair according to the wind speed range in the tunnel;
1.1.5 Disposing a gas concentration sensor group and a wind speed sensor group in the tunnel, and acquiring three-dimensional coordinate values [ (x 1, y1, z 1), (x 2, y2, z 2), … …, (xn, yn, zn) ] of all the sensors;
1.1.6 And (3) carrying out a gas leakage experiment according to the set tunnel inlet wind speed parameter and the leakage source parameter, and carrying out experiments of j1 working conditions altogether. Recording the fuel gas leakage starting time trelease, and recording the time sequence data of the concentration and the wind speed of the fuel gas in the tunnel through a sensor group;
1.1.7 Searching a gas concentration sensor L with the concentration higher than a preset gas concentration alarm lower limit Clow detected for the first time in a gas concentration sensor group and the moment tdetect at the moment in the wind speed time sequence data and the gas concentration time sequence data samples acquired by the j1 group sensor group. The difference between the time tdetect and the fuel gas leakage start time trelease is the leakage start time t= tdetect-trelease predicted by the model. The input time sequence length of the model is tlength, and for the wind speed time sequence data and the combustible gas concentration time sequence data acquired by each group of sensor groups, the time sequence data serving as a data set is extracted by taking tdetect as a starting time and tdetect + tlength as a termination time.
S2, constructing a tunnel simulation model, arranging a plurality of gas concentration sensors and wind speed sensors in a simulation tunnel, simulating to obtain time sequence concentration field data and time sequence wind field data in the tunnel under different leakage diffusion conditions and tunnel environment conditions according to set key leakage parameters and tunnel environment parameters, establishing a coordinate system in the simulation tunnel, extracting the time sequence data and wind speed time sequence data of the combustible gas concentration at the coordinate positions of each sensor from the time sequence concentration field data and the time sequence wind field data according to the set coordinates of the gas concentration sensors and the wind speed sensors in the simulation tunnel, and recording the coordinates of a leakage source, the coordinates of each sensor, the starting time of gas leakage and the leakage direction, wherein the data form a data set;
The specific process is as follows:
1.2 A three-dimensional or two-dimensional tunnel simulation model is constructed by a computational fluid dynamics simulation method, and the time sequence concentration field and time sequence wind field data in the tunnel under different leakage diffusion conditions and tunnel environment conditions are obtained through simulation according to set key leakage parameters (leakage rate, leakage point positions and leakage directions) and tunnel environment parameters (wind fields in the tunnel). And constructing a coordinate system in the simulated tunnel, and extracting the time sequence data of the concentration and the wind speed of the combustible gas at the deployment coordinate position of the sensor group according to the coordinates of the set n gas concentrations and wind speed sensor groups in the tunnel from the time sequence concentration field and the time sequence wind field data in the tunnel obtained by the model. The coordinates of the leakage point and each sensor in the experiment [ (x 1, y1, z 1), (x 2, y2, z 2), … …, (xn, yn, zn) ] are recorded, and the moment trelease when the gas starts to leak is recorded;
Specific:
1.2.1 Constructing a three-dimensional or two-dimensional tunnel simulation model by a computational fluid dynamics simulation method;
1.2.2 For a three-dimensional tunnel model, constructing an x-y-z three-dimensional coordinate system in the tunnel by taking the elevation of the bottom of the tunnel of the model as a reference, selecting the length direction of the tunnel as an x axis, the width direction as a y axis, the height direction as a z axis, selecting the central position of the tunnel along the width direction, locating on the ground, and locating at the same position with a leakage point in the length direction, wherein the position is defined as a coordinate origin (xs=0, ys=0, zs=0);
1.2.3 For a two-dimensional tunnel model, constructing an x-z two-dimensional coordinate system in the tunnel by taking the elevation of the bottom of the tunnel of the model as a reference, selecting the length direction of the tunnel as an x axis, selecting the height direction as a z axis, selecting the central position of the tunnel along the width direction, positioning on the ground, and positioning at the same position with a leakage point in the length direction, wherein the position is defined as a coordinate origin (xs=0, zs=0);
1.2.4 Setting and controlling different combustible gas leakage flow rates vgas according to the actual conditions of the gas leakage accidents in the tunnel;
1.2.5 Setting and controlling different combustible gas leakage directions;
1.2.6 Setting and controlling different tunnel inlet wind speeds vair according to the wind speed range in the tunnel;
1.2.7 According to the set tunnel inlet wind speed parameter, the leakage source parameter simulates the gas leakage process in the tunnel, and j2 working conditions are simulated altogether. Recording the fuel gas leakage starting time trelease, and recording time sequence data of a fuel gas concentration field and a wind speed field in a tunnel;
1.2.8 N gas concentration sensor groups and wind speed sensor groups are initially deployed in the tunnel, and the coordinates of the n gas concentration sensor groups and the wind speed sensor groups are [ (x 1, y1, z 1) 0, (x 2, y2, z 2) 0, … …, (xn, yn, zn) 0]. And extracting values of the coordinate points at all moments on the wind field and concentration field data in the tunnel under j2 working conditions obtained in the numerical simulation calculation, so that the wind speed time sequence data or the flammable gas concentration time sequence data acquired by the sensors at the corresponding coordinates can be obtained. Meanwhile, selecting a sensor from the sensor group as a reference sensor k, wherein the coordinates of the sensor are (xk, yk, zk) 0, and calculating the relative coordinates (xk, yk, zk) 0- (0, 0) = (xk, yk, zk) 0 between the coordinates of the leakage source and the reference sensor;
1.2.9 Taking Deltax as a translation distance of the sensor group in the tunnel length direction (x direction), translating the coordinate of the sensor group to [ (x1+Deltax, y1, z 1) 1, (x2+Deltax, y2, z 2) 1, … …, (xn+Deltax, yn, zn) 1], acquiring wind speed time sequence data or combustible gas concentration time sequence data acquired by the sensor under the sensor group coordinate in j2 working conditions according to the method of the step 1.2.8), and obtaining relative coordinates (xk+Deltax, yk, zk) 0- (0, 0) = (xk+Deltax, yk, zk) 0 between the leakage source and the reference sensor k. Translating the coordinates of the sensor group for m times, repeating the operation of the step 1.2.8), and finally obtaining the wind speed time sequence data and the combustible gas concentration time sequence data acquired by the (m+1) x j2 group sensor group;
1.2.10 Searching a gas concentration sensor l with the concentration higher than a preset gas concentration alarm lower limit Clow detected for the first time in a gas concentration sensor group and the moment tdetect at the moment in the time in wind speed time sequence data and combustible gas concentration time sequence data samples acquired by the (m+1) x j2 group sensor group. The difference between the time tdetect and the fuel gas leakage start time trelease is the leakage start time t= tdetect-trelease predicted by the model. The input time sequence length of the model is tlength, and for the wind speed time sequence data and the combustible gas concentration time sequence data acquired by each group of sensor groups, the time sequence data serving as a data set is extracted by taking tdetect as a starting time and tdetect + tlength as a termination time.
S3, processing the data set into a training set, a verification set and a test set for training a prediction model, and training, verifying and testing the prediction model to obtain a deep learning prediction model;
The specific process is as follows:
1.3 The acquired time sequence data of the concentration and the wind speed of the combustible gas on the deployment position of the sensor group, and the recorded leakage source coordinates, sensor coordinates, leakage starting time and leakage direction are processed into a training set and a testing set for training a combustible gas leakage source positioning and leakage rate prediction model in the tunnel.
The data obtained in steps 1.1.7) and 1.2.10) (the data obtained in the two steps can be used independently or simultaneously) are collated. The wind speed time sequence data and the combustible gas concentration time sequence data acquired by the sensor group and the coordinate information of each sensor are used as input data of a model, and the relative coordinates (namely the position of a leakage source) between a leakage source corresponding to a sample and a reference sensor k, the gas leakage flow speed vgas, the leakage starting time t= tdetect-leakage and the leakage direction are used as output data of the model. And the data set is proportionally divided into a training set, a verification set and a test set according to working conditions or samples.
The construction and training process of the deep learning prediction model is as follows:
2.1 Taking the sensor group wind speed time sequence data and the combustible gas concentration time sequence data which are finally obtained in the step S3 as an input layer, and taking a leakage source position, a gas leakage flow rate, a leakage starting time and a leakage direction which correspond to an input layer sample as an output layer to construct a deep learning prediction model;
2.2 The deep learning prediction model is composed of a multi-layer convolutional neural network, a full-connection network, a long-term and short-term memory network and a full-connection network, and finally outputs prediction results of a leakage source position, a gas leakage flow rate and a leakage starting time, wherein the model structure is not fixed to be a combination of a certain super parameter and a model framework;
2.3 Preprocessing samples of the training set, the verification set and the test set, including but not limited to data merging, outlier processing and normalization processing;
2.4 Inputting the preprocessed training set sample into a preset prediction model for training to obtain a trained deep learning prediction model.
S4, arranging a plurality of gas concentration sensors and air speed sensors in a tunnel in which the leakage of the combustible gas occurs to be predicted, inputting the acquired time sequence data of the concentration of the combustible gas and the acquired time sequence data of the air speed into a deep learning prediction model, and obtaining prediction results of positioning, leakage rate, leakage starting time and leakage direction of a plurality of groups of combustible gas leakage sources; the specific process is as follows:
The concentration and wind speed data acquisition process in the tunnel is as follows:
3.1 Disposing m concentration and wind speed sensor groups in the tunnel according to the relative coordinates [ (x 1, y1, z 1), (x 2, y2, z 2), … …, (xn, yn, zn) ] and the spacing of the n concentration and wind speed sensor groups disposed in the step S2, wherein m is greater than or equal to n, so that the n concentration and wind speed sensor groups disposed in the step 1.2) are satisfied by the n sensor disposition positions (or the sensor disposition positions with certain intervals) in the tunnel, and the relative coordinates [ (x 1, y1, z 1), (x 2, y2, z 2), … …, (xn, yn, zn) ] and the spacing requirement are satisfied;
3.2 After the sensor deployment is completed, recording the concentration time sequence data and the wind speed time sequence data of the combustible gas in the tunnel in real time;
the determination process of the position of the leakage source, the leakage rate, the leakage starting time and the leakage direction in the tunnel is as follows:
4.1 When gas leakage accidents occur in the tunnel, a certain gas concentration sensor l in the tunnel detects that the gas concentration is larger than the gas concentration alarm lower limit Clow value at the time tdetect. Determining an o-group sensor group which comprises a gas concentration sensor and of which all sensor relative coordinates meet the sensor group relative coordinate relation required by a deep learning prediction model;
4.2 The time sequence data of the concentration of the combustible gas and the time sequence data of the wind speed, which are acquired from the time point tdetect to the time point tdetect + tlength of the o-group sensor group determined in the step 4.1), are input into a trained deep learning prediction model for o times, and the deep learning prediction model obtains the prediction results of the o-group on the position of the leakage source, the leakage rate, the leakage starting time and the leakage direction corresponding to the o-group input.
S5, synthesizing all groups of prediction results, and obtaining final prediction results of the combustible gas leakage source positioning, the leakage rate, the leakage starting time and the leakage direction by adopting a mode including and not limited to averaging.
A system using the method for positioning and predicting leakage rate of combustible gas leakage sources in tunnels, comprising: the data acquisition module is used for acquiring the concentration time sequence data of the combustible gas, the wind speed time sequence data and the coordinate information of each sensor in real time; the prediction module is used for inputting the data into a deep learning prediction model to obtain prediction results of the positioning of the combustible gas leakage source, the leakage rate, the leakage starting time and the leakage direction; and the output module is used for outputting the prediction result.
The data acquisition module comprises a data experiment unit and a data simulation unit, the data experiment unit comprises a plurality of gas concentration sensors, a plurality of wind speed sensors and a host, the host consists of a central processor, a coordinate system construction module, an alarm module and a display module, the coordinate system construction module, the alarm module and the display module are respectively connected with the central processor, and the gas concentration sensors and the wind speed sensors are respectively connected with the host; the data simulation unit is used for obtaining a data set by simulating a calculation fluid simulation method according to the simulated tunnel geometric parameters and the simulated accident working condition parameters.
The sensor capable of collecting the gas concentration is a plurality of gas concentration sensors which can be connected with a host machine through an optical cable (laser sensor) or a cable (electrochemical sensor) according to categories, and the installation position coordinates of the sensors are obtained; the wind speed sensors are a plurality of wind speed sensors which can be connected with the host computer through cables, and the installation position coordinates of the sensors are obtained; under special conditions, the wind speed sensor can be not deployed, and only the concentration sensor can be deployed.
The prediction module comprises: a processor comprising computer program instructions which, when executed by the processor, are adapted to carry out the steps corresponding to the method for positioning a combustible gas leakage source, leakage rate and leakage start time in a tunnel described above. A computer readable storage medium having stored thereon computer program instructions which, when executed by a processor, are adapted to carry out the steps corresponding to the method for positioning a combustible gas leakage source, leakage rate leakage start time and leakage orientation within a tunnel described above.
Specific examples:
As shown in fig. 1, the method for positioning a leakage source and predicting a leakage rate of combustible gas in a tunnel according to the embodiment specifically includes the following steps:
s1, acquiring a data set for model training by means of experiments, fluid numerical simulation and the like, wherein the data set comprises the following specific steps:
1.1 Data for training are obtained by adopting a fluid numerical simulation mode, and two-dimensional tunnel geometric simulation is constructed by adopting computational fluid dynamics software, as shown in figure 2. The tunnel is 300m long and 6m high, and the tunnel left side is tunnel air inlet, and the tunnel right side is tunnel air outlet. The gas leakage port is positioned at the center of the bottom of the tunnel, namely 150m away from the tunnel inlet, and the width of the gas leakage port is 0.1m;
1.2 Tunnel air inlet is a velocity inlet boundary condition, air flow rate is 0.2m/s, and the component is air. The tunnel air outlet is a pressure outlet boundary condition and the pressure is atmospheric. The gas leakage port is a boundary condition of a speed inlet, and the component is methane;
1.3 An x-z coordinate system is built in the tunnel, the position of the leakage source is taken as an original point (0, 0), x represents the direction along the length of the tunnel and is forward towards the air outlet of the tunnel, z represents the height direction of the tunnel and is forward towards the top of the tunnel. For the leak orientation, 0 ° forward in the x-direction, 90 ° forward in the z-axis;
1.4 Starting from 0.2m/s at the gas leakage rate, gradually increasing by 0.2m/s up to 20m/s. The gas leakage direction is 90 degrees, 67.5 degrees, 45 degrees and 22.5 degrees, and the total direction is 4 directions. 400 simulation working conditions are set in total, and the leakage time of fuel gas is trelease=0s. Transient simulation of gas leakage was performed for 200s, and the calculation results were stored every 0.5 s. And finally, 400 time sequence speed fields and methane concentration field data in the tunnel gas leakage 200s under the 400 working conditions can be obtained. The concentration distribution of methane with x marked in the range of-30 m to 120m in the tunnel under one working condition is shown in figure 3;
1.5 The installation positions of the wind speed sensor and the methane concentration sensor are preset in the tunnel model, and the coordinate parameters of the sensors are determined. Sensor coordinates (-30 m,5.8 m), (-10 m,5.8 m), (10 m,5.8 m) and (30 m,5.8 m) were deployed at 4 positions, and as shown in fig. 4, there is a single concentration sensor at the 4 positions, and there is a speed sensor and a concentration sensor at the same time. And reading the time sequence data of the wind speed and the gas concentration at the coordinates of the sensor from the time sequence data of the wind field and the concentration field in the tunnel under 100 working conditions obtained by calculation, as shown in fig. 5 and 6. Selecting the sensor (-30 m,5.8 m) at the 1 st position as a reference sensor, wherein the relative coordinates between the leakage source and the reference sensor are (-30 m,5.8 m);
1.6 Δx=0.1m as a sensor group in the tunnel length direction
The translation distance in the (x-direction) is 500 translations of the sensor set, i.e. the sensor coordinates are stepwise translated from (-30 m,5.8 m), (-10 m,5.8 m), (10 m,5.8 m) and (30 m,5.8 m) to (20 m,5.8 m), (40 m,5.8 m), (60 m,5.8 m) and (80 m,5.8 m). Recording the speed and concentration time sequence data recorded by the sensor and the relative coordinates between the leakage source and the reference sensor in the 500 translations for each working condition to obtain 200000 groups of sample data;
1.7 Searching for a sensor l with methane concentration higher than a preset methane concentration alarm lower limit value Clow=0.005 detected for the first time in 200000 groups of sample data, and recording the moment tdetect at the moment as the initial moment of inputting the sensor sample data into a model. And tdetect as a leak start time of the model output corresponding to the sample;
1.8 For each sample, extracting time series data from tdetect up to tdetect +15s as an input of the model;
1.9 200000 sets of sample data are collated, wherein the speed and concentration time series data and sensor coordinates acquired by each sensor from tdetect to tdetect +15s in the sample form an input tensor of the model, and the input tensor has the dimension of [200000,30,4,3 ]. Wherein 200000 represents the number of samples; 30 represents a time sequence length, and the acquisition frequency of time sequence data of 15s is 2Hz;4 represents a sensor at 4; and3 represents a data channel of each sensor, 1 channel is the x coordinate of the sensor, 2 channel is the speed data acquired by the sensor, 3 channel is the methane concentration data acquired by the sensor, and when the sensor at a certain position is only a speed sensor or a concentration sensor, the return value of the channel corresponding to the sensor which is not provided at the position is 0.
1.10 200000 Sets of sample data are consolidated, where the relative coordinates between the sample's leak source and reference sensor, the leak point leak rate, and the leak start time tdetect are taken as the model's output tensor, with dimensions of [200000,4 ]. Wherein 200000 represents the number of samples; 4 represents 4 output results of each sample, namely the relative coordinates between the leakage source and the reference sensor, the leakage speed of the leakage point, the leakage starting time tdetect and the leakage direction;
1.11 200000 groups of sample data are divided into 80% of training set, 10% of verification set and 10% of test set according to working conditions.
S2, constructing and training a prediction model of the leakage source positioning, the leakage rate and the leakage starting time of the combustible gas in the tunnel, wherein the method specifically comprises the following steps:
2.1 The prediction model of the combustible gas leakage source positioning, the leakage rate and the leakage starting time in the tunnel is composed of a preset multi-layer neural network deep learning model, input data are tensors with the dimension of [30,4,3], and the tensors respectively represent time sequence length, sensor and sensor channel data; the output data is tensor with the dimension of [4], which respectively represents the relative coordinates between the leakage source and the reference sensor, the leakage speed of the leakage point, the leakage starting time tdetect and the leakage direction;
2.2 The deep learning model mainly comprises a convolutional neural network, a fully-connected network, a long-term and short-term memory network and a fully-connected network, and fig. 7 is a detailed model structure diagram of the deep learning prediction model. The model structure, the model depth and the super parameters can be adjusted according to actual data and service conditions by only being used as a model framework of the display two displays of the embodiment;
2.3 Before the training set, the testing set and the verification set are input into the model for training and prediction, abnormal value processing and data normalization processing are carried out;
2.4 The input tensor with the dimension of [30,4,3] in the model is firstly subjected to a two-layer convolutional neural network, and the characteristics of the position of a single sensor are extracted, wherein the characteristic dimension is; then passing through a fully connected neural network and a Dropout layer; then the characteristic time sequence data enters a long-short-term memory network (LSTM) layer to extract time sequence characteristics; the output results of the final long short term memory network (LSTM) layer are respectively sent to four groups of full-connection networks with 4 layers, and the prediction output of the relative coordinates, the leakage speed, the leakage starting time tdetect and the leakage direction of the reference sensors, which are respectively corresponding to the models, is carried out; finally, the output results are combined into tensors with the dimension of [4] to be used as output data of the model;
2.5 All convolutional neural networks in the model are initialized by adopting a tanh activation function and He;
2.6 The last output layer activation function in the three groups of fully connected networks at the end of the model is Sigmoid, limiting the output result to between 0 and 1. The other fully connected networks all adopt a tanh activation function;
2.7 Training the model by adopting an Adam optimizer and adopting an MSE loss function;
2.8 Inputting the training set into a preset deep learning model, and executing 500 rounds of training to obtain a prediction model of combustible gas leakage source positioning, leakage rate and leakage starting time in the tunnel after completing training;
2.9 For the prediction result of the model after training, the final prediction result of the leakage source position, the leakage speed and the leakage starting time can be obtained through inverse normalization processing.
S3, deploying sensors and systems in the tunnel, wherein the deployment of the sensors and the systems in the tunnel specifically comprises the following steps:
3.1 Fig. 8 shows the deployment of sensors within a two-dimensional tunnel. Firstly, an x-z coordinate system is established in a tunnel, the bottom of the entrance of the tunnel is taken as the origin (0 m,0 m), and the tunnel length is 150m;
3.2 8 gas concentration sensors and wind speed sensors are deployed in the tunnel, and the deployment coordinates of the sensors are (0 m,5.8 m), (20 m,5.8 m), (40 m,5.8 m), (60 m,5.8 m), (80 m,5.8 m), (100, 5.8 m), (120 m,5.8 m), (140 m,5.8 m). The deployment coordinates are such that any adjacent 4 sensor deployment positions satisfy the sensor relative coordinate relationship preset in step 1.5) (-30 m,5.8 m), (-10 m,5.8 m), (10 m,5.8 m) and (30 m,5.8 m), i.e. each sensor deployment point is spaced 20m apart in the x-direction. For example, sensor deployment points 1,2,3,4 and sensor deployment points 2,3,4,5 all satisfy the sensor deployment relative coordinate relationship described above;
3.3 Associated sensor data reading devices, sensor data recording devices, as well as sensor data processing terminals and computer terminals executing predictive algorithms, along with sensor deployment;
s4, acquiring concentration and wind speed time sequence data acquired by a specific sensor in a tunnel to be predicted for gas leakage, wherein the concentration and wind speed time sequence data are specifically as follows: after the sensor, the related equipment and the computer terminal are deployed, the gas concentration and the wind speed information in the tunnel are monitored in real time.
S5: and the concentration and wind speed time sequence data acquired by the specific sensor are input into a model to obtain a plurality of groups of prediction results. Synthesizing a plurality of groups of prediction results to obtain the position coordinates of the combustible gas leakage source, the leakage rate value, the gas leakage starting time and the leakage direction prediction results, wherein the method specifically comprises the following steps:
5.1 Fig. 9 shows the data acquisition and prediction process. When gas leakage occurs in the tunnel, the gas diffuses in the tunnel and is detected by a first methane concentration sensor, the first sensor detecting that the methane concentration is greater than a set concentration lower limit value Clow=0.005 is marked as a sensor a;
5.2 Dividing the sensor groups which comprise the sensor a and meet the relative coordinate relation of the sensors preset in the step 1.5), and dividing 4 groups of sensor groups;
5.3 Continuously collecting 15s concentration data and wind speed data from the moment that the sensor a detects that the methane concentration is greater than the set concentration lower limit value;
5.4 15s concentration data and wind speed time sequence data acquired by the 4 groups of sensors divided in the step 5.2) and coordinate information of the sensors are formed into tensor data with the dimension of [4,30,4,3], and the tensor data are input into a deep learning prediction model trained in the step 2.8) after being normalized;
5.5 For the input data of the 4 groups of sensors, the model predicts 4 groups of prediction results of the position, the leakage speed, the leakage starting time and the leakage direction of the leakage points, averages the 4 groups of prediction results and then carries out inverse normalization processing, so that the position, the leakage speed and the leakage starting time of the leakage points predicted by the system according to the input of the sensors can be obtained.
The invention discloses a prediction result display of a method for predicting the leakage source, the leakage rate, the leakage starting time and the leakage direction of combustible gas in a tunnel, which comprises the following steps:
1. Input data from 30 time steps of sensors at 4 in one sample of the validation set:
(Table 1)
(Table 2)
2: Carrying out normalization processing on input data, and inputting the input data into a trained prediction model of the leakage source location, the leakage rate and the leakage starting time of the combustible gas in the tunnel to obtain a prediction result;
3: performing inverse normalization processing on the predicted result to obtain the following predicted result: leakage source position 18.753m, leakage rate: 16.932m/s, leakage start time: -3.536s, leakage angle: 88.733 DEG; actual leakage source conditions: leakage source location: 18.5m, leak rate 16.8m/s, leak start time: -3.5s, leakage angle: 90 deg..

Claims (8)

1. The method for positioning and predicting the leakage rate of the combustible gas leakage source in the tunnel is characterized by comprising the following steps of:
S1, arranging a plurality of gas concentration sensors and air velocity sensors in an experimental tunnel, acquiring combustible gas concentration time sequence data at the positions of the gas concentration sensors and wind velocity time sequence data at the positions of the wind velocity sensors under different leakage diffusion conditions and tunnel environment conditions, establishing a coordinate system in the experimental tunnel, and recording the coordinates of a leakage source, the coordinates of the sensors, the gas leakage starting time and the leakage direction in the experiment, wherein the data form a data set;
S2, constructing a tunnel simulation model, arranging a plurality of gas concentration sensors and wind speed sensors in a simulation tunnel, simulating to obtain time sequence concentration field data and time sequence wind field data in the tunnel under different leakage diffusion conditions and tunnel environment conditions according to set key leakage parameters and tunnel environment parameters, establishing a coordinate system in the simulation tunnel, extracting the time sequence data and wind speed time sequence data of the combustible gas concentration at the coordinate positions of each sensor from the time sequence concentration field data and the time sequence wind field data according to the set coordinates of the gas concentration sensors and the wind speed sensors in the simulation tunnel, and recording the coordinates of a leakage source, the coordinates of each sensor, the starting time of gas leakage and the leakage direction, wherein the data form a data set;
S3, processing the data set in the step S1 or/and the data set in the step S2 into a training set, a verification set and a test set for training a prediction model, and training, verifying and testing the prediction model to obtain a deep learning prediction model;
The deep learning prediction model is composed of a multi-layer convolutional neural network, a full-connection network, a long-term and short-term memory network and a full-connection network, wind speed time sequence data, combustible gas concentration time sequence data and coordinates of each sensor, which are processed by a data set, are used as input layers, and a leakage source position, a gas leakage flow rate, a leakage starting time and a leakage direction, which correspond to samples of the input layers, are used as output layers to construct the deep learning prediction model;
S4, arranging a plurality of gas concentration sensors and air speed sensors in a tunnel in which the leakage of the combustible gas occurs to be predicted, inputting the acquired time sequence data of the concentration of the combustible gas and the acquired time sequence data of the air speed into a deep learning prediction model, and obtaining prediction results of positioning, leakage rate, leakage starting time and leakage direction of a plurality of groups of combustible gas leakage sources;
S5, synthesizing all groups of prediction results to obtain final prediction results of the positioning of the combustible gas leakage source, the leakage rate, the leakage starting time and the leakage direction.
2. The method for positioning and predicting leakage rate of combustible gas leakage sources in tunnels according to claim 1, wherein the method comprises the following steps: in the step S2, one sensor is selected from the plurality of sensors and is used as a reference sensor, the relative coordinates between the coordinates of the leakage source and the reference sensor are calculated, and the positional relationship between the reference sensor and the leakage source is used as a basis for positioning the leakage source in the prediction result.
3. The method for positioning and predicting leakage rate of combustible gas leakage sources in tunnels according to claim 2, wherein the method comprises the following steps: in the step S2, the gas concentration sensor and the wind speed sensor are translated several times along the length direction of the simulated tunnel, so as to obtain wind speed time series data and combustible gas concentration time series data of the current coordinates after each translation, and relative coordinates between the leakage source and the reference sensor.
4. The method for positioning and predicting leakage rate of combustible gas leakage sources in tunnels according to claim 3, wherein the method comprises the following steps: in the steps S1 and S2, a gas concentration sensor with a concentration higher than a preset gas concentration alarm lower limit and a current time detected for the first time are searched in the wind speed time sequence data and the combustible gas concentration time sequence data acquired by the sensor, and the difference between the current time and the gas leakage starting time is the leakage starting time predicted by the model.
5. The method for positioning and predicting leakage rate of combustible gas leakage sources in tunnels according to claim 4, wherein the method comprises the following steps: in the step S4, the sensors in the tunnel are deployed according to the relative coordinates and the distance of the sensors set for the simulated tunnel in the step S2.
6. The method for positioning and predicting leakage rate of combustible gas leakage sources in tunnels according to claim 5, wherein the method comprises the following steps: in the step S3, the data set is the data set in the step S1 or/and the data set in the step S2.
7. The method for positioning and predicting leakage rate of combustible gas leakage sources in tunnels according to claim 6, wherein the method comprises the following steps: in the steps S1 and S2, a coordinate system is constructed with the elevation of the tunnel floor as a reference, and the origin of coordinates of the coordinate system is a combustible gas leakage source on or above the floor.
8. A system employing the method for positioning and predicting leakage rate of combustible gas leakage sources in a tunnel as claimed in any one of claims 1 to 7, comprising:
The data acquisition module is used for acquiring the concentration time sequence data of the combustible gas, the wind speed time sequence data and the coordinate information of each sensor in real time;
The prediction module is used for inputting the data into a deep learning prediction model to obtain prediction results of the positioning of the combustible gas leakage source, the leakage rate, the leakage starting time and the leakage direction;
The output module is used for outputting a prediction result;
The data acquisition module comprises a data experiment unit and a data simulation unit, the data experiment unit comprises a plurality of gas concentration sensors, a plurality of wind speed sensors and a host, the host consists of a central processor, a coordinate system construction module, an alarm module and a display module, the coordinate system construction module, the alarm module and the display module are respectively connected with the central processor, and the gas concentration sensors and the wind speed sensors are respectively connected with the host; the data simulation unit is used for obtaining a data set by simulating a calculation fluid simulation method according to the simulated tunnel geometric parameters and the simulated accident working condition parameters.
CN202310838809.0A 2023-07-10 2023-07-10 Method and system for positioning combustible gas leakage source and predicting leakage rate in tunnel Active CN117034740B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310838809.0A CN117034740B (en) 2023-07-10 2023-07-10 Method and system for positioning combustible gas leakage source and predicting leakage rate in tunnel

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310838809.0A CN117034740B (en) 2023-07-10 2023-07-10 Method and system for positioning combustible gas leakage source and predicting leakage rate in tunnel

Publications (2)

Publication Number Publication Date
CN117034740A CN117034740A (en) 2023-11-10
CN117034740B true CN117034740B (en) 2024-06-14

Family

ID=88621612

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310838809.0A Active CN117034740B (en) 2023-07-10 2023-07-10 Method and system for positioning combustible gas leakage source and predicting leakage rate in tunnel

Country Status (1)

Country Link
CN (1) CN117034740B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117629534B (en) * 2023-11-28 2024-07-02 北京航天试验技术研究所 Gas leakage detection method and device and gas leakage sample collection system
CN118517643A (en) * 2024-07-25 2024-08-20 海纳云物联科技有限公司 Method and apparatus for gas leak location, and computer readable storage medium

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108280849A (en) * 2018-01-23 2018-07-13 中国矿业大学(北京) A kind of correction of pipe gallery gas leakage concentration field prediction and slip method of estimation
CN116187033A (en) * 2023-01-14 2023-05-30 西安交通大学 Long-distance natural gas pipeline leakage positioning method, computer equipment and computer readable storage medium

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2740718B2 (en) * 1992-09-30 1998-04-15 三井東圧化学株式会社 Leakage point and leak amount estimation system for gas, steam, etc.
CN110398320B (en) * 2019-07-09 2020-11-20 北京讯腾智慧科技股份有限公司 Gas leakage detection positioning method and system easy for continuous optimization
CN114383054A (en) * 2021-01-27 2022-04-22 福州大学 Pipe gallery gas pipeline leakage experiment system and method
US20230176023A1 (en) * 2021-12-08 2023-06-08 Schlumberger Technology Corporation Method and apparatus for methane leakage detection
CN115773473B (en) * 2022-04-29 2024-06-04 重庆大学 Monitoring system and monitoring method for natural gas leakage in tunnel
CN115164115B (en) * 2022-07-05 2024-08-02 福州大学 Gas pipeline leakage rate prediction method based on physical model driven machine learning

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108280849A (en) * 2018-01-23 2018-07-13 中国矿业大学(北京) A kind of correction of pipe gallery gas leakage concentration field prediction and slip method of estimation
CN116187033A (en) * 2023-01-14 2023-05-30 西安交通大学 Long-distance natural gas pipeline leakage positioning method, computer equipment and computer readable storage medium

Also Published As

Publication number Publication date
CN117034740A (en) 2023-11-10

Similar Documents

Publication Publication Date Title
CN117034740B (en) Method and system for positioning combustible gas leakage source and predicting leakage rate in tunnel
Zhao et al. Leak localization using distributed sensors and machine learning for hydrogen releases from a fuel cell vehicle in a parking garage
CN111797572B (en) Urban accident disaster evolution simulation and risk prediction early warning method
CN107505267B (en) Gas detector distribution point analysis method and device
CN114321740B (en) Combustible gas leakage point positioning method, system and readable storage module
US11982613B2 (en) Methods and internet of things (IOT) systems for corrosion protection optimization of pipeline of smart gas
CN111912937B (en) Intelligent real-time tracing method and system for indoor sudden pollutants
CN114019110B (en) Workplace gas detector end cloud integration platform based on big data
CN106777893A (en) A kind of method for recognizing city space multi-pollutant source using removable contamination detector
CN114638048A (en) Three-dimensional spray pipe flow field rapid prediction and sensitivity parameter analysis method and device
CN116882268B (en) Data-driven tunnel fire smoke development prediction method and intelligent control system
CN114880935A (en) Tunnel fire advanced prediction method
KR102413399B1 (en) Leak diagnosis system for offshore plant pipelines based on machine learning
CN117763934A (en) Micro pipe gallery gas monitoring method and system based on deep learning
Yang et al. Prediction of hydrogen leakage location and intensity in hydrogen refueling stations based on deep learning
CN115730535A (en) Artificial intelligence reverse tracing and leakage determining method based on gas monitoring
CN113743015B (en) Fire scene data acquisition method, medium and electronic equipment
CN117291293B (en) Tunnel fire scene disaster perception and situation development prediction method
JPH06117600A (en) System for estimating leakage point and leakage volume of gas, steam or the like
CN118228414A (en) Automatic type selection system for pipeline support and hanger based on BIM technology and use method
CN116228770A (en) Method and system for identifying and monitoring pipeline leakage
CN118484767B (en) Underground monitoring data intelligent processing method and system
Zhao et al. A leakage detection method for hydrogen-blended natural gas pipelines in utility tunnels based on multi-task LSTM and CFD simulation
Liu et al. Feature extraction and pattern recognition of gas pipeline flow noise signals in a strong noisy background
CN117390520B (en) Transformer state monitoring method and system

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant