US-20260126336-A1 - METHODS AND INTERNET OF THINGS SYSTEMS FOR GAS PIPELINE NETWORK MAINTENANCE

US20260126336A1US 20260126336 A1US20260126336 A1US 20260126336A1US-20260126336-A1

Abstract

The embodiments of the present disclosure provide methods and Internet of Thing systems for determining a gas leakage based on smart gas. The method may be implemented by a processor of a smart gas safety management platform based on an Internet of Things system for determining a gas leakage, comprising: obtaining first pipeline data located at a plurality of points of a gas pipeline network; determining a target pipeline section based on the first pipeline data; obtaining second pipeline data at both ends of the target pipeline section; determining a location of a gas leakage of the target pipeline section based on the second pipeline data; and feeding the location of the gas leakage back to a terminal of a gas management user.

Inventors

Zehua Shao
Haitang XIANG
Bin Liu

Assignees

CHENGDU QINCHUAN IOT TECHNOLOGY CO., LTD.

Dates

Publication Date: 20260507
Application Date: 20251228
Priority Date: 20230202

Claims (14)

1 . An Internet of Things system for gas pipeline network maintenance, comprising: a smart gas safety management platform, a smart gas pipeline network device sensor network platform, and a smart gas pipeline network device object platform, wherein the smart gas pipeline network device object platform, including a gas device and a monitoring device, is configured to collect first pipeline data and second pipeline data at a plurality of points of a gas pipeline network; and transmit the first pipeline data and the second pipeline data to the smart gas safety management platform through the smart gas pipeline network device sensor network platform; the first pipeline data includes sound data or vibration data; the second pipeline data includes a vibration frequency at any point of a target pipeline section; the smart gas pipeline network device sensor network platform is configured to connect the smart gas safety management platform with the smart gas pipeline network device object platform; and the smart gas safety management platform includes a process, wherein the process is configured to: obtain first pipeline data located at a plurality of points of the gas pipeline network; obtain a location where an intersection of pipeline sections, an inflection point of each pipeline section, or an installation location of a vibration sensor of each pipeline section is located, a torsion angle between the pipeline sections, and a global feature of each pipeline section; obtain the location where the intersection of the pipeline sections, the inflection point of each pipeline section, or a vibration feature of the installation location of the vibration sensor of each pipeline section through one or more vibration sensors arranged in the gas pipeline network; determine feature information of the gas pipeline network based on the location where the intersection of the pipeline sections, the inflection point of each pipeline section, or the installation location and the vibration feature of the installation location of the vibration sensor of each pipeline section is located, the torsion angle between the pipeline sections, and the global feature of each pipeline section; construct a pipeline map of the gas pipeline network based on the feature information of the gas pipeline network, wherein the pipeline map comprises a node and an edge, the node corresponds to the intersection of the pipeline sections, the inflection point of teach pipeline section, or the installation location of the vibration sensor of each pipeline section, the edge corresponds to the pipeline section, a node feature of the node includes the torsional angle and the vibration feature, and an edge feature of the edge includes the global feature of each pipeline section; determine, based on the pipeline map, a leakage suspicious degree of each pipeline in the pipeline map through a suspicious pipeline prediction model, wherein the suspicious pipeline prediction model is a machine learning model; and determine the target pipeline section based on the leakage suspicious degree; obtain second pipeline data at both ends of the target pipeline section; determine a location of the gas leakage of the target pipeline section based on the second pipeline data; feed the location of the gas leakage back to a terminal of a gas management user; and the terminal of the gas management is configured to in response to the location of the gas leak, generate a maintenance engineer dispatching instruction to dispatch a maintenance engineer in the smart gas pipeline network device object platform to the location of the gas leakage and a terminal of the maintenance engineer performs maintenance on the location of the gas leakage based on the dispatching instruction.
2 . The system of claim 1 , wherein the system further includes a smart gas user platform and a smart gas service platform, the smart gas user platform includes a gas user sub-platform and a supervision user sub-platform; the smart gas user platform is configured as a mobile device and a tablet computer; the smart gas service platform includes a smart gas use service sub-platform corresponding to the gas user sub-platform and a smart supervision service sub-platform corresponding to the supervision user sub-platform; the smart gas service platform is configured as a processing device, the processing device is a server or a server group; the smart gas safety management platform includes a smart gas pipeline network management sub-platform and a smart gas data center, wherein the smart gas pipeline network management sub-platform includes a pipeline network gas leakage monitoring module, a safety emergency management module, and a pipeline network geographic information management module; the smart gas data center includes a storage device; the smart gas data center is configured to send relevant safety data to the corresponding pipeline network gas leakage monitoring module by identifying a safety parameter category, the safety parameter category includes a gas pipeline sound feature and a vibration feature; and in response to the relevant safety data exceeding a safety monitoring threshold preset by the pipeline network gas leakage monitoring module, alarm automatically and push alarm information to a user automatically.
3 . The system of claim 1 , wherein the processor is further configured to: divide the target pipeline section into a left and a right of the target pipeline section by dichotomization; if the second pipeline data at the left of the target pipeline section is greater than the second pipeline data at the right of the target pipeline section, determine a midpoint of the target pipeline section as a right endpoint of a new target pipeline section, determine a left endpoint of an original target pipeline section as a left endpoint of the new target pipeline section; and determine the new target pipeline section based on the second pipeline data at both ends of the new target pipeline section until the location of the gas leakage of the target pipeline section is finally determined.
4 . The system of claim 1 , wherein the processor is further configured to: obtain the suspicious pipeline prediction model through a great number of first training samples with labels, wherein the first training sample includes a sample pipeline map, and the label of the first training sample may be a tag that indicates whether a leakage occurs in a pipeline section corresponding to an edge of the sample pipeline map; and when the suspicious pipeline prediction model is trained, input the sample pipeline map of the first training sample into an initial suspicious pipeline prediction model, construct a first loss function based on the leakage suspicious degree output by the initial suspicious pipeline prediction model and the label of the first training sample and update a parameter of the initial suspicious pipeline prediction model iteratively based on the first loss function until a preset condition is satisfied, determine the parameter of the initial suspicious pipeline prediction model and obtain a trained suspicious pipeline prediction model; wherein the preset condition includes convergence of the first loss function and a training period reaching a threshold.
5 . The system of claim 1 , wherein the edge feature further includes a pipeline section feature vector and the pipeline section feature vector is determined by a pipeline feature embedding layer in a leakage suspicious degree determination model based on a pipeline unit feature sequence.
6 . The system of claim 5 , wherein the edge feature further includes: a sequence element feature of the pipeline unit feature sequence.
7 . The system of claim 1 , wherein the processor is further configured to: performing one or more rounds of iteration based on the second pipeline data until a preset iteration condition may be satisfied and an iteration result may be obtained; and determine the location of the gas leakage based on the iteration result, wherein at least one of the one or more rounds of iteration includes: issue an instruction to an inspection robot to control the inspection robot to move to a target location of the target pipeline section; wherein the inspection robot includes a camera and a vibration sensor; divide the target pipeline section into at least one target pipeline subsection based on the target location; identify the target location on the pipeline gas network through the camera of the inspection robot automatically, and obtain target data of the target location through the vibration sensor of the inspection robot; obtain a new target pipeline section by removing a target pipeline subsection satisfying a first preset condition from the target pipeline section based on the target data; the first preset condition includes a gas leakage risk of the target pipeline subsection being smaller than a leakage risk threshold; in response to a determination that the preset iteration condition is not satisfied, determine the new target pipeline section as the target pipeline section for a next round of iteration; or in response to a determination that the preset iteration condition is satisfied, stop the iteration, wherein the iteration result includes the new target pipeline section or the target data.
8 . The system of claim 7 , wherein the processor is further configured to: predict, based on the target data, a leakage suspicious degree of each target pipeline subsection through a leakage suspicious degree determination model, wherein the leakage suspicious degree determination model is a machine learning model; and remove the target pipeline subsection with the leakage suspicious degree satisfies the first preset condition from the target pipeline section.
9 . The system of claim 8 , wherein the leakage suspicious degree determination model includes an embedding layer and a leakage suspicious degree determination layer and the embedding layer includes an endpoint embedding layer; and the endpoint embedding layer is configured to determine an endpoint feature vector by processing endpoint feature information of the target pipeline subsection.
10 . The system of claim 9 , wherein the embedding layer further includes a pipeline feature embedding layer and the pipeline feature embedding layer is configured to determine a pipeline section feature vector by processing a pipeline unit feature sequence of the target pipeline section; and the leakage suspicious degree determination layer is configured to determine the leakage suspicious degree by processing the endpoint feature vector and the pipeline section feature vector.
11 . The system of claim 10 , wherein the leakage suspicious degree determination model is obtained by joint training of the embedding layer and the leakage suspicious degree determination layer, a second training sample of the leakage suspicious degree determination model includes a sample pipeline unit feature sequence including the target pipeline section and sample endpoint feature information at both ends of the target pipeline subsection in the sample pipeline unit feature sequence, the label of the second training sample is a tag that indicates whether a gas leakage occurs in a target pipeline section subsection corresponding to an element and the sample endpoint feature information of the element in the sample pipeline unit feature sequence; when the leakage suspicious degree determination model is trained, the sample pipeline unit feature sequence and the sample endpoint feature information of the second training sample are input into the pipeline feature embedding layer and the endpoint embedding layer of the leakage suspicious degree determination model, respectively, the pipeline section feature vector output by the pipeline feature embedding layer and the endpoint feature vector output by the endpoint embedding layer are input into the leakage suspicious degree determination layer of the leakage suspicious degree determination model, a second loss function is constructed based on the leakage suspicious degree output by the leakage suspicious degree determination layer and the label of a second sample, parameters of an initial leakage suspicious degree determination model is iteratively updated based on the second loss function until a second preset condition is satisfied, the parameters in the leakage suspicious degree determination model is determined, and a trained leakage suspicious degree determination model is obtained, the second preset condition includes convergence of the second loss function and a training period reaching a threshold.
12 . The system of claim 7 , wherein the processor is further configured to: determine a plurality of candidate locations based on a length of the target pipeline section and a preset step length; preset a plurality of leakage features for the plurality of candidate locations based on the plurality of candidate locations; determine prediction vibration features at both ends of the target pipeline section corresponding to each candidate location through a vibration feature determination model based on the leakage feature, wherein the vibration feature determination model is the machine learning model; calculate vibration compliance degrees at both ends of the target pipeline section corresponding to the each candidate location based on the prediction vibration features and actual vibration features at both ends of the target pipeline section; determine a first confidence point based on the vibration compliance degree; and obtain data of the first confidence point as the target data by controlling the inspection robot to move to the first confidence point and determining a location where the first confidence point is located as the target location of the target pipeline section.
13 . The system of claim 12 , wherein the first confidence point is determined by weighted summation based on weights of the vibration compliance degrees at both ends of the each candidate location, wherein the weights are determined based on distances between the candidate location and two endpoints and a sequence element feature of a pipeline unit feature subsequence, wherein the closer the candidate location is to the endpoint, the greater the weight of the vibration compliance degree at the endpoint is set; and the higher a density of the sequence element of the pipeline unit feature subsequence between the candidate location and the endpoint, the greater the weights of the vibration compliance degrees at the end is set.
14 . The system of claim 12 , wherein the vibration feature determination model includes a pipeline feature embedding layer and a vibration feature determination layer; the pipeline feature embedding layer is configured to determine a pipeline section feature vector based on a pipeline unit feature subsequence; and the vibration feature determination layer is configured to determine a prediction vibration feature based on a plurality of leakage locations, leakage features corresponding to the plurality of candidate locations, an endpoint location of the target pipeline section, and the pipeline section feature vector output by the pipeline feature embedding layer.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS This application is a continuation of a U.S. application Ser. No. 18/183,132, filed on Mar. 13, 2023, which claims priority to Chinese Patent Application No. 202310103582.5, filed on Feb. 2, 2023, the entire contents of which are hereby incorporated by reference. TECHNICAL FIELD The present disclosure relates to the field of gas safety, and in particular, to methods and Internet of Things systems for gas pipeline network maintenance. BACKGROUND A gas pipeline network is generally laid underground, together with a complex urban environment, which invisibly increases the difficulty of risk investigation of the gas pipeline network and brings huge personal and property losses once a safety accident occurs. In the prior art, it is generally to determine whether a gas leakage occurs in the gas pipeline network through manual inspection, deployment of an on-site monitoring device such as a sensor, etc., combined with the upper monitoring software, etc., which is relatively difficult to monitor the gas leakage of some concealed locations or invisible underground locations. Therefore, it is necessary to provide methods for determining a gas leakage based on smart gas to ensure reliable operation of the gas pipeline network. SUMMARY According to one or more embodiments of the present disclosure, an Internet of Things system for gas pipeline network maintenance is provided. An Internet of Things system for gas pipeline network maintenance, comprising: a smart gas safety management platform, a smart gas pipeline network device sensor network platform, and a smart gas pipeline network device object platform, wherein the smart gas pipeline network device object platform, including a gas device and a monitoring device, is configured to collect first pipeline data and second pipeline data at a plurality of points of a gas pipeline network; and transmit the first pipeline data and the second pipeline data to the smart gas safety management platform through the smart gas pipeline network device sensor network platform; the first pipeline data includes sound data or vibration data; the second pipeline data includes a vibration frequency at any point of a target pipeline section; the smart gas pipeline network device sensor network platform is configured to connect the smart gas safety management platform with the smart gas pipeline network device object platform; and the smart gas safety management platform includes a process, wherein the process is configured to: obtain first pipeline data located at a plurality of points of the gas pipeline network; obtain a location where an intersection of pipeline sections, an inflection point of each pipeline section, or an installation location of a vibration sensor of each pipeline section is located, a torsion angle between the pipeline sections, and a global feature of each pipeline section; obtain the location where the intersection of the pipeline sections, the inflection point of each pipeline section, or a vibration feature of the installation location of the vibration sensor of each pipeline section through one or more vibration sensors arranged in the gas pipeline network; determine feature information of the gas pipeline network based on the location where the intersection of the pipeline sections, the inflection point of each pipeline section, or the installation location and the vibration feature of the installation location of the vibration sensor of each pipeline section is located, the torsion angle between the pipeline sections, and the global feature of each pipeline section; construct a pipeline map of the gas pipeline network based on the feature information of the gas pipeline network, wherein the pipeline map comprises a node and an edge, the node corresponds to the intersection of the pipeline sections, the inflection point of teach pipeline section, or the installation location of the vibration sensor of each pipeline section, the edge corresponds to the pipeline section, a node feature of the node includes the torsional angle and the vibration feature, and an edge feature of the edge includes the global feature of each pipeline section; determine, based on the pipeline map, a leakage suspicious degree of each pipeline in the pipeline map through a suspicious pipeline prediction model, wherein the suspicious pipeline prediction model is a machine learning model; and determine the target pipeline section based on the leakage suspicious degree; obtain second pipeline data at both ends of the target pipeline section; determine a location of the gas leakage of the target pipeline section based on the second pipeline data; feed the location of the gas leakage back to a terminal of a gas management user; and the terminal of the gas management is configured to in response to the location of the gas leak, generate a maintenance engineer dispatching instruction to dispatch a maintenance engineer in the smart gas pipeline network de