Evaluation system and method for analyzing health state of pressure pipeline based on multi-modal data
By obtaining pipeline laying route information and dividing areas, combined with deep learning models, the problem of inaccurate processing of pipeline monitoring data in the existing technology is solved, and the accuracy and personalized adjustment of pipeline health status evaluation are achieved.
Patent Information
- Application Number
- CN202411003931.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-25
- Publication Date
- 2024-11-08
AI Technical Summary
The existing pipeline monitoring technology cannot adaptive data processing based on environmental factors and liquid types in different regions, resulting in inaccurate health assessment results.
By obtaining pipeline laying route information, calculating the installation location of the fixed-point monitoring device, dividing pipeline areas, establishing partition databases, setting monitoring data thresholds, and using deep learning models to evaluate health status, extracting key feature data in leaked events, and adjusting similarity The monitoring data threshold for the area.
It realizes personalized monitoring data processing based on the characteristic data of different regions, improves the accuracy of pipeline health status assessment and ensures the accuracy and reliability of evaluation results.
Smart Images

Figure CN118912393A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of pipeline monitoring, and in particular to a pressure pipeline health status assessment system and method based on multimodal data analysis. Background Art
[0002] During the long-term operation of the pipeline, internal corrosion defects and pipeline cracks that are difficult to detect will occur due to the transportation of chemicals; at the same time, when physical conditions such as pressure, temperature, and pH change, special reactions will occur between liquids and liquids, and between liquids and solids, and scaling will occur inside the pipeline. Scaling blockage not only reduces the transportation flow, but also causes pipeline rupture and under-scale corrosion. In recent years, various pipeline accidents have occurred frequently, causing huge economic losses and very serious environmental problems. If a pipeline accident cannot be known, located, and repaired early, it will bring very serious consequences. Therefore, it is of great significance to evaluate the health status of the pipeline and to prevent and remedy it in time before and after secondary disasters. In recent years, domestic and foreign scholars have successively carried out research on pipeline health monitoring issues, and many types of pipeline monitoring technologies have been proposed and applied.
[0003] However, currently when pipelines are monitored, the same processing method is used for monitoring data in different regions. However, the environmental factors and types of transported liquids in pipelines in different regions are different, which will lead to inaccurate monitoring data. The data processing method cannot be adaptively adjusted according to different regions, resulting in inaccurate pipeline health assessment results. Summary of the invention
[0004] The purpose of the present invention is to provide a pressure pipeline health status assessment system and method based on multimodal data analysis to solve the problems raised in the above background technology.
[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: a pressure pipeline health status assessment method based on multimodal data analysis, comprising:
[0006] Obtaining pipeline laying route information, calculating the distribution of installation positions of fixed-point monitoring devices on the pipeline according to the route information, and installing the fixed-point monitoring devices at corresponding positions of the pipeline according to the installation positions, wherein the fixed-point monitoring devices are used to monitor pipeline data to obtain pipeline monitoring data;
[0007] Divide the pipeline into regions, monitor the pipeline in real time, and establish a regional database. Store the pipeline monitoring data in each region in the regional database. Initially set the monitoring data threshold of each region. When the pipeline monitoring data reaches the monitoring data threshold, an alarm signal is issued.
[0008] When there is no leakage, the pipeline monitoring data is uploaded, pre-processed, and then input into the deep learning model to output the pipeline health status assessment results;
[0009] When a leakage incident occurs, the pipeline monitoring data of the leaking pipeline in the leakage incident is obtained, the key feature data in the pipeline monitoring data is extracted, and the proportional factor is calculated based on the number of occurrences of the key feature data in the leakage incident recorded in the historical database. The leakage area where the leaking pipeline is located is marked, and the characteristic factors of the leakage area are extracted. The characteristic factors are input into the deep learning model, the similar area is output, and the monitoring data threshold in the similar area is adjusted according to the proportional factor.
[0010] Furthermore, the process of calculating the distribution of the installation positions of the fixed-point monitoring devices on the pipeline according to the route information is as follows:
[0011] The valve on the pipeline is marked as node one, and a fixed-point monitoring device is installed at each node. The distance threshold between two adjacent fixed-point monitoring devices is set as D, and the path length between two adjacent nodes one is obtained as d;
[0012] If d<D, there is no need to add a fixed-point monitoring device between two adjacent output nodes;
[0013] If d>D, it is further determined whether the pipeline between two adjacent nodes has a bend. If the pipeline between two adjacent nodes has no bend, it is calculated that m fixed-point monitoring devices need to be installed between two adjacent nodes. The calculation formula is: , and install m fixed-point monitoring devices equidistantly between two adjacent nodes;
[0014] If the pipe between two adjacent nodes 1 has a bend, mark the bend as node 2, and calculate the path length between node 2 and the two adjacent nodes 1. and ,like , then set up a fixed-point monitoring device at node 2; if , then compare and The path length is greater than the other path length, and the node corresponding to the path length is taken as the end point. Fixed-point monitoring devices are set up at the locations.
[0015] Furthermore, the fixed-point monitoring device is used to monitor the change data of the pipeline, and the process of obtaining the pipeline monitoring data is:
[0016] Set the monitoring time of the fixed-point monitoring device, monitor the pipeline on time according to the monitoring time, monitor the type of liquid transmitted by the pipeline, flow rate, pressure, deformation force, corrosiveness, scaling degree, temperature, humidity data, and form pipeline monitoring data;
[0017] The monitoring time includes automatic monitoring time and manual inspection time. The automatic monitoring time is to set the monitoring frequency of the fixed-point monitoring device so that it can automatically perform regular monitoring of the pipeline; the manual inspection time is to set the time point for manual inspection. When the manual inspection time is reached, the workers will conduct a comprehensive inspection of the pipeline.
[0018] Furthermore, the process of dividing the pipeline into regions is as follows:
[0019] The three-dimensional model after pipeline laying is simulated in the three-dimensional coordinate system. The pipelines at the same depth are divided into the same area. The pipelines that are connected in the same area and form the same flow path are marked as the same group. The pipelines in the same group are divided into several sections according to their service life. One section of the pipeline is marked as an area. Starting from the end of the pipeline, several sections of the pipeline are marked as several areas in turn to complete the regional division of the pipeline.
[0020] Furthermore, the initial setting of the monitoring data threshold for each area includes:
[0021] An initial monitoring data threshold is set for each area, wherein the monitoring data threshold includes a flow rate threshold, a pressure threshold, a deformation force threshold, a corrosiveness threshold, a scaling degree threshold, a temperature threshold, and a humidity threshold;
[0022] When the monitoring pipeline data reaches the monitoring data threshold of the corresponding data type, an alarm signal is issued.
[0023] Furthermore, the process of outputting the health status assessment result of the pipeline is as follows:
[0024] The monitored data on liquid type, flow rate, pressure, deformation force, corrosiveness, scaling degree, temperature, and humidity are used as training sets and input into the deep learning model to output pipeline health status assessment results, which include poor, general, good, and very good.
[0025] Furthermore, the process of extracting key feature data from pipeline monitoring data is as follows:
[0026] The pipeline monitoring data is divided into multiple data sets according to the data type, and the data type includes liquid type data, flow rate data, pressure data, deformation force data, corrosiveness data, scaling degree data, temperature data and humidity data. The historical pipeline monitoring data is extracted, and the data in each data set are used to form a corresponding fitting equation F(x). The time period before the leakage event occurs is used as the marked time period, and the slope K of each fitting equation F(x) in the marked time period is calculated. The slope K greater than the threshold K is extracted. 阈 The data type corresponding to the data set is used as the key feature data.
[0027] Furthermore, the process of calculating the proportional factor by combining the number of occurrences of key characteristic data in the leakage events recorded in the historical database is as follows:
[0028] Get the total number of leakage events N in the historical database, extract the pipeline monitoring data of the leaking pipeline in each leakage event, and calculate the key feature data therein, extract the key feature data of the same data type, get the number of occurrences of the same type of key feature data n, and calculate the proportional factor , the calculation formula is: , where C is a constant.
[0029] Furthermore, the characteristic factors include area factors, group factors, temperature factors, humidity factors and service life factors of the pipe segment. The extracted characteristic factors are input into the learning model to output the similarity between other areas and the leakage area. , set the similarity threshold ,Will > The corresponding areas are marked as similar areas;
[0030] The initial monitoring data threshold corresponding to the key feature data in the similar area is extracted and set as , get the proportional factor corresponding to the key feature data , calculate the new monitoring data threshold corresponding to the key feature data in similar areas , and its calculation formula is: , where k is a constant; the new monitoring data threshold after calculation Adjust to the corresponding monitoring data threshold in similar areas.
[0031] The pressure pipeline health status assessment system based on multimodal data analysis is applied to the pressure pipeline health status assessment method based on multimodal data analysis, including:
[0032] A monitoring device layout module, which calculates the installation distribution positions of various fixed-point monitoring devices on the pipeline according to the route information of the pipeline laying, and installs the fixed-point monitoring devices at the corresponding positions of the pipeline to monitor the data of the pipeline and obtain the pipeline monitoring data;
[0033] A zone division module, which is used to divide the pipeline into zones, monitor the pipeline in zones, and store the pipeline monitoring data obtained by the zone monitoring in a corresponding zone database;
[0034] A threshold alarm module, which is used to set the monitoring data threshold of each area, and when the monitoring pipeline data reaches the monitoring data threshold, an alarm signal is issued;
[0035] A threshold adjustment module, which performs personalized adjustment on the monitoring data threshold in each area according to the number of occurrences of key feature data in historical leakage events;
[0036] A health status assessment module includes a deep learning model, which inputs the collected monitoring pipeline data into the deep learning model, and after normalization processing, outputs the health status assessment result of the corresponding pipeline.
[0037] Compared with the prior art, the beneficial effects achieved by the present invention are:
[0038] 1. The present invention obtains key feature data in historical leakage events of the same leakage area by analyzing, and adjusts the monitoring data thresholds of other similar areas similar to the leakage area according to the occurrence frequency of the key feature data, thereby achieving the purpose of adaptively adjusting the monitoring data thresholds of different areas to adopt different data processing methods, so as to ensure that the obtained pipeline health status assessment results are more accurate.
[0039] 2. The present invention accurately calculates the distribution position of the fixed-point monitoring device through the route information of the pipeline, so that when monitoring, the fixed-point monitoring device with different distribution position can be used to verify whether the pipeline monitoring data of other fixed-point monitoring devices are accurate, thereby improving the accuracy of the pipeline health status assessment results. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:
[0041] Figure 1 It is a flow chart of adjusting the monitoring data threshold of similar areas in the pressure pipeline health status assessment method based on multimodal data analysis of the present invention. DETAILED DESCRIPTION
[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0043] Example 1: Please refer to Figure 1 The present invention provides a technical solution: a pressure pipeline health status assessment method based on multimodal data analysis, comprising:
[0044] Obtaining pipeline laying route information, calculating the distribution of installation positions of fixed-point monitoring devices on the pipeline according to the route information, and installing the fixed-point monitoring devices at corresponding positions of the pipeline according to the installation positions, wherein the fixed-point monitoring devices are used to monitor pipeline data to obtain pipeline monitoring data;
[0045] In a preferred embodiment, the process of calculating the distribution of installation positions of fixed-point monitoring devices on the pipeline according to route information is as follows:
[0046] The valve on the pipeline is marked as node one, and a fixed-point monitoring device is installed at each node. The distance threshold between two adjacent fixed-point monitoring devices is set as D, and the path length between two adjacent nodes one is obtained as d;
[0047] If d<D, there is no need to add a fixed-point monitoring device between two adjacent output nodes;
[0048] If d>D, it is further determined whether the pipeline between two adjacent nodes has a bend. If the pipeline between two adjacent nodes has no bend, it is calculated that m fixed-point monitoring devices need to be installed between two adjacent nodes. The calculation formula is: , and install m fixed-point monitoring devices equidistantly between two adjacent nodes;
[0049] If the pipe between two adjacent nodes 1 has a bend, mark the bend as node 2, and calculate the path length between node 2 and the two adjacent nodes 1. and ,like , then set up a fixed-point monitoring device at node 2; if , then compare and The path length is greater than the other path length, and the node corresponding to the path length is taken as the end point. Fixed-point monitoring devices are set up at the locations.
[0050] For example, a fixed-point monitoring device is set at each valve. If the pipeline between the two valves is bent and the path length between the two valves is greater than the set distance threshold, such as the path distance between the two valves is 30 meters and the distance threshold is 20 meters, then a fixed-point monitoring device needs to be installed between the two valves. At this time, the turning point needs to be marked. If the turning point is at the midpoint, a fixed-point monitoring device is installed at the turning point. If the turning point is at a different distance from the valve path on both sides, such as the left side is 19 meters, on the right The distance is 11 meters, then a fixed-point monitoring device is installed at the position of 19-11=8 meters on the left side;
[0051] The distribution of fixed-point monitoring devices on the pipeline can be obtained through the above calculations, and the fixed-point monitoring devices are installed according to the calculated distribution positions, wherein the fixed-point monitoring devices include temperature sensors, humidity sensors, flow rate sensors, and other devices that can monitor the pipeline;
[0052] During routine inspection, monitoring is performed using a fixed-point monitoring device at the valve. When the difference between the pipeline monitoring data and the monitoring data threshold is less than the difference threshold, a signal is sent to start monitoring the fixed-point monitoring device between the current valve and the adjacent valve to further verify whether the pipeline monitoring data monitored at this valve is accurate; or the monitoring frequency of the fixed-point monitoring device between the two valves is set to monitor the pipeline within a fixed time to assist in verifying whether the pipeline monitoring data at the valve is accurate.
[0053] The above-mentioned difference threshold is a set value, and the monitoring frequency of the fixed-point monitoring device is also a set value.
[0054] In a preferred embodiment, the fixed-point monitoring device is used to monitor the change data of the pipeline, and the process of obtaining the pipeline monitoring data is:
[0055] Set the monitoring time of the fixed-point monitoring device, monitor the pipeline on time according to the monitoring time, monitor the type of liquid transmitted by the pipeline, flow rate, pressure, deformation force, corrosiveness, scaling degree, temperature, humidity data, and form pipeline monitoring data;
[0056] The monitoring time includes automatic monitoring time and manual inspection time. The automatic monitoring time is to set the monitoring frequency of the fixed-point monitoring device so that it can automatically perform regular monitoring of the pipeline; the manual inspection time is to set the time point for manual inspection. When the manual inspection time is reached, the workers will conduct a comprehensive inspection of the pipeline.
[0057] When the health status of the pipeline deteriorates, the workers' manual inspection frequency can be adjusted in time.
[0058] Divide the pipeline into regions, monitor the pipeline in real time, and establish a regional database. Store the pipeline monitoring data in each region in the regional database. Initially set the monitoring data threshold of each region. When the pipeline monitoring data reaches the monitoring data threshold, an alarm signal is issued.
[0059] In a preferred embodiment, the process of dividing the pipeline into regions is as follows:
[0060] The three-dimensional model after pipeline laying is simulated in the three-dimensional coordinate system. The pipelines at the same depth are divided into the same area. The pipelines that are connected in the same area and form the same flow path are marked as the same group. The pipelines in the same group are divided into several sections according to their service life. One section of the pipeline is marked as an area. Starting from the end of the pipeline, several sections of the pipeline are marked as several areas in turn to complete the regional division of the pipeline.
[0061] In a preferred embodiment, the initial setting of the monitoring data threshold of each area includes:
[0062] An initial monitoring data threshold is set for each area, wherein the monitoring data threshold includes a flow rate threshold, a pressure threshold, a deformation force threshold, a corrosiveness threshold, a scaling degree threshold, a temperature threshold, and a humidity threshold;
[0063] When the monitoring pipeline data reaches the monitoring data threshold of the corresponding data type, an alarm signal is issued.
[0064] When there is no leakage, the pipeline monitoring data is uploaded, pre-processed, and then input into the deep learning model to output the pipeline health status assessment results;
[0065] In a preferred embodiment, the process of outputting the health status assessment result of the pipeline is:
[0066] The monitored data on liquid type, flow rate, pressure, deformation force, corrosiveness, scaling degree, temperature, and humidity are used as training sets and input into the deep learning model to output pipeline health status assessment results, which include poor, general, good, and very good.
[0067] When a leakage incident occurs, the pipeline monitoring data of the leaking pipeline in the leakage incident is obtained, the key feature data in the pipeline monitoring data is extracted, and the proportional factor is calculated based on the number of occurrences of the key feature data in the leakage incident recorded in the historical database. The leakage area where the leaking pipeline is located is marked, and the characteristic factors of the leakage area are extracted. The characteristic factors are input into the deep learning model, the similar area is output, and the monitoring data threshold in the similar area is adjusted according to the proportional factor.
[0068] In a preferred embodiment, the process of extracting key feature data from pipeline monitoring data is as follows:
[0069] The pipeline monitoring data is divided into multiple data sets according to the data type, and the data type includes liquid type data, flow rate data, pressure data, deformation force data, corrosiveness data, scaling degree data, temperature data and humidity data. The historical pipeline monitoring data is extracted, and the data in each data set are used to form a corresponding fitting equation F(x). The time period before the leakage event occurs is used as the marked time period, and the slope K of each fitting equation F(x) in the marked time period is calculated. The slope K greater than the threshold K is extracted. 阈 The data type corresponding to the data set is used as the key feature data.
[0070] For example, the collected pressure data and scaling data form fitting equations to form pressure equations and scaling equations, extract the slopes K of the pressure equations and scaling equations in a period of time before the leakage event occurs, and compare them with the corresponding threshold K 阈 By comparison, if both are greater than the threshold, the pressure data and scaling data are taken as key feature data, and the changes of these two data need to be monitored during monitoring.
[0071] In a preferred embodiment, the process of calculating the proportional factor in combination with the number of occurrences of key characteristic data in the leakage events recorded in the historical database is as follows:
[0072] Get the total number of leakage events N in the historical database, extract the pipeline monitoring data of the leaking pipeline in each leakage event, and calculate the key feature data therein, extract the key feature data of the same data type, get the number of occurrences of the same type of key feature data n, and calculate the proportional factor , the calculation formula is: , where C is a constant.
[0073] In a preferred embodiment, the characteristic factors include area factors, group factors, temperature factors, humidity factors and service life factors of the pipe segment. The extracted characteristic factors are input into the learning model, and the similarity between other areas and the leakage area is output. , set the similarity threshold ,Will > The corresponding areas are marked as similar areas;
[0074] The initial monitoring data threshold corresponding to the key feature data in the similar area is extracted, set as , get the proportional factor corresponding to the key feature data , calculate the new monitoring data threshold corresponding to the key feature data in similar areas , and its calculation formula is: , where k is a constant; the new monitoring data threshold after calculation Adjust to the corresponding monitoring data threshold in similar areas.
[0075] For example, after extracting the characteristic factors in the leakage area, other areas with the same or similar characteristic factors are extracted as similar areas. Since similar areas have some similar characteristic factors as the leakage area, the key feature data extracted from the leakage area can be used as the key monitoring object of the similar areas to ensure that the health status of the pipeline is predicted in advance.
[0076] Embodiment 2: A pressure pipeline health status assessment system based on multimodal data analysis is applied to a pressure pipeline health status assessment method based on multimodal data analysis, comprising:
[0077] A monitoring device layout module, which calculates the installation distribution positions of various fixed-point monitoring devices on the pipeline according to the route information of the pipeline laying, and installs the fixed-point monitoring devices at the corresponding positions of the pipeline to monitor the data of the pipeline and obtain the pipeline monitoring data;
[0078] A zone division module, which is used to divide the pipeline into zones, monitor the pipeline in zones, and store the pipeline monitoring data obtained by the zone monitoring in a corresponding zone database;
[0079] A threshold alarm module, which is used to set the monitoring data threshold of each area and send out an alarm signal when the monitoring pipeline data reaches the monitoring data threshold;
[0080] A threshold adjustment module, which performs personalized adjustment on the monitoring data threshold in each area according to the number of occurrences of key feature data in historical leakage events;
[0081] A health status assessment module includes a deep learning model, which inputs the collected monitoring pipeline data into the deep learning model, and after normalization processing, outputs the health status assessment result of the corresponding pipeline.
[0082] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. A pressure pipeline health status assessment method based on multimodal data analysis, characterized in that: include: Obtaining pipeline laying route information, calculating the distribution of installation positions of fixed-point monitoring devices on the pipeline according to the route information, and installing the fixed-point monitoring devices at corresponding positions of the pipeline according to the installation positions, wherein the fixed-point monitoring devices are used to monitor pipeline data to obtain pipeline monitoring data; Divide the pipeline into regions, monitor the pipeline in real time, and establish a regional database. Store the pipeline monitoring data in each region in the regional database. Initially set the monitoring data threshold of each region. When the pipeline monitoring data reaches the monitoring data threshold, an alarm signal is issued. When there is no leakage, the pipeline monitoring data is uploaded, pre-processed, and then input into the deep learning model to output the pipeline health status assessment results; When a leakage incident occurs, the pipeline monitoring data of the leaking pipeline in the leakage incident is obtained, the key feature data in the pipeline monitoring data is extracted, and the proportional factor is calculated based on the number of occurrences of the key feature data in the leakage incident recorded in the historical database. The leakage area where the leaking pipeline is located is marked, and the characteristic factors of the leakage area are extracted. The characteristic factors are input into the deep learning model, the similar area is output, and the monitoring data threshold in the similar area is adjusted according to the proportional factor.
2. The method for assessing the health status of a pressure pipeline based on multimodal data analysis according to claim 1 is characterized in that: The process of calculating the distribution of the installation positions of the fixed-point monitoring devices on the pipeline according to the route information is as follows: The valve on the pipeline is marked as node one, and a fixed-point monitoring device is installed at each node. The distance threshold between two adjacent fixed-point monitoring devices is set as D, and the path length between two adjacent nodes one is obtained as d; If d<D, there is no need to add a fixed-point monitoring device between two adjacent output nodes; If d>D, it is further determined whether the pipeline between two adjacent nodes has a bend. If the pipeline between two adjacent nodes has no bend, it is calculated that m fixed-point monitoring devices need to be installed between two adjacent nodes. The calculation formula is: , and install m fixed-point monitoring devices equidistantly between two adjacent nodes; If the pipeline between two adjacent nodes 1 has a bend, mark the bend as node 2, and calculate the path length between node 2 and the two adjacent nodes 1. and ,like , then set up a fixed-point monitoring device at node 2; if , then compare and The path length is greater than the other path length, and the node corresponding to the path length is taken as the end point. Fixed-point monitoring devices are set up at the locations.
3. The method for assessing the health status of a pressure pipeline based on multimodal data analysis according to claim 1 is characterized in that: The fixed-point monitoring device is used to monitor the change data of the pipeline, and the process of obtaining the pipeline monitoring data is as follows: Set the monitoring time of the fixed-point monitoring device, monitor the pipeline on time according to the monitoring time, monitor the type of liquid transmitted by the pipeline, flow rate, pressure, deformation force, corrosiveness, scaling degree, temperature, humidity data, and form pipeline monitoring data; The monitoring time includes automatic monitoring time and manual inspection time. The automatic monitoring time is to set the monitoring frequency of the fixed-point monitoring device so that it can automatically perform regular monitoring of the pipeline; the manual inspection time is to set the time point for manual inspection. When the manual inspection time is reached, the workers will conduct a comprehensive inspection of the pipeline.
4. The method for assessing the health status of a pressure pipeline based on multimodal data analysis according to claim 3 is characterized in that: The process of dividing the pipeline into regions is as follows: The three-dimensional model after pipeline laying is simulated in the three-dimensional coordinate system. The pipelines at the same depth are divided into the same area. The pipelines that are connected in the same area and form the same flow path are marked as the same group. The pipelines in the same group are divided into several sections according to their service life. One section of the pipeline is marked as an area. Starting from the end of the pipeline, several sections of the pipeline are marked as several areas in turn to complete the regional division of the pipeline.
5. The method for assessing the health status of a pressure pipeline based on multimodal data analysis according to claim 1 is characterized in that: The initial setting of monitoring data thresholds for each area includes: An initial monitoring data threshold is set for each area, wherein the monitoring data threshold includes a flow rate threshold, a pressure threshold, a deformation force threshold, a corrosiveness threshold, a scaling degree threshold, a temperature threshold, and a humidity threshold; When the monitoring pipeline data reaches the monitoring data threshold of the corresponding data type, an alarm signal is issued.
6. The method for evaluating the health status of a pressure pipeline based on multimodal data analysis according to claim 1 is characterized in that: The process of outputting the health status assessment results of the pipeline is as follows: The monitored data on liquid type, flow rate, pressure, deformation force, corrosiveness, scaling degree, temperature, and humidity are used as training sets and input into the deep learning model to output the pipeline health status assessment results.
7. The method for assessing the health status of a pressure pipeline based on multimodal data analysis according to claim 5 is characterized in that: The process of extracting key feature data from pipeline monitoring data is as follows: The pipeline monitoring data is divided into multiple data sets according to the data type, and the data type includes liquid type data, flow rate data, pressure data, deformation force data, corrosiveness data, scaling degree data, temperature data and humidity data. The historical pipeline monitoring data is extracted, and the data in each data set are used to form a corresponding fitting equation F(x). The time period before the leakage event occurs is used as the marked time period, and the slope K of each fitting equation F(x) in the marked time period is calculated. The slope K greater than the threshold K is extracted. 阈 The data type corresponding to the data set is used as the key feature data.
8. The method for assessing the health status of a pressure pipeline based on multimodal data analysis according to claim 7 is characterized in that: The process of calculating the proportional factor by combining the number of occurrences of key characteristic data in the leakage events recorded in the historical database is as follows: Get the total number of leakage events N in the historical database, extract the pipeline monitoring data of the leaking pipeline in each leakage event, and calculate the key feature data therein, extract the key feature data of the same data type, get the number of occurrences of the same type of key feature data n, and calculate the proportional factor , the calculation formula is: , where C is a constant.
9. The method for assessing the health status of a pressure pipeline based on multimodal data analysis according to claim 8, characterized in that: The characteristic factors include area factors, group factors, temperature factors, humidity factors and service life factors of the pipe segment. The extracted characteristic factors are input into the learning model to output the similarity between other areas and the leakage area. , set the similarity threshold ,Will > The corresponding areas are marked as similar areas; The initial monitoring data threshold corresponding to the key feature data in the similar area is extracted and set as , get the proportional factor corresponding to the key feature data , calculate the new monitoring data threshold corresponding to the key feature data in similar areas , and its calculation formula is: , where k is a constant; the new monitoring data threshold after calculation Adjust to the corresponding monitoring data threshold in similar areas.
10. A pressure pipeline health status assessment system based on multimodal data analysis, applied to a pressure pipeline health status assessment method based on multimodal data analysis as claimed in any one of claims 1 to 9, characterized in that: include: A monitoring device layout module, which calculates the installation distribution positions of various fixed-point monitoring devices on the pipeline according to the route information of the pipeline laying, and installs the fixed-point monitoring devices at the corresponding positions of the pipeline to monitor the data of the pipeline and obtain the pipeline monitoring data; A zone division module, which is used to divide the pipeline into zones, monitor the pipeline in zones, and store the pipeline monitoring data obtained by the zone monitoring in a corresponding zone database; A threshold alarm module, which is used to set the monitoring data threshold of each area, and when the monitoring pipeline data reaches the monitoring data threshold, an alarm signal is issued; A threshold adjustment module, which performs personalized adjustment on the monitoring data threshold in each area according to the number of occurrences of key feature data in historical leakage events; A health status assessment module includes a deep learning model, which inputs the collected monitoring pipeline data into the deep learning model, and after normalization processing, outputs the health status assessment result of the corresponding pipeline.
Citation Information
Patent Citations
Water supply network pipe explosion risk prediction method and system
CN112101811A
Pipe network leakage risk assessment method
CN116384740A
Monitoring and early warning processing method, device and equipment for water supply network and medium
CN117649752A
Pipeline optical fiber early warning method, device and equipment based on multiple modes and storage medium
CN117668721A
VOCs visual monitoring and early warning system based on artificial intelligence
CN118351987A
Cited By
Intelligent heat supply network cloud control information platform system based on big data analysis
CN119648190A
Pipeline health analysis system for high pour-point oil delivery
CN120351456A
