Oil and gas pipeline leakage diagnosis method and system based on feature extraction

Abstract

The invention discloses an oil and gas pipeline leakage diagnosis method and system based on feature extraction, and relates to the technical field of oil and gas transmission. A leakage diffusion model is established according to basic data, and all oil and gas pipelines are classified and numbered according to the basic data; real-time operation data of the oil and gas pipeline are obtained through a sensor set arranged on the monitoring node, feature extraction is conducted on the real-time operation data, then an operation evaluation coefficient is calculated, and the operation evaluation coefficient is judged according to a preset operation evaluation threshold value to generate a leakage searching signal; the method comprises the following steps: acquiring a monitoring node coordinate generated when leakage is detected according to a leakage searching signal, further searching a corresponding leakage diffusion model, acquiring real-time update data at a target node to optimize the leakage diffusion model, and positioning a leakage site according to the optimized leakage diffusion model. And generating a visual reminding signal to mark on the pipeline three-dimensional model, and visually generating an oil and gas diffusion effect at the leakage site according to the leakage diffusion model.

Description

Technical Field

[0001] The present invention relates to the technical field of oil and gas transportation, and in particular to a method and system for diagnosing oil and gas pipeline leakage based on feature extraction. Background Art

[0002] Since the buried depth of oil and gas pipelines is usually about 1 to 2 meters, they are easily affected by natural disasters such as landslides, earthquakes, floods, or human factors such as construction activities and deliberate sabotage, resulting in pipeline damage or even oil and gas leakage. Oil and gas leakage will not only directly cause huge property losses, but also have serious impacts on the environment, such as soil and water pollution, ecosystem damage, etc. At the same time, there is a risk of ignition and explosion, which seriously threatens the safety of life and property. Oil and gas pipeline leakage detection technology can quickly identify leakage problems, accurately locate the leakage location, and provide timely response and treatment measures, which is of great significance to ensure the safe operation of pipelines.

[0003] At present, the mainstream oil and gas pipeline leakage detection technologies mainly include manual line patrol method, acoustic wave detection method, stress wave detection method, distributed optical fiber sensor detection method, etc. Although these oil and gas pipeline leakage detection technologies are widely used, they still have some shortcomings and deficiencies: 1) Poor continuity and real-time performance: Due to the long laying distance of oil and gas pipelines, wide coverage and complex working environment, some detection technologies such as manual line patrol method and acoustic wave detection method can only partially cover the pipeline, and cannot achieve real-time monitoring of the entire pipeline; 2) Poor positioning ability: Limited by the sensitivity and resolution of the sensor, the complexity of the pipeline, the characteristics of the leakage and environmental conditions, the existing leakage detection technology has low positioning accuracy for the leakage point. For example, the positioning accuracy of the acoustic wave detection method is only about 100 meters;

[0004] In view of the above-mentioned technical defects, a solution is now proposed. Summary of the invention

[0005] The purpose of the present invention is to locate the leakage site according to the optimized leakage diffusion model, generate a visual reminder signal to mark it on the pipeline three-dimensional model, and visualize the oil and gas diffusion effect at the leakage site according to the leakage diffusion model.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: a method and system for diagnosing oil and gas pipeline leakage based on feature extraction, comprising the following steps:

[0007] Step 1: Obtain and process the basic data of the oil and gas pipelines, and establish a leakage diffusion model based on the basic data. Classify and number each oil and gas pipeline according to the basic data, and store the corresponding leakage diffusion model according to the number;

[0008] Step 2: Establish a three-dimensional pipeline model based on the basic data of the oil and gas pipeline, and define a number of monitoring nodes on the three-dimensional pipeline model. According to the location of the monitoring node in the oil and gas pipeline, the monitoring node is also assigned a classification number mark, and the real-time operation data of the oil and gas pipeline is obtained through the sensor group set on the monitoring node;

[0009] Step 3: Acquire and process real-time operation data, extract features from the real-time operation data, and then calculate the operation evaluation coefficient, and determine the operation evaluation coefficient according to a preset operation evaluation threshold to generate a leak search signal;

[0010] Step 4: Obtain and process the leakage search signal, obtain the coordinates of the monitoring node where the leakage is detected according to the leakage search signal, and mark it as the target node, find the corresponding leakage diffusion model according to the number of the target node, obtain the real-time update data at the target node to optimize the leakage diffusion model, and locate the leakage site according to the optimized leakage diffusion model, and obtain the coordinates of the leakage site and send them to the trajectory output unit;

[0011] Step 5: Obtain the coordinates of the leakage site, generate a visual reminder signal and mark it on the three-dimensional model of the pipeline, and visualize the oil and gas diffusion effect at the leakage site according to the leakage diffusion model.

[0012] Furthermore, the specific process of establishing a leakage diffusion model based on basic data is as follows:

[0013] S101, obtaining basic data of the oil and gas pipeline, wherein the basic data includes diameter data of the oil and gas pipeline, thickness data of the oil and gas pipeline, and concentration data of the transport medium in the oil and gas pipeline;

[0014] S102, based on the basic data of the oil and gas pipeline, setting the physical constraints of the initial network model based on the fully connected neural network architecture, defining the loss function of the initial network model, and embedding the unsteady-state convection-diffusion equation of oil and gas pipeline leakage diffusion and its residual gradient into the loss function of the initial network model to obtain the leakage diffusion model of the oil and gas pipeline leakage;

[0015] S103, detecting leaked oil and gas in the leakage area when the oil and gas pipeline leaks and spreads by mid-infrared light waves, and integrating multiple groups of concentration grayscale images into a data set as training samples based on the concentration grayscale image of the leakage area when the oil and gas pipeline leaks and spreads;

[0016] S104, dividing the generated training samples into a training set and a test set according to a ratio of 8:2;

[0017] S105, downloading the weight file and loading it onto the corresponding network to initialize the migration network parameters, and determining the number of hidden layer nodes of the BP neural network model according to the number of normal images in the training set;

[0018] S106. Modify the last fully connected layer of the network, keep the input unchanged, set the output to the maximum diameter of the leakage area, initialize the weight of the last layer, use the gradient descent algorithm for learning, and use fixed step size decay to optimize the training parameters, retrain the entire network, and obtain the anomaly recognition model;

[0019] S305, during the training process, a small batch of normal images are randomly and non-repeatedly extracted from the training set. Extracting all the normal images in the training set is considered a training cycle. The training is completed after a certain period of iteration, and then the test set is used to evaluate the effect of the leakage diffusion model.

[0020] Furthermore, the sensor group includes a pressure sensor and a flow sensor arranged on the inner wall of the oil and gas pipeline, and a temperature sensor and an acoustic wave sensor arranged on the outer wall of the oil and gas pipeline.

[0021] Furthermore, the specific process of generating a leak search signal is as follows:

[0022] S201, acquiring real-time operation data, wherein the real-time operation data includes real-time pressure data Fi, real-time flow data Qi, real-time temperature data Ti and real-time sound wave data;

[0023] S202, extracting features from real-time acoustic wave data to obtain time-domain feature values ​​of the acoustic wave data including amplitude A, zero-crossing rate η and signal energy E, and calculating the acoustic wave characteristic coefficient Ui of the oil and gas pipeline according to the following formula: , where e1, e2 and e3 are preset weight coefficients, and the acoustic wave characteristic coefficient is used to reflect the comprehensive performance of various time domain characteristic values ​​in real-time acoustic wave data;

[0024] S203, and then calculate the operation evaluation coefficient Wi according to the following formula: , where α, β, and γ are preset proportional coefficients, is the standard pressure value of the oil and gas pipeline under normal conditions, is the standard flow value of the oil and gas pipeline under normal conditions, is the standard temperature value of the oil and gas pipeline under normal conditions, is the normal characteristic coefficient value of the acoustic wave data under the normal state of the oil and gas pipeline. The operation evaluation coefficient is used to reflect the actual operation state of the oil and gas pipeline. The smaller the operation evaluation coefficient is, the closer the operation state of the oil and gas pipeline is to the normal state. On the contrary, the larger the operation evaluation coefficient is, the closer the operation state of the oil and gas pipeline is to the abnormal state.

[0025] S204, obtaining a preset operation evaluation threshold, if the operation evaluation coefficient is greater than or equal to the operation evaluation threshold, searching for a corresponding monitoring node according to the source of the real-time operation data, and generating a leakage search signal in combination with the coordinates of the monitoring node.

[0026] Furthermore, the specific process of obtaining the coordinates of the leakage site is as follows:

[0027] S301, searching for a corresponding leakage diffusion model from a model storage unit according to the serial number of a target node;

[0028] S302, obtaining real-time updated data at the target node, wherein the real-time updated data includes real-time data of the oil and gas pipeline at the target node and updated concentration data of the transport medium in the oil and gas pipeline, and reconstructing the oil and gas concentration field during the leakage and diffusion of the oil and gas pipeline based on the updated concentration data of the transport medium in the oil and gas pipeline to obtain a simulation diagram of the oil and gas concentration field;

[0029] S304, acquiring real-time data of the oil and gas pipeline, obtaining the oil and gas concentration field when the oil and gas pipeline diffuses based on the leakage diffusion model, visualizing the oil and gas concentration field when the oil and gas pipeline diffuses, and intercepting several groups of oil and gas concentration field prediction maps one by one along the distribution position of the oil and gas pipeline;

[0030] S305, comparing the oil and gas concentration field simulation map with several groups of oil and gas concentration field prediction maps, obtaining an oil and gas concentration field prediction map that is consistent with the oil and gas concentration field simulation map, and the location of the oil and gas concentration field prediction map is the leakage site, and the coordinates of the leakage site are obtained.

[0031] The present invention also provides an oil and gas pipeline leakage diagnosis system based on feature extraction, a model storage unit, a data monitoring unit, a data analysis unit, a leakage diagnosis unit and a trajectory output unit;

[0032] The model storage unit is used to obtain and process the basic data of the oil and gas pipelines, establish a leakage diffusion model based on the basic data, classify and number each oil and gas pipeline based on the basic data, and store the corresponding leakage diffusion model according to the number;

[0033] The data monitoring unit is used to establish a three-dimensional model of the pipeline according to the basic data of the oil and gas pipeline, and to define a number of monitoring nodes on the three-dimensional model of the pipeline. According to the location of the monitoring node in the oil and gas pipeline, the monitoring node is also assigned a classification number mark, and the real-time operation data of the oil and gas pipeline is obtained through the sensor group set on the monitoring node, and the real-time operation data is sent to the data analysis unit;

[0034] The data analysis unit is used to obtain and process real-time operation data, extract features from the real-time operation data, and then calculate the operation evaluation coefficient, and determine the operation evaluation coefficient according to a preset operation evaluation threshold to generate a leak search signal and send it to the leak diagnosis unit;

[0035] The leakage diagnosis unit is used to obtain and process the leakage search signal, obtain the coordinates of the monitoring node where the leakage is detected according to the leakage search signal, and mark it as the target node, search the corresponding leakage diffusion model from the model storage unit according to the number of the target node, obtain the real-time update data at the target node to optimize the leakage diffusion model, and locate the leakage site according to the optimized leakage diffusion model, and obtain the coordinates of the leakage site and send them to the trajectory output unit;

[0036] The trajectory output unit is used to obtain the coordinates of the leakage site, generate a visual reminder signal to mark it on the three-dimensional model of the pipeline, and visualize the oil and gas diffusion effect at the leakage site according to the leakage diffusion model.

[0037] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0038] The oil and gas pipeline leakage diagnosis method and system based on feature extraction establish a leakage diffusion model according to basic data, classify and number each oil and gas pipeline according to the basic data, and obtain real-time operation data of the oil and gas pipeline through a sensor group arranged on a monitoring node, extract features from the real-time operation data, and then calculate the operation evaluation coefficient, judge the operation evaluation coefficient according to a preset operation evaluation threshold to generate a leakage search signal, obtain the coordinates of the monitoring node where the leakage is detected according to the leakage search signal, and then find the corresponding leakage diffusion model, obtain real-time updated data at the target node to optimize the leakage diffusion model, and locate the leakage site according to the optimized leakage diffusion model, generate a visual reminder signal to mark it on the pipeline three-dimensional model, and visualize the oil and gas diffusion effect at the leakage site according to the leakage diffusion model. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 A schematic diagram of the method flow of the present invention is shown;

[0040] Figure 2 A schematic diagram of the system structure of the present invention is shown. DETAILED DESCRIPTION

[0041] 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.

[0042] Embodiment 1:

[0043] like Figure 1 As shown, a method for diagnosing oil and gas pipeline leakage based on feature extraction comprises the following steps:

[0044] Step 1: Obtain and process the basic data of the oil and gas pipelines, and establish a leakage diffusion model based on the basic data. Classify and number each oil and gas pipeline according to the basic data, and store the corresponding leakage diffusion model according to the number;

[0045] The specific process of establishing a leakage diffusion model based on basic data is as follows:

[0046] S101, obtaining basic data of the oil and gas pipeline, the basic data including diameter data of the oil and gas pipeline, thickness data of the oil and gas pipeline, and concentration data of the transport medium in the oil and gas pipeline;

[0047] S102, based on the basic data of the oil and gas pipeline, setting the physical constraints of the initial network model based on the fully connected neural network architecture, defining the loss function of the initial network model, and embedding the unsteady-state convection diffusion equation of oil and gas pipeline leakage diffusion and its residual gradient into the loss function of the initial network model to obtain the leakage diffusion model of the oil and gas pipeline leakage;

[0048] S103, detecting leaked oil and gas in the leakage area when the oil and gas pipeline leaks and spreads by mid-infrared light waves, and integrating multiple groups of concentration grayscale images into a data set as training samples based on the concentration grayscale image of the leakage area when the oil and gas pipeline leaks and spreads;

[0049] S104, dividing the generated training samples into a training set and a test set according to a ratio of 8:2;

[0050] S105, downloading the weight file and loading it onto the corresponding network to initialize the migration network parameters, and determining the number of hidden layer nodes of the BP neural network model according to the number of normal images in the training set;

[0051] S106. Modify the last fully connected layer of the network, keep the input unchanged, set the output to the maximum diameter of the leakage area, initialize the weight of the last layer, use the gradient descent algorithm for learning, and use fixed step size decay to optimize the training parameters, retrain the entire network, and obtain the anomaly recognition model;

[0052] S305, during the training process, a small batch of normal images are randomly and non-repeatedly extracted from the training set. Extracting all the normal images in the training set is considered a training cycle. The training is completed after a certain period of iteration, and then the test set is used to evaluate the effect of the leakage diffusion model.

[0053] Step 2: Establish a three-dimensional pipeline model based on the basic data of the oil and gas pipeline, and define a number of monitoring nodes on the three-dimensional pipeline model. According to the location of the monitoring node in the oil and gas pipeline, the monitoring node is also assigned a classification number mark, and the real-time operation data of the oil and gas pipeline is obtained through the sensor group set on the monitoring node;

[0054] The sensor group includes a pressure sensor and a flow sensor arranged on the inner wall of the oil and gas pipeline, and a temperature sensor and an acoustic wave sensor arranged on the outer wall of the oil and gas pipeline.

[0055] Step 3: Acquire and process real-time operation data, extract features from the real-time operation data, and then calculate the operation evaluation coefficient, and determine the operation evaluation coefficient according to a preset operation evaluation threshold to generate a leak search signal;

[0056] The specific process of generating a leak search signal is as follows:

[0057] S201, acquiring real-time operation data, where the real-time operation data includes real-time pressure data Fi, real-time flow data Qi, real-time temperature data Ti and real-time sound wave data;

[0058] S202, extracting features from real-time acoustic wave data to obtain time-domain feature values ​​of the acoustic wave data including amplitude A, zero-crossing rate η and signal energy E, and calculating the acoustic wave characteristic coefficient Ui of the oil and gas pipeline according to the following formula: , where e1, e2 and e3 are preset weight coefficients, and the acoustic wave characteristic coefficient is used to reflect the comprehensive performance of various time domain characteristic values ​​in real-time acoustic wave data;

[0059] S203, and then calculate the operation evaluation coefficient Wi according to the following formula: , where α, β, and γ are preset proportional coefficients, is the standard pressure value of the oil and gas pipeline under normal conditions, is the standard flow value of the oil and gas pipeline under normal conditions, is the standard temperature value of the oil and gas pipeline under normal conditions, is the normal characteristic coefficient value of the acoustic wave data under the normal state of the oil and gas pipeline. The operation evaluation coefficient is used to reflect the actual operation state of the oil and gas pipeline. The smaller the operation evaluation coefficient is, the closer the operation state of the oil and gas pipeline is to the normal state. On the contrary, the larger the operation evaluation coefficient is, the closer the operation state of the oil and gas pipeline is to the abnormal state.

[0060] S204, obtaining a preset operation evaluation threshold, if the operation evaluation coefficient is greater than or equal to the operation evaluation threshold, searching for a corresponding monitoring node according to the source of the real-time operation data, and generating a leakage search signal in combination with the coordinates of the monitoring node.

[0061] Step 4: Obtain and process the leakage search signal, obtain the coordinates of the monitoring node where the leakage is detected according to the leakage search signal, and mark it as the target node, find the corresponding leakage diffusion model according to the number of the target node, obtain the real-time update data at the target node to optimize the leakage diffusion model, and locate the leakage site according to the optimized leakage diffusion model, and obtain the coordinates of the leakage site and send them to the trajectory output unit;

[0062] The specific process of obtaining the coordinates of the leakage site is as follows:

[0063] S301, searching for a corresponding leakage diffusion model from a model storage unit according to the serial number of a target node;

[0064] S302, obtaining real-time updated data at the target node, the real-time updated data including real-time data of the oil and gas pipeline at the target node and updated concentration data of the transport medium in the oil and gas pipeline, reconstructing the oil and gas concentration field during the leakage and diffusion of the oil and gas pipeline based on the updated concentration data of the transport medium in the oil and gas pipeline, and obtaining a simulation diagram of the oil and gas concentration field;

[0065] S304, acquiring real-time data of the oil and gas pipeline, obtaining the oil and gas concentration field when the oil and gas pipeline diffuses based on the leakage diffusion model, visualizing the oil and gas concentration field when the oil and gas pipeline diffuses, and intercepting several groups of oil and gas concentration field prediction maps one by one along the distribution position of the oil and gas pipeline;

[0066] S305, comparing the oil and gas concentration field simulation map with several groups of oil and gas concentration field prediction maps, obtaining an oil and gas concentration field prediction map that is consistent with the oil and gas concentration field simulation map, and the location of the oil and gas concentration field prediction map is the leakage site, and the coordinates of the leakage site are obtained.

[0067] Step 5: Obtain the coordinates of the leakage site, generate a visual reminder signal and mark it on the three-dimensional model of the pipeline, and visualize the oil and gas diffusion effect at the leakage site according to the leakage diffusion model.

[0068] The present invention establishes a leakage diffusion model according to basic data, classifies and numbers each oil and gas pipeline according to the basic data, obtains real-time operation data of the oil and gas pipeline through a sensor group arranged on a monitoring node, extracts features from the real-time operation data, and then calculates an operation evaluation coefficient, determines the operation evaluation coefficient according to a preset operation evaluation threshold to generate a leakage search signal, obtains the coordinates of the monitoring node where the leakage is detected according to the leakage search signal, and then searches for the corresponding leakage diffusion model, obtains real-time updated data at the target node to optimize the leakage diffusion model, locates the leakage site according to the optimized leakage diffusion model, generates a visual reminder signal to mark it on the three-dimensional model of the pipeline, and visually generates the oil and gas diffusion effect at the leakage site according to the leakage diffusion model.

[0069] Embodiment 2:

[0070] like Figure 2 As shown, the present invention also provides an oil and gas pipeline leakage diagnosis system based on feature extraction, including a model storage unit, a data monitoring unit, a data analysis unit, a leakage diagnosis unit and a trajectory output unit;

[0071] The model storage unit is used to obtain and process the basic data of the oil and gas pipelines, establish a leakage diffusion model based on the basic data, classify and number each oil and gas pipeline based on the basic data, and store the corresponding leakage diffusion model according to the number;

[0072] The data monitoring unit is used to establish a three-dimensional model of the pipeline according to the basic data of the oil and gas pipeline, and to define a number of monitoring nodes on the three-dimensional model of the pipeline. According to the location of the monitoring node in the oil and gas pipeline, the monitoring node is also assigned a classification number mark, and the real-time operation data of the oil and gas pipeline is obtained through the sensor group set on the monitoring node, and the real-time operation data is sent to the data analysis unit;

[0073] The data analysis unit is used to obtain and process real-time operation data, extract features from the real-time operation data, and then calculate the operation evaluation coefficient, and determine the operation evaluation coefficient according to a preset operation evaluation threshold to generate a leak search signal and send it to the leak diagnosis unit;

[0074] The leakage diagnosis unit is used to obtain and process the leakage search signal, obtain the coordinates of the monitoring node where the leakage is detected according to the leakage search signal, and mark it as the target node, search the corresponding leakage diffusion model from the model storage unit according to the number of the target node, obtain the real-time update data at the target node to optimize the leakage diffusion model, and locate the leakage site according to the optimized leakage diffusion model, and obtain the coordinates of the leakage site and send them to the trajectory output unit;

[0075] The trajectory output unit is used to obtain the coordinates of the leakage site, generate a visual reminder signal to mark it on the three-dimensional model of the pipeline, and visualize the oil and gas diffusion effect at the leakage site according to the leakage diffusion model.

[0076] The threshold is set to facilitate comparison. The size of the threshold depends on the amount of sample data and the number of bases set by technicians in this field for each group of sample data; as long as it does not affect the proportional relationship between the parameter and the quantized value.

[0077] The above formulas are all dimensionless and numerical calculations. The formula is a formula obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0078] In the two embodiments provided in the present application, it should be understood that the disclosed method and system can be implemented in other ways; for example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed; another point, the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the modules can be electrical or other forms;

[0079] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for diagnosing oil and gas pipeline leakage based on feature extraction, characterized in that: The following steps are involved: Step 1: Obtain and process the basic data of the oil and gas pipelines, and establish a leakage diffusion model based on the basic data. Classify and number each oil and gas pipeline according to the basic data, and store the corresponding leakage diffusion model according to the number; Step 2: Establish a three-dimensional pipeline model based on the basic data of the oil and gas pipeline, and define a number of monitoring nodes on the three-dimensional pipeline model. According to the location of the monitoring node in the oil and gas pipeline, the monitoring node is also assigned a classification number mark, and the real-time operation data of the oil and gas pipeline is obtained through the sensor group set on the monitoring node; Step 3: Acquire and process real-time operation data, extract features from the real-time operation data, and then calculate the operation evaluation coefficient, and determine the operation evaluation coefficient according to a preset operation evaluation threshold to generate a leak search signal; Step 4: Obtain and process the leakage search signal, obtain the coordinates of the monitoring node where the leakage is detected according to the leakage search signal, and mark it as the target node, find the corresponding leakage diffusion model according to the number of the target node, obtain the real-time update data at the target node to optimize the leakage diffusion model, and locate the leakage site according to the optimized leakage diffusion model, and obtain the coordinates of the leakage site and send them to the trajectory output unit; Step 5: Obtain the coordinates of the leakage site, generate a visual reminder signal and mark it on the three-dimensional model of the pipeline, and visualize the oil and gas diffusion effect at the leakage site according to the leakage diffusion model.

2. The oil and gas pipeline leakage diagnosis method based on feature extraction according to claim 1 is characterized in that: The specific process of establishing a leakage diffusion model based on basic data is as follows: S101, obtaining basic data of the oil and gas pipeline, wherein the basic data includes diameter data of the oil and gas pipeline, thickness data of the oil and gas pipeline, and concentration data of the transport medium in the oil and gas pipeline; S102, based on the basic data of the oil and gas pipeline, setting the physical constraints of the initial network model based on the fully connected neural network architecture, defining the loss function of the initial network model, and embedding the unsteady-state convection-diffusion equation of oil and gas pipeline leakage diffusion and its residual gradient into the loss function of the initial network model to obtain the leakage diffusion model of the oil and gas pipeline leakage; S103, detecting leaked oil and gas in the leakage area when the oil and gas pipeline leaks and spreads by mid-infrared light waves, and integrating multiple groups of concentration grayscale images into a data set as training samples based on the concentration grayscale image of the leakage area when the oil and gas pipeline leaks and spreads; S104, dividing the generated training samples into a training set and a test set according to a ratio of 8:2; S105, downloading the weight file and loading it onto the corresponding network to initialize the migration network parameters, and determining the number of hidden layer nodes of the BP neural network model according to the number of normal images in the training set; S106. Modify the last fully connected layer of the network, keep the input unchanged, set the output to the maximum diameter of the leakage area, initialize the weight of the last layer, use the gradient descent algorithm for learning, and use fixed step size decay to optimize the training parameters, retrain the entire network, and obtain the anomaly recognition model; S305, during the training process, a small batch of normal images are randomly and non-repeatedly extracted from the training set. Extracting all the normal images in the training set is considered a training cycle. The training is completed after a certain period of iteration, and then the test set is used to evaluate the effect of the leakage diffusion model.

3. The oil and gas pipeline leakage diagnosis method based on feature extraction according to claim 1 is characterized in that: The sensor group includes a pressure sensor and a flow sensor arranged on the inner wall of the oil and gas pipeline, and a temperature sensor and an acoustic wave sensor arranged on the outer wall of the oil and gas pipeline.

4. The oil and gas pipeline leakage diagnosis method based on feature extraction according to claim 1 is characterized in that: The specific process of generating a leak search signal is as follows: S201, acquiring real-time operation data, wherein the real-time operation data includes real-time pressure data Fi, real-time flow data Qi, real-time temperature data Ti and real-time sound wave data; S202, extracting features from real-time acoustic wave data to obtain time-domain feature values ​​of the acoustic wave data including amplitude A, zero-crossing rate η and signal energy E, and calculating the acoustic wave characteristic coefficient Ui of the oil and gas pipeline according to the following formula: , where e1, e2 and e3 are preset weight coefficients, and the acoustic wave characteristic coefficient is used to reflect the comprehensive performance of various time domain characteristic values ​​in real-time acoustic wave data; S203, and then calculate the operation evaluation coefficient Wi according to the following formula: , where α, β, and γ are preset proportional coefficients. is the standard pressure value of the oil and gas pipeline under normal conditions, is the standard flow value of the oil and gas pipeline under normal conditions, is the standard temperature value of the oil and gas pipeline under normal conditions, is the normal characteristic coefficient value of the acoustic wave data of the oil and gas pipeline under normal conditions; S204, obtaining a preset operation evaluation threshold, if the operation evaluation coefficient is greater than or equal to the operation evaluation threshold, searching for a corresponding monitoring node according to the source of the real-time operation data, and generating a leakage search signal in combination with the coordinates of the monitoring node.

5. The oil and gas pipeline leakage diagnosis method based on feature extraction according to claim 1 is characterized in that: The specific process of obtaining the coordinates of the leakage site is as follows: S301, searching for a corresponding leakage diffusion model from a model storage unit according to the serial number of a target node; S302, obtaining real-time updated data at the target node, wherein the real-time updated data includes real-time data of the oil and gas pipeline at the target node and updated concentration data of the transport medium in the oil and gas pipeline, and reconstructing the oil and gas concentration field during the leakage and diffusion of the oil and gas pipeline based on the updated concentration data of the transport medium in the oil and gas pipeline to obtain a simulation diagram of the oil and gas concentration field; S304, acquiring real-time data of the oil and gas pipeline, obtaining the oil and gas concentration field when the oil and gas pipeline diffuses based on the leakage diffusion model, visualizing the oil and gas concentration field when the oil and gas pipeline diffuses, and intercepting several groups of oil and gas concentration field prediction maps one by one along the distribution position of the oil and gas pipeline; S305, comparing the oil and gas concentration field simulation map with several groups of oil and gas concentration field prediction maps, obtaining an oil and gas concentration field prediction map that is consistent with the oil and gas concentration field simulation map, and the location of the oil and gas concentration field prediction map is the leakage site, and the coordinates of the leakage site are obtained.

6. An oil and gas pipeline leakage diagnosis system based on feature extraction, characterized in that: Model storage unit, data monitoring unit, data analysis unit, leakage diagnosis unit and trajectory output unit; The model storage unit is used to obtain and process the basic data of the oil and gas pipelines, establish a leakage diffusion model based on the basic data, classify and number each oil and gas pipeline based on the basic data, and store the corresponding leakage diffusion model according to the number; The data monitoring unit is used to establish a three-dimensional model of the pipeline according to the basic data of the oil and gas pipeline, and to define a number of monitoring nodes on the three-dimensional model of the pipeline. According to the location of the monitoring node in the oil and gas pipeline, the monitoring node is also assigned a classification number mark, and the real-time operation data of the oil and gas pipeline is obtained through the sensor group set on the monitoring node, and the real-time operation data is sent to the data analysis unit; The data analysis unit is used to obtain and process real-time operation data, extract features from the real-time operation data, and then calculate the operation evaluation coefficient, and determine the operation evaluation coefficient according to a preset operation evaluation threshold to generate a leak search signal and send it to the leak diagnosis unit; The leakage diagnosis unit is used to obtain and process the leakage search signal, obtain the coordinates of the monitoring node where the leakage is detected according to the leakage search signal, and mark it as the target node, search the corresponding leakage diffusion model from the model storage unit according to the number of the target node, obtain the real-time update data at the target node to optimize the leakage diffusion model, and locate the leakage site according to the optimized leakage diffusion model, and obtain the coordinates of the leakage site and send them to the trajectory output unit; The trajectory output unit is used to obtain the coordinates of the leakage site, generate a visual reminder signal to mark it on the three-dimensional model of the pipeline, and visualize the oil and gas diffusion effect at the leakage site according to the leakage diffusion model.

Images