Pipeline damage behavior prediction and protection method based on AI large model

Abstract

The invention discloses a pipeline damage behavior prediction and protection method based on an AI large model. The pipeline damage behavior prediction and protection method based on the AI large model comprises the following steps: S1, acquiring gas pipeline operation data; s2, gas pipeline data analysis; s3, AI model training; s4, detecting the running state of the gas pipeline; and S5, improving and optimizing the AI model. According to the method, the gas pipeline damage characteristic data is obtained through the gas pipeline historical operation data, the pipeline state monitoring index is obtained according to the gas pipeline damage characteristic data, the pipeline state monitoring model is obtained based on supervised learning training according to the pipeline state monitoring index, the pipeline state monitoring model is subjected to repetitive training, and the pipeline state monitoring accuracy is improved. The identification capability of the pipeline state monitoring model on the pipeline state is improved, the purpose of training by learning a large amount of data through the AI large model so as to improve the prediction capability of the pipeline damage behavior is achieved, and the problem of low prediction capability of the pipeline damage behavior in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of pipeline destruction behavior prediction and protection, and in particular to a pipeline destruction behavior prediction and protection method based on an AI large model. Background Art

[0002] Pipelines are widely used in transporting gas, liquid and solid devices, especially for drainage, heating, gas supply and long-distance transportation of natural gas and various industrial devices. Pipelines are pipelines for transmitting working fluids in hydraulic systems. It is also a reasonably arranged pipeline system. Because of the flexibility of pipelines, they are often used in mechanical equipment driven by liquids. Pipelines have a service life. If damaged pipelines are not detected in time, safety hazards may occur, causing casualties, property losses or environmental damage. It can be seen that the prediction of pipeline damage behavior is imminent.

[0003] The existing pipeline damage behavior prediction and protection system collects pipeline information to establish a model for numerical simulation, obtains the carbon dioxide corrosion rate, and calculates the pipeline corrosion rate based on the pipeline corrosion damage probability, thereby predicting the pipeline service life.

[0004] For example, a pipeline internal corrosion prediction method disclosed in the invention patent announcement with publication number: CN109816133B includes: dividing a target pipeline into multiple sub-pipelines; obtaining the water accumulation probability and the corrosion probability of the sub-pipelines; multiplying the water accumulation probability by the corrosion probability to obtain the total corrosion probability of the sub-pipelines; and predicting whether each of the sub-pipelines of the target pipeline is prone to internal corrosion based on the total corrosion probability.

[0005] For example, a patent application with publication number: CN114066267A discloses an intelligent method for evaluating the integrity of an oil and gas pipeline network, which includes the following construction steps: Step 1: collecting oil and gas pipeline network integrity evaluation data; Step 2: constructing an oil and gas pipeline network integrity evaluation index system; Step 3: designing an improved chicken swarm algorithm; Step 4: designing a Gabor wavelet neural network; Step 5: optimizing the parameters of the Gabor wavelet neural network using the improved chicken swarm algorithm; and Step 6: testing a test sample using the trained optimal pipeline network integrity evaluation model.

[0006] However, in the process of implementing the technical solution of the invention in the embodiments of the present application, the present application found that the above technology has at least the following technical problems:

[0007] In the prior art, the amount of data processed by the mathematical model for calculating the corrosion rate based on the factors of pipeline damage is too small, errors are easily generated during the calculation process, and the potential pipeline damage behavior pattern cannot be calculated, resulting in a low ability to predict pipeline damage behavior. Summary of the invention

[0008] The embodiment of the present application solves the problem of low pipeline destructive behavior prediction ability in the prior art by providing a pipeline destructive behavior prediction and protection method based on an AI big model, and realizes the improvement of pipeline destructive behavior prediction ability by learning a large amount of data for training through the AI ​​big model.

[0009] The embodiment of the present application provides a pipeline destruction behavior prediction and protection method based on an AI large model, comprising the following steps: S1. Acquisition of gas pipeline operation data: Acquisition of historical operation data of each gas pipeline in a specified gas pipeline, and preprocessing of the historical operation data of the gas pipeline; S2. Analysis of gas pipeline data: Feature extraction and screening of gas pipeline destruction feature data based on the preprocessed historical operation data of the gas pipeline; S3. AI model training: Based on the gas pipeline destruction feature data, supervised learning technology is used to analyze and obtain pipeline feature index data; S4. Gas pipeline operation status detection: Based on the pipeline feature index, a pipeline status monitoring index is obtained, and the pipeline status monitoring index is used to determine whether the current gas pipeline is in an abnormal state, and an early warning is issued when the gas pipeline is in an abnormal state; S5. Improvement and optimization of the AI ​​model: Collect real-time operation data of the gas pipeline and repeat the above steps to optimize the AI ​​model.

[0010] Furthermore, the analysis process for preprocessing the historical operation data of the gas pipeline is: setting the monitoring time, dividing the historical operation data of the gas pipeline according to the monitoring time, and obtaining multiple groups of historical operation data of the gas pipeline; normalizing the historical operation data of the gas pipeline to obtain the preprocessed historical operation data of the gas pipeline.

[0011] Furthermore, the gas pipeline damage characteristic data includes appearance parameter data, corrosion data and pressure data; the pipeline characteristic index data includes appearance quality compliance index, corrosion damage probability index and pressure distribution status index; the appearance quality compliance index is used to describe the damage on the appearance of the gas pipeline; the corrosion damage probability index is used to reflect the degree of corrosion marks inside the gas pipeline; the pressure distribution status index is used to reflect the internal pressure of the gas pipeline.

[0012] Furthermore, the analysis process of the appearance quality compliance index is as follows: obtaining appearance parameter data, the appearance parameter data including data of the inner diameter of the gas pipeline, the outer diameter of the gas pipeline, the wall thickness of the gas pipeline and the length of the gas pipeline; obtaining appearance parameter correction factors, the appearance parameter correction factors including the gas pipeline diameter correction factor, the gas pipeline wall thickness correction factor and the gas pipeline length correction factor; according to the appearance parameter data and the appearance parameter correction factors, obtaining the appearance quality compliance index; the appearance quality compliance index is: Where W represents the appearance quality compliance index, i is the number of multiple sets of gas pipeline historical operation data, i = 1, 2, ..., n, n is the total number of multiple sets of gas pipeline historical operation data, ir i The inner diameter of the gas pipeline representing the historical operation data of the i-th group of gas pipelines, or i represents the outer diameter of the gas pipeline of the i-th group of gas pipeline historical operation data, r' represents the gas pipeline diameter correction factor, Indicates the standard value of the gas pipeline diameter for the historical operation data of the gas pipeline, H i represents the wall thickness of the gas pipeline of the i-th group of historical operation data of the gas pipeline, H' represents the correction factor of the wall thickness of the gas pipeline, Indicates the standard value of the gas pipeline wall thickness for the historical operation data of the gas pipeline, h i represents the length of the gas pipeline of the i-th group of historical operation data of the gas pipeline, h' represents the correction factor of the gas pipeline length, The standard value of the gas pipeline length representing the historical operation data of the gas pipeline.

[0013] Furthermore, the analysis process of the corrosion damage probability index is as follows: obtaining corrosion data, the corrosion data including corrosion rate, liquid water existence coefficient and corrosion depth coefficient; the corrosion rate is used to describe the degree of corrosion of the gas pipeline in unit time; the liquid water existence coefficient is used to describe the liquid water residue inside the gas pipeline; the corrosion depth coefficient is used to describe the degree of corrosion of the gas pipeline; according to the objective weight method, setting the weight ratio value of the liquid water existence coefficient for the corrosion damage probability index and the weight ratio value of the corrosion depth coefficient for the corrosion damage probability index; according to the liquid water existence coefficient, the weight ratio value of the liquid water existence coefficient for the corrosion damage probability index, the corrosion depth coefficient, and the weight ratio value of the corrosion depth coefficient for the corrosion damage probability index, the corrosion damage probability index is obtained.

[0014] Furthermore, the analysis process of the pressure distribution status index is as follows: obtain pressure data, the pressure data including threshold pressure, test pressure and bursting pressure; the threshold pressure is used to describe the maximum pressure that the gas pipeline can withstand under normal operation; the test pressure is used to describe the pressure applied to the gas pipeline during a pressure test; the bursting pressure is used to describe the bursting pressure that occurs when the gas pipeline exceeds its pressure bearing capacity; obtain a pressure correction factor; according to an objective weighting method, set a weight ratio of the threshold pressure to the pressure distribution status index, a weight ratio of the test pressure to the pressure distribution status index and a weight ratio of the bursting pressure to the pressure distribution status index; according to the threshold pressure, the weight ratio of the threshold pressure to the pressure distribution status index, the test pressure, the weight ratio of the test pressure to the pressure distribution status index, the bursting pressure, and the weight ratio of the bursting pressure to the pressure distribution status index, the pressure distribution status index is obtained.

[0015] Furthermore, the analysis process of the pipeline state monitoring index is: extracting the appearance quality compliance index, the corrosion damage probability index, and the pressure distribution status index, and the appearance quality compliance index, the corrosion damage probability index, and the pressure distribution status index are used to establish a pipeline state monitoring model; according to the objective weight method, the weight ratio of the appearance quality compliance index to the pipeline state monitoring index, the weight ratio of the corrosion damage probability index to the pipeline state monitoring index, and the weight ratio of the pressure distribution status index to the pipeline state monitoring index are set; according to the appearance quality compliance index, the weight ratio of the appearance quality compliance index to the pipeline state monitoring index, the corrosion damage probability index, the weight ratio of the corrosion damage probability index to the pipeline state monitoring index, the pressure distribution status index, and the weight ratio of the pressure distribution status index to the pipeline state monitoring index, the pipeline state monitoring index is obtained.

[0016] Furthermore, the calculation process of the pipeline status monitoring index is: obtaining the pipeline damage threshold; according to the appearance quality compliance index, the corrosion damage probability index, and the pressure distribution index; the pipeline status monitoring index is: Wherein, W represents the appearance quality compliance index, F represents the corrosion damage probability index, Y represents the pressure distribution index, e represents the natural constant, W' represents the weight ratio of the appearance quality compliance index to the pipeline condition monitoring index, F' represents the weight ratio of the corrosion damage probability index to the pipeline condition monitoring index, Y' represents the weight ratio of the pressure distribution index to the pipeline condition monitoring index, and G represents the pipeline condition monitoring index.

[0017] Furthermore, the analysis process for determining whether the current gas pipeline is in an abnormal state is as follows: obtaining a pipeline damage threshold value, which is used for comparing the pipeline state monitoring index to determine the pipeline operation state; when the pipeline state monitoring index is greater than the pipeline damage threshold value, it indicates that the pipeline operation is abnormal, and the pipeline state monitoring model issues an early warning; when the pipeline state monitoring index is not greater than the pipeline damage threshold value, it indicates that the pipeline operation is normal, and the pipeline state monitoring model does not issue an early warning.

[0018] Furthermore, the analysis process of collecting real-time operation data of the gas pipeline and repeating the above steps to optimize the AI ​​model is: training the pipeline status monitoring model based on the historical operation data of the gas pipeline; obtaining the real-time operation data of the gas pipeline, and repeatedly training the pipeline status monitoring model according to the real-time operation data of the gas pipeline.

[0019] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0020] 1. By obtaining the historical operation data of the gas pipeline, pre-processing the historical operation data of the gas pipeline and extracting features to obtain appearance parameter data, corrosion data and pressure data, the appearance parameter data, corrosion data and pressure data are trained based on supervised learning to obtain a pipeline status monitoring model, and the real-time operation data of the gas pipeline is obtained. The historical operation data of the gas pipeline is replaced with the real-time operation data of the gas pipeline. The pipeline status monitoring model is trained repeatedly to keep pace with the pipeline status and improve the pipeline status recognition ability of the pipeline status monitoring model. Then, a large amount of data is learned through the AI ​​large model for training, thereby improving the pipeline damage behavior prediction ability, which effectively solves the problem of low pipeline damage behavior prediction ability in the prior art.

[0021] 2. By extracting features from the pre-processed historical operating data of each gas pipeline in the designated gas pipeline, the gas pipeline damage feature data is obtained. The pipeline feature index data is obtained based on the analysis of the gas pipeline damage feature data, and the pipeline status monitoring model is trained with the pipeline feature index data, thereby achieving the goal of not having to modify the pipeline itself and reducing maintenance costs.

[0022] 3. By obtaining the pipeline damage threshold, when the pipeline status monitoring index is greater than the pipeline damage threshold, it means that the pipeline operation is abnormal, and the pipeline status monitoring model issues an early warning. When the pipeline status monitoring index is not greater than the pipeline damage threshold, it means that the pipeline operation is normal, and the pipeline status monitoring model does not issue an early warning, thereby performing real-time monitoring and early warning of the pipeline, thereby reducing human intervention and improving the safety of immediate discovery of pipeline damage. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 A flow chart of a pipeline damage behavior prediction and protection method based on an AI large model provided in an embodiment of the present application. DETAILED DESCRIPTION

[0024] The embodiment of the present application solves the problem of low pipeline destruction behavior prediction ability in the prior art by providing a pipeline destruction behavior prediction and protection method based on an AI large model. The method obtains historical operation data of the gas pipeline, pre-processes the historical operation data of the gas pipeline and extracts features to obtain appearance parameter data, corrosion data and pressure data. The appearance parameter data, corrosion data and pressure data are trained based on supervised learning to obtain a pipeline status monitoring model. The real-time operation data of the gas pipeline is obtained, and the historical operation data of the gas pipeline is replaced with the real-time operation data of the gas pipeline. The pipeline status monitoring model is trained repetitively to keep pace with the changes in the pipeline status, thereby improving the pipeline status recognition ability of the pipeline status monitoring model, and realizing the improvement of the pipeline destruction behavior prediction ability by learning a large amount of data for training through the AI ​​large model.

[0025] The technical solution in the embodiment of the present application is to solve the problem of low prediction ability of pipeline damage behavior. The overall idea is as follows:

[0026] By obtaining the historical operation data of the gas pipeline, the historical operation data of the gas pipeline is preprocessed to obtain complete and available data; based on the complete and available data, feature extraction is performed to obtain the gas pipeline damage feature data, and the gas pipeline damage feature data includes appearance parameter data, corrosion data and pressure data; based on the analysis of the gas pipeline damage feature data, the pipeline characteristic index data is obtained, and the pipeline status monitoring index is obtained based on the pipeline characteristic index data. The pipeline status monitoring index is used to determine whether the current gas pipeline is in an abnormal state, and to issue an early warning when the gas pipeline is in an abnormal state; the pipeline status monitoring index is obtained by analyzing the appearance quality compliance index, the corrosion damage probability index, and the pressure distribution index. The data is obtained by correlation analysis of gas pipeline damage feature data; the appearance quality compliance index is obtained by analyzing the appearance parameter data of external damage to the gas pipeline; the corrosion damage probability index is obtained by analyzing the corrosion data; the pressure distribution status is obtained by analyzing the pressure data; the pipeline status monitoring model trained by supervised learning can monitor the operating status of the gas pipeline in real time and issue an early warning when an abnormality is detected; the pipeline status monitoring model has supervised learning capabilities, and can continue to improve the training model while analyzing the real-time operation data of the pipeline in the later stage, thereby improving the pipeline damage prediction capability, achieving the goal of learning a large amount of pipeline operation data through the training model to improve the pipeline damage behavior prediction capability.

[0027] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0028] like Figure 1 As shown, it is a flow chart of a pipeline destruction behavior prediction and protection method based on an AI large model provided in an embodiment of the present application, and the method includes the following steps: S1. Gas pipeline operation data acquisition: acquiring the historical operation data of each gas pipeline in the specified gas pipeline, and preprocessing the historical operation data of the gas pipeline; S2. Gas pipeline data analysis: performing feature extraction and screening of gas pipeline destruction feature data based on the preprocessed historical operation data of the gas pipeline; S3. AI model training: based on the gas pipeline destruction feature data, using supervised learning technology to analyze and obtain pipeline feature index data; S4. Gas pipeline operation status detection: obtaining a pipeline status monitoring index based on the pipeline feature index, and the pipeline status monitoring index is used to determine whether the current gas pipeline is in an abnormal state, and to issue an early warning when the gas pipeline is in an abnormal state; S5. AI model improvement and optimization: collecting real-time operation data of the gas pipeline and repeating the above steps to optimize the AI ​​model.

[0029] Furthermore, a monitoring duration is set, and the historical operation data of the gas pipeline is divided according to the monitoring duration to obtain multiple groups of historical operation data of the gas pipeline; the historical operation data of the gas pipeline is normalized to obtain pre-processed historical operation data of the gas pipeline.

[0030] In this embodiment, multiple groups of gas pipeline historical operation data are preprocessed using the Z-score normalization method, and the specific analysis is as follows:

[0031] In the formula, x i is the i-th group of gas pipeline historical operation data, i=1,2,...,n, n represents the total number of multiple groups of gas pipeline historical operation data, and Z represents the standardized data of the gas pipeline historical operation data.

[0032] Furthermore, the pipeline characteristic index data includes appearance quality compliance index, corrosion damage probability index, and pressure distribution status index; the appearance quality compliance index is used to describe the damage on the appearance of the gas pipeline; the corrosion damage probability index is used to reflect the degree of corrosion marks inside the gas pipeline; the pressure distribution status index is used to reflect the internal pressure of the gas pipeline.

[0033] Further, the appearance parameter data is obtained, and the appearance parameter data includes data of the inner diameter of the gas pipeline, the outer diameter of the gas pipeline, the wall thickness of the gas pipeline and the length of the gas pipeline; the appearance parameter correction factor is obtained, and the appearance parameter correction factor includes a gas pipeline diameter correction factor, a gas pipeline wall thickness correction factor and a gas pipeline length correction factor; according to the appearance parameter data and the appearance parameter correction factor, the appearance quality compliance index is obtained; the appearance quality compliance index is: Where W represents the appearance quality compliance index, i is the number of multiple sets of gas pipeline historical operation data, i = 1, 2, ..., n, n is the total number of multiple sets of gas pipeline historical operation data, ir i The inner diameter of the gas pipeline representing the historical operation data of the i-th group of gas pipelines, or i represents the outer diameter of the gas pipeline of the i-th group of gas pipeline historical operation data, r' represents the gas pipeline diameter correction factor, Indicates the standard value of the gas pipeline diameter for the historical operation data of the gas pipeline, H i represents the wall thickness of the gas pipeline of the i-th group of historical operation data of the gas pipeline, H' represents the correction factor of the wall thickness of the gas pipeline, Indicates the standard value of the gas pipeline wall thickness for the historical operation data of the gas pipeline, h i represents the length of the gas pipeline of the i-th group of historical operation data of the gas pipeline, h' represents the gas pipeline length correction factor, The standard value of the gas pipeline length representing the historical operation data of the gas pipeline.

[0034] In this embodiment, the appearance quality compliance index analysis method includes but is not limited to infrared thermal imaging technology, laser scanning technology, ultrasonic technology and magnetic particle detection technology, wherein infrared thermal imaging technology is used to detect the temperature distribution on the surface of the gas pipeline to identify the problem, laser scanning technology is used to generate a three-dimensional model of the gas pipeline to detect surface defects, ultrasonic technology is used to detect the wall thickness and internal defects of the gas pipeline to evaluate the appearance quality, and magnetic particle detection technology is used to detect quality problems on the surface of the gas pipeline; the inner diameter of the gas pipeline is measured by a three-point contact inner diameter gauge. Compared with the traditional two-point contact inner diameter gauge, the three-point contact inner diameter gauge uses three measuring points, and the three measuring points are evenly distributed on the side head of the gas pipeline; the thickness of the gas pipeline is obtained by measuring the propagation time of the ultrasonic wave in the gas pipeline; the length of the gas pipeline is obtained by analyzing the reflection signal of the ultrasonic pulse sent by the ultrasonic wave inside the pipeline; the appearance parameter correction factor is obtained through the sensor calibration certificate.

[0035] Furthermore, corrosion data is obtained, and the corrosion data includes corrosion rate, liquid water existence coefficient and corrosion depth coefficient; the corrosion rate is used to describe the degree of corrosion of the gas pipeline in unit time; the liquid water existence coefficient is used to describe the liquid water residue inside the gas pipeline; the corrosion depth coefficient is used to describe the degree of corrosion of the gas pipeline; according to the objective weight method, the weight ratio value of the liquid water existence coefficient to the corrosion damage probability index and the weight ratio value of the corrosion depth coefficient to the corrosion damage probability index are set; according to the liquid water existence coefficient, the weight ratio value of the liquid water existence coefficient to the corrosion damage probability index, the corrosion depth coefficient, and the weight ratio value of the corrosion depth coefficient to the corrosion damage probability index, the corrosion damage probability index is obtained.

[0036] In this embodiment, the corrosion damage probability index analysis method includes but is not limited to ultrasonic testing, magnetic particle testing, magnetic flow testing and electrochemical corrosion monitoring, the corrosion rate analysis method includes but is not limited to using a corrosion detector to directly measure the corrosion condition of the gas pipeline surface, the temperature analysis method in the pipeline includes but is not limited to infrared thermal imaging technology, thermocouple and thermistor temperature sensors, ultrasonic technology and optical fiber temperature measurement technology, the liquid water existence coefficient analysis method includes but is not limited to multiphase flow model, sensor monitoring, microwave technology and acoustic wave technology, the surface air pressure analysis method includes but is not limited to barometer, smartphone application, weather station and satellite data, the corrosion depth coefficient analysis method includes but is not limited to visual detection and corrosion detector; the corrosion damage probability index is: fs=arctan(vf*t),f y represents the liquid water existence coefficient, f y ' represents the weight ratio of the liquid water existence coefficient to the corrosion damage probability index, f s represents the corrosion depth factor, f s' represents the weight ratio of the corrosion depth coefficient to the corrosion damage probability index, t represents the monitoring time, T g Indicates the temperature inside the pipe, P d represents the surface pressure, v f represents the corrosion rate, e represents the natural constant, F represents the corrosion damage probability index, P d =Liquid water exists when the pressure is 1000hPa.

[0037] Furthermore, pressure data is obtained, and the pressure data includes threshold pressure, test pressure and bursting pressure; the threshold pressure is used to describe the maximum pressure that the gas pipeline can withstand under normal operation; the test pressure is used to describe the pressure applied to the gas pipeline during a pressure test; the bursting pressure is used to describe the bursting pressure that occurs when the gas pipeline exceeds its pressure bearing capacity; the pressure correction factor is obtained; according to the objective weighting method, the weight ratio of the threshold pressure to the pressure distribution index, the weight ratio of the test pressure to the pressure distribution index and the weight ratio of the bursting pressure to the pressure distribution index are set; according to the threshold pressure, the weight ratio of the threshold pressure to the pressure distribution index, the test pressure, the weight ratio of the test pressure to the pressure distribution index, the bursting pressure, and the weight ratio of the bursting pressure to the pressure distribution index, the pressure distribution index is obtained.

[0038] In this embodiment, the pressure distribution index analysis method includes but is not limited to a pressure sensor network, a fluid mechanics simulation, and a remote monitoring system, wherein the pressure sensor network is used to monitor the pressure changes at different locations of the gas pipeline, record the pressure data in the gas pipeline and transmit it to the monitoring center for processing and analysis, the fluid mechanics simulation is used to model the gas pipeline system, thereby simulating and analyzing the pressure distribution inside the gas pipeline, and the remote monitoring system is used to transmit the pressure data inside the gas pipeline to the remote monitoring center to monitor the pressure condition of the gas pipeline in real time. The pressure distribution index is: In the formula, y y <0.2MPa,y y represents the threshold pressure, y y ' represents the weight ratio of the threshold pressure to the pressure distribution index, y b >4.0MPa,y b Indicates the bursting pressure, y b ' represents the weight ratio of burst pressure to pressure distribution index, y s Indicates the test pressure, y s ' represents the weight ratio of the test pressure to the pressure distribution index, and Y represents the pressure distribution index.

[0039] Furthermore, the appearance quality compliance index, corrosion damage probability index and pressure distribution status index are extracted, and the appearance quality compliance index, corrosion damage probability index and pressure distribution status index are used to establish a pipeline state monitoring model; according to the objective weight method, the weight ratio of the appearance quality compliance index to the pipeline state monitoring index, the weight ratio of the corrosion damage probability index to the pipeline state monitoring index and the weight ratio of the pressure distribution status index to the pipeline state monitoring index are set; according to the appearance quality compliance index, the weight ratio of the appearance quality compliance index to the pipeline state monitoring index, the corrosion damage probability index, the weight ratio of the corrosion damage probability index to the pipeline state monitoring index, the pressure distribution status index and the weight ratio of the pressure distribution status index to the pipeline state monitoring index, the pipeline state monitoring index is obtained.

[0040] Furthermore, the pipeline damage threshold is obtained; according to the appearance quality compliance index, corrosion damage probability index, and pressure distribution index; the pipeline status monitoring index can be obtained by sensor measurement and analysis methods, and can also be calculated by formula; the pipeline status monitoring index is: Wherein, W represents the appearance quality compliance index, F represents the corrosion damage probability index, Y represents the pressure distribution index, e represents the natural constant, W' represents the weight ratio of the appearance quality compliance index to the pipeline condition monitoring index, F' represents the weight ratio of the corrosion damage probability index to the pipeline condition monitoring index, Y' represents the weight ratio of the pressure distribution index to the pipeline condition monitoring index, and G represents the pipeline condition monitoring index.

[0041] Furthermore, a pipeline damage threshold is obtained and used to compare the pipeline status monitoring index to determine the pipeline operation status; when the pipeline status monitoring index is greater than the pipeline damage threshold, it indicates that the pipeline operation is abnormal, and the pipeline status monitoring model issues an early warning; when the pipeline status monitoring index is not greater than the pipeline damage threshold, it indicates that the pipeline operation is normal, and the pipeline status monitoring model does not issue an early warning.

[0042] In this embodiment, the pipeline damage threshold analysis method includes but is not limited to statistical analysis by numerical simulation software, based on the damage characteristic data of the simulated pipeline under different conditions, to determine the pipeline damage threshold; the pipeline damage threshold is: Where W represents the appearance quality compliance index, F represents the corrosion damage probability index, Y represents the pressure distribution index, e represents the natural constant, S y Indicates the pipeline damage threshold.

[0043] Furthermore, a pipeline status monitoring model is trained based on historical operation data of the gas pipeline; real-time operation data of the gas pipeline is obtained, and the pipeline status monitoring model is repeatedly trained according to the real-time operation data of the gas pipeline.

[0044] In this embodiment, the pipeline status monitoring model is trained repeatedly, learns according to the real-time operation data of the pipeline, and adjusts and optimizes according to the feedback, so as to keep pace with the pipeline status, thereby improving the pipeline status recognition ability of the pipeline status monitoring model and reducing the false alarm rate.

[0045] The technical scheme in the above-mentioned embodiment of the present application has at least the following technical effects or advantages: relative to a pipeline internal corrosion prediction method announced by the invention patent with publication number: CN109816133B, the embodiment of the present application extracts features from the historical operation data of each gas pipeline in the specified gas pipeline after pre-processing, obtains the gas pipeline damage feature data, and obtains the pipeline feature index data according to the analysis of the gas pipeline damage feature data, so as to train the pipeline state monitoring model with the pipeline feature index data, thereby achieving the goal of not needing to modify the pipeline itself and reducing maintenance costs; relative to a method for evaluating the integrity of an oil and gas pipeline network disclosed in the patent application with publication number: CN114066267A An intelligent method is provided. The embodiment of the present application obtains historical operation data of the gas pipeline, pre-processes the historical operation data of the gas pipeline and extracts features to obtain appearance parameter data, corrosion data and pressure data, trains the appearance parameter data, corrosion data and pressure data based on supervised learning to obtain a pipeline state monitoring model, obtains real-time operation data of the gas pipeline, replaces the historical operation data of the gas pipeline with the real-time operation data of the gas pipeline, and adopts repetitive training for the pipeline state monitoring model to keep pace with the changes of the pipeline state, thereby improving the pipeline state recognition ability of the pipeline state monitoring model, and further realizing the improvement of the pipeline destructive behavior prediction ability by learning a large amount of data through the AI ​​large model.

[0046] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0047] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0048] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0049] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0050] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0051] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A pipeline damage behavior prediction and protection method based on AI large model, characterized in that: The following steps are involved: S1. Acquisition of gas pipeline operation data: Acquisition of historical operation data of each gas pipeline in a designated gas pipeline, and preprocessing of the historical operation data of the gas pipeline; S2. Gas pipeline data analysis: Based on the pre-processed historical operation data of the gas pipeline, feature extraction is performed to screen the gas pipeline damage feature data; S3, AI model training: Based on the gas pipeline damage feature data, supervised learning technology is used to analyze and obtain pipeline feature index data; S4. Gas pipeline operation status detection: A pipeline status monitoring index is obtained based on the pipeline characteristic index. The pipeline status monitoring index is used to determine whether the current gas pipeline is in an abnormal state, and to issue an early warning when the gas pipeline is in an abnormal state; S5. AI model improvement and optimization: Collect real-time operation data of the gas pipeline and repeat the above steps to optimize the AI ​​model.

2. The pipeline damage behavior prediction and protection method based on the AI ​​large model as claimed in claim 1 is characterized in that: The analysis process of preprocessing the historical operation data of the gas pipeline is as follows: Set the monitoring time, divide the historical operation data of the gas pipeline according to the monitoring time, and obtain multiple groups of historical operation data of the gas pipeline; The historical operation data of the gas pipeline is normalized to obtain the preprocessed historical operation data of the gas pipeline.

3. The pipeline damage behavior prediction and protection method based on the AI ​​large model as claimed in claim 2 is characterized by: The gas pipeline damage characteristic data includes appearance parameter data, corrosion data and pressure data; The pipeline characteristic index data include appearance quality compliance index, corrosion damage probability index, and pressure distribution index; The appearance quality compliance index is used to describe the situation where the appearance of the gas pipeline is damaged; The corrosion damage probability index is used to reflect the extent of corrosion traces inside the gas pipeline; The pressure distribution index is used to reflect the internal pressure of the gas pipeline.

4. The pipeline damage behavior prediction and protection method based on the AI ​​large model as claimed in claim 3 is characterized in that: The analysis process of the appearance quality compliance index is as follows: Acquire appearance parameter data, wherein the appearance parameter data includes data of the inner diameter of the gas pipeline, the outer diameter of the gas pipeline, the wall thickness of the gas pipeline, and the length of the gas pipeline; Obtaining appearance parameter correction factors, wherein the appearance parameter correction factors include a gas pipeline diameter correction factor, a gas pipeline wall thickness correction factor, and a gas pipeline length correction factor; According to the appearance parameter data and the appearance parameter correction factor, the appearance quality compliance index is obtained; The appearance quality compliance index is: Where W represents the appearance quality compliance index, i is the number of multiple sets of gas pipeline historical operation data, i = 1, 2, ..., n, n is the total number of multiple sets of gas pipeline historical operation data, ir i The inner diameter of the gas pipeline representing the historical operation data of the i-th group of gas pipelines, or i represents the outer diameter of the gas pipeline of the i-th group of gas pipeline historical operation data, r' represents the gas pipeline diameter correction factor, Indicates the standard value of the gas pipeline diameter for the historical operation data of the gas pipeline, H i represents the wall thickness of the gas pipeline of the i-th group of historical operation data of the gas pipeline, H' represents the correction factor of the wall thickness of the gas pipeline, Indicates the standard value of the gas pipeline wall thickness for the historical operation data of the gas pipeline, h i represents the length of the gas pipeline of the i-th group of historical operation data of the gas pipeline, h' represents the gas pipeline length correction factor, The standard value of the gas pipeline length representing the historical operation data of the gas pipeline.

5. The pipeline damage behavior prediction and protection method based on AI large model as claimed in claim 3 is characterized in that: The analysis process of the corrosion damage probability index is: Acquiring corrosion data, wherein the corrosion data includes a corrosion rate, a liquid water existence coefficient, and a corrosion depth coefficient; The corrosion rate is used to describe the degree of corrosion of the gas pipeline per unit time; The liquid water existence coefficient is used to describe the liquid water remaining inside the gas pipeline; The corrosion depth coefficient is used to describe the degree of corrosion of the gas pipeline; According to the objective weight method, the weight ratio of the liquid water existence coefficient to the corrosion damage probability index and the weight ratio of the corrosion depth coefficient to the corrosion damage probability index are set; The corrosion damage probability index is obtained based on the liquid water existence coefficient, the weight ratio of the liquid water existence coefficient to the corrosion damage probability index, the corrosion depth coefficient, and the weight ratio of the corrosion depth coefficient to the corrosion damage probability index.

6. The pipeline damage behavior prediction and protection method based on the AI ​​large model as claimed in claim 3 is characterized in that: The analysis process of the pressure distribution index is as follows: Acquiring pressure data, wherein the pressure data includes a threshold pressure, a test pressure, and a burst pressure; The threshold pressure is used to describe the maximum pressure that the gas pipeline can withstand under normal operation; The test pressure is used to describe the pressure applied to the gas pipeline during the pressure test; The bursting pressure is used to describe the bursting pressure of a gas pipeline when it exceeds its pressure bearing capacity; Get the pressure correction factor; According to the objective weight method, the weight ratio of the threshold pressure to the pressure distribution index, the weight ratio of the test pressure to the pressure distribution index and the weight ratio of the burst pressure to the pressure distribution index are set; The pressure distribution index is obtained based on the threshold pressure, the weight ratio of the threshold pressure to the pressure distribution index, the test pressure, the weight ratio of the test pressure to the pressure distribution index, the bursting pressure, and the weight ratio of the bursting pressure to the pressure distribution index.

7. The pipeline damage behavior prediction and protection method based on the AI ​​large model as claimed in claim 3 is characterized in that: The analysis process of the pipeline status monitoring index is as follows: Extracting the appearance quality compliance index, the corrosion damage probability index, and the pressure distribution status index, wherein the appearance quality compliance index, the corrosion damage probability index, and the pressure distribution status index are used to establish a pipeline status monitoring model; According to the objective weight method, set the weight ratio of the appearance quality compliance index to the pipeline condition monitoring index, the weight ratio of the corrosion damage probability index to the pipeline condition monitoring index, and the weight ratio of the pressure distribution index to the pipeline condition monitoring index; The pipeline condition monitoring index is obtained based on the appearance quality compliance index, the weight ratio of the appearance quality compliance index to the pipeline condition monitoring index, the corrosion damage probability index, the weight ratio of the corrosion damage probability index to the pipeline condition monitoring index, the pressure distribution status index, and the weight ratio of the pressure distribution status index to the pipeline condition monitoring index.

8. The pipeline damage behavior prediction and protection method based on the AI ​​large model as claimed in claim 7 is characterized in that: The calculation process of the pipeline status monitoring index is: Get pipeline damage threshold; According to the appearance quality compliance index, corrosion damage probability index, and pressure distribution index; The pipeline status monitoring index is: Wherein, W represents the appearance quality compliance index, F represents the corrosion damage probability index, Y represents the pressure distribution index, e represents the natural constant, W' represents the weight ratio of the appearance quality compliance index to the pipeline condition monitoring index, F' represents the weight ratio of the corrosion damage probability index to the pipeline condition monitoring index, Y' represents the weight ratio of the pressure distribution index to the pipeline condition monitoring index, and G represents the pipeline condition monitoring index.

9. The pipeline damage behavior prediction and protection method based on AI large model as claimed in claim 1 is characterized in that: The analysis process for determining whether the current gas pipeline is in an abnormal state is as follows: Obtain pipeline damage threshold value, which is used for pipeline status monitoring index comparison to determine pipeline operation status; When the pipeline status monitoring index is greater than the pipeline damage threshold, it indicates that the pipeline is operating abnormally, and the pipeline status monitoring model issues an early warning; When the pipeline status monitoring index is not greater than the pipeline damage threshold, it means that the pipeline is operating normally and the pipeline status monitoring model does not issue an early warning.

10. The pipeline damage behavior prediction and protection method based on AI large model as claimed in claim 7 is characterized in that: The analysis process of collecting real-time operation data of the gas pipeline and repeating the above steps to optimize the AI ​​model is as follows: Training pipeline status monitoring model based on historical operation data of gas pipelines; Obtain real-time operation data of the gas pipeline and perform repetitive training on the pipeline status monitoring model based on the real-time operation data of the gas pipeline.

Images