A method and system for evaluating high-consequence areas of oil and gas pipelines based on machine vision

Abstract

The present invention relates to the field of intelligent monitoring technology, and discloses a method and system for evaluating high-consequence areas of oil and gas pipelines based on machine vision, the method comprising: setting a number of information collection points in the area to be monitored, deploying CCD cameras to collect real-time image data, and constructing an image data set; comparing with a standard image, and splitting the image data set to obtain a defect image data chain and a normal image data chain; splitting the defect image data chain into a number of sub-defect image data chains; analyzing and processing the defect image data chain to obtain a defect assessment value, and determining the allocation weight, and finally obtaining the overall defect assessment value of the area to be monitored; collecting dynamic parameters of the area to be monitored, determining the risk valuation of the area to be monitored based on the dynamic parameters and the overall defect assessment value, and issuing an early warning based on the risk valuation. This application improves the intelligence level of pipeline monitoring and the reliability of early warning, and solves the problems existing in traditional monitoring methods.

Description

Technical Field

[0001] The present invention relates to the field of intelligent monitoring technology, and in particular to a method and system for evaluating high-consequence areas of oil and gas pipelines based on machine vision. Background Art

[0002] With the development of industrial automation and intelligence, the safe operation and quality control of pipelines have become key issues. In order to ensure that pipelines can operate safely and for a long time, it is often necessary to monitor and analyze the operating status of pipelines in real time. Pipeline damage can lead to serious accidents such as leakage and explosion. In a complex industrial environment, the safety and stability of the pipeline system are crucial factors in the production process, especially in the fields of oil, natural gas, and chemicals. Therefore, how to effectively monitor the operating status of pipelines and take preventive measures in time at an early stage has become a key issue in pipeline safety management.

[0003] However, in actual industrial environments, the operating environment of pipelines is relatively harsh and requires manual inspection. The manual inspection cycle is long, and it is difficult to achieve real-time monitoring. There is a lag, and due to factors such as fatigue and negligence of inspectors, it may not be possible to discover subtle defects in the equipment in the early stage, resulting in the failure to timely check hidden dangers. Although sensors are installed in certain key locations, they can only monitor a specific type of parameter (such as temperature, vibration, etc.) and cannot fully reflect the defects of the pipeline.

[0004] Therefore, it is necessary to design a high consequence area assessment method and system for oil and gas pipelines based on machine vision to solve the problems existing in current technology. Summary of the invention

[0005] In view of this, the present invention proposes a method and system for evaluating high consequence areas of oil and gas pipelines based on machine vision, aiming to solve the problems existing in current oil and gas pipeline detection technology that rely on manual monitoring or single sensor monitoring, resulting in low monitoring efficiency and poor effect.

[0006] The present invention proposes a method for evaluating high-consequence areas of oil and gas pipelines based on machine vision, comprising:

[0007] Set up several information collection points in the area to be monitored, deploy a CCD camera at each collection point, collect real-time image data at each collection point, and build an image data set based on all real-time image data;

[0008] Compare each real-time image data in the image data set with the standard image respectively, split the image data set according to the comparison result, and obtain a defect image data chain and a normal image data chain;

[0009] Analyze each real-time image data in the defect image data chain based on edge recognition, and split the defect image data chain into a plurality of sub-defect image data chains according to defect types;

[0010] Analyze and process each sub-defect image data chain to obtain a defect assessment value of each sub-defect image data chain, determine an allocation weight according to the sub-defect image data chain, and obtain an overall defect assessment value of the area to be monitored according to the allocation weight and all defect assessment values;

[0011] The dynamic parameters of the area to be monitored are collected, the risk valuation of the area to be monitored is determined according to the dynamic parameters and the overall defect assessment value, and an early warning is issued to the area to be monitored according to the risk valuation.

[0012] Furthermore, the image data set is split according to the comparison result to obtain a defect image data chain and a normal image data chain, including:

[0013] Acquire a first initial blank data chain and a second initial blank data chain, wherein each initial blank data chain is provided with a plurality of data chain nodes and connection nodes;

[0014] Comparing the data volume of the real-time image data in the image data set with the data volume of the standard image, and generating a link switching mark when the difference between the data volume of the real-time image data and the data volume of the standard image is greater than a times the data volume of the standard image;

[0015] Transferring all real-time image data carrying the chain change mark to the first initial blank data chain to obtain the defective image data chain;

[0016] The unlabeled real-time image data in the image data set is transferred to the second initial blank data chain to obtain the normal image data chain.

[0017] Furthermore, when the defect image data chain is split into a plurality of sub-defect image data chains according to defect types, it includes:

[0018] Convert each real-time image data in the defect image data chain into a grayscale image;

[0019]

[0020] Using Gaussian filtering to smooth the grayscale image;

[0021]

[0022] The Sobel operator is used to calculate the horizontal and vertical gradients of the smoothed image, and the Canny algorithm is used to extract the image edge;

[0023]

[0024]

[0025] Acquire the contour of the edge image based on contour detection, extract the contour area, contour perimeter and contour aspect ratio from each contour, and determine the defect type according to the contour area, contour perimeter and contour aspect ratio;

[0026] in, represents the gray value at position (x, y), , , Represent the pixel values ​​of the red, green, and blue channels at position (x, y), respectively. represents the pixel value of the smoothed image at position (x, y), k represents the radius of the Gaussian kernel, represents the value of the Gaussian kernel, represents the processed grayscale image, represents the horizontal gradient of the image at position (x, y), Represents the vertical gradient of the image at position (x, y).

[0027] Further, when determining the defect type according to the contour area, contour perimeter and contour aspect ratio, it includes:

[0028] When the contour area is less than or equal to the area threshold and the contour perimeter is greater than the perimeter threshold, it is determined to be a scratch;

[0029] When the contour is curved and the contour area is greater than the area threshold, and the perimeter is greater than the perimeter threshold, it is judged as corrosion;

[0030] When the aspect ratio of the outline is greater than 1, it is determined to be a crack.

[0031] Further, when analyzing and processing each sub-defect image data chain to obtain the defect evaluation value of each sub-defect image data chain, the defect evaluation value is calculated by the following formula:

[0032] ;

[0033] Among them, E represents the defect evaluation value, Ai represents the area of ​​the i-th contour in the sub-defect image data chain, Pi represents the circumference of the i-th contour in the sub-defect image data chain, Wi represents the length of the i-th contour in the sub-defect image data chain, Hi represents the width of the i-th contour in the sub-defect image data chain, ARi represents the aspect ratio of the i-th contour in the sub-defect image data chain, and t represents the number of real-time image data in the sub-defect image data chain.

[0034] Furthermore, when determining the allocation weight according to the sub-defect image data chain, it includes:

[0035] The defect assessment mean is obtained according to the defect assessment values ​​of all sub-defect image data chains, and the difference between the defect assessment value of each sub-defect image data chain and the defect assessment mean is calculated by the following formula;

[0036]

[0037] Wherein, Qa represents the difference value, Ea represents the defect evaluation value of the a-th sub-defect image data chain, and E0 represents the defect evaluation mean value;

[0038] The weight of each sub-defect image data chain is determined according to the difference value:

[0039]

[0040] Wherein, Za represents the weight of the a-th sub-defect image data chain, and w represents the standard deviation.

[0041] Further, when the overall defect assessment value of the area to be monitored is obtained according to the allocated weights and all defect assessment values, the overall defect assessment value is calculated by the following formula:

[0042] ;

[0043] Wherein, Ej represents the defect evaluation value of the j-th sub-defect image data chain, Zj represents the weight of the j-th sub-defect image data chain, and X represents the overall defect evaluation value.

[0044] Further, when determining the risk estimation of the area to be monitored according to the dynamic parameter and the overall defect assessment value, it includes:

[0045] The dynamic parameters include the change in population density and the change in building density;

[0046] The risk valuation is obtained by calculating the following formula:

[0047]

[0048] Among them, F represents the risk valuation, X represents the overall defect assessment value, represents the change in population density, It represents the change of building density, w1 and w2 represent weight coefficients, and w1+w2=1; Dp represents the population density mitigation factor, and Db represents the building density mitigation factor.

[0049] Furthermore, when issuing an early warning to the area to be monitored according to the risk valuation, it includes:

[0050] Obtain a risk excess value according to the risk estimation and the risk threshold, wherein the risk excess value is the difference between the risk estimation and the risk threshold, and compare the risk excess value with a first preset risk excess value and a second preset risk excess value respectively, and issue an early warning to the area to be monitored according to the comparison result;

[0051] When the risk excess value is less than or equal to the first preset risk excess value, a third-level warning is issued for the area to be monitored; when the risk excess value is greater than the first preset risk excess value and less than or equal to the second preset risk excess value, a second-level warning is issued for the area to be monitored; when the risk excess value is greater than the second preset risk excess value, a first-level warning is issued for the area to be monitored; the third-level warning is less than the second-level warning, and the second-level warning is less than the first-level warning.

[0052] Compared with the prior art, the beneficial effects of the present invention are as follows: by deploying multiple information collection points in the area to be monitored, and using a CCD camera to collect real-time image data, an image data set is constructed, thereby realizing real-time monitoring of the pipeline and its operating status. By comparing the collected real-time image with the standard image, the defect image data of the pipeline can be accurately identified, and the defects can be further classified and analyzed based on edge recognition technology, thereby improving the accuracy and efficiency of defect identification. The defect image data is subdivided and weighted according to the defect type to obtain a more accurate defect assessment value. In addition, by combining the dynamic parameters of the pipeline, the overall risk level of the area is comprehensively assessed, and ultimately an intelligent early warning of potential risks is achieved. The intelligence level of pipeline monitoring and the reliability of early warning are improved, and the problems of lag, limitations and manual inspection errors existing in traditional monitoring methods are solved, ensuring the safe operation of the pipeline.

[0053] On the other hand, the present application also provides a high-consequence area assessment system for oil and gas pipelines based on machine vision, which is used to apply the above-mentioned high-consequence area assessment method for oil and gas pipelines based on machine vision, including:

[0054] A plurality of CCD cameras are provided, wherein the CCD cameras are deployed at information collection points, and the CCD cameras are used to collect real-time image data of each collection point, and to construct an image data set based on all the real-time image data;

[0055] A judgment unit is configured to compare each real-time image data in the image data set with a standard image, and split the image data set according to the comparison result to obtain a defective image data chain and a normal image data chain;

[0056] a classification unit, configured to analyze each real-time image data in the defect image data chain based on edge recognition, and split the defect image data chain into a plurality of sub-defect image data chains according to defect types;

[0057] a processing unit configured to analyze and process each sub-defect image data chain to obtain a defect assessment value of each sub-defect image data chain, determine an allocation weight according to the sub-defect image data chain, and obtain an overall defect assessment value of the to-be-monitored area according to the allocation weight and all defect assessment values;

[0058] The early warning unit is configured to collect dynamic parameters of the area to be monitored, determine the risk valuation of the area to be monitored according to the dynamic parameters and the overall defect assessment value, and issue an early warning to the area to be monitored according to the risk valuation.

[0059] It is understandable that the above-mentioned machine vision-based high-consequence area assessment method and system for oil and gas pipelines have the same beneficial effects and will not be described in detail here. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0061] Figure 1 A flow chart of a method for evaluating high-consequence areas of oil and gas pipelines based on machine vision provided by an embodiment of the present invention;

[0062] Figure 2 A structural block diagram of a machine vision-based oil and gas pipeline high consequence area assessment system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0063] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art. It should be noted that, in the absence of conflict, the embodiments of the present invention and the features described in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0064] In some embodiments of the present application, see Figure 1 As shown, a method for assessing high-consequence areas of oil and gas pipelines based on machine vision includes:

[0065] S100: several information collection points are set in the area to be monitored, a CCD camera is deployed at each collection point, real-time image data of each collection point is collected, and an image data set is constructed based on all the real-time image data.

[0066] S200: Compare each real-time image data in the image data set with the standard image respectively, split the image data set according to the comparison result, and obtain a defect image data chain and a normal image data chain.

[0067] S300: Analyze each real-time image data in the defect image data chain based on edge recognition, and split the defect image data chain into a plurality of sub-defect image data chains according to defect types.

[0068] S400: Analyze and process each sub-defect image data chain to obtain the defect assessment value of each sub-defect image data chain, determine the allocation weight according to the sub-defect image data chain, and obtain the overall defect assessment value of the area to be monitored according to the allocation weight and all defect assessment values.

[0069] S500: Collect dynamic parameters of the area to be monitored, determine the risk valuation of the area to be monitored according to the dynamic parameters and the overall defect assessment value, and issue an early warning to the area to be monitored according to the risk valuation.

[0070] Specifically, in S10, multiple information collection points are set up in the area to be monitored, and CCD cameras are deployed at each collection point. The CCD camera collects image data of the area in real time to ensure all-round real-time monitoring in the entire area. By collecting image data from all collection points, a complete image data set is constructed to provide a basis for subsequent analysis. In S200, each frame of real-time image is compared with a pre-set standard image, and abnormal points in the image are identified through difference detection. According to the comparison results, the image data set is divided into two categories: one is the "defect image data chain" containing potential defects, and the other is the "normal image data chain" without abnormal detection. The preliminary classification can help the system to centrally process defect image data and effectively reduce the complexity of data processing. In S300, each frame of the defect image data chain is analyzed more deeply, and edge recognition technology is used to accurately detect the defect contour of the image. Edge recognition can accurately locate the shape, size and position of the defect, and then the defect image is further classified according to these features to generate several "sub-defect image data chains", each of which represents a different type of defect. The accuracy of defect identification is ensured and the foundation for subsequent defect evaluation is laid. In S400, the defects in each sub-defect image data chain are analyzed, and the defect assessment value of each defect type is calculated. The assessment value is calculated based on the parameters of the defect. Subsequently, a corresponding weight is assigned to each defect according to the attributes of each sub-defect image data chain. Finally, all defect assessment values ​​and corresponding weights are combined to calculate the overall defect assessment value of the entire monitored area. The impact of different types of defects is combined to provide a more comprehensive defect assessment result. In addition to static defects, S500 also collects dynamic parameters of the monitored area (such as changes in personnel and building density, etc.), combines these dynamic factors with the overall defect assessment value, and forms a more accurate regional risk assessment. The risk valuation of the area is obtained. If the risk reaches a certain threshold, an early warning will be issued to prompt preventive measures to ensure the safety of pipeline operation.

[0071] It is understandable that by combining real-time image acquisition, image comparison, edge recognition and multi-dimensional data analysis, efficient and accurate defect identification and risk assessment are achieved. Compared with traditional methods, it can fully cover the monitoring area, detect and classify defects in real time, and improve the timeliness and accuracy of monitoring; by introducing edge recognition technology, the accuracy of defect analysis is ensured; at the same time, by comprehensively considering dynamic parameters and defect data, an intelligent regional risk assessment and early warning mechanism is realized. It effectively solves the problems of manual inspection lag and sensor monitoring limitations in traditional monitoring methods, and improves the level of pipeline safety management in high-risk areas such as oil and natural gas.

[0072] In some embodiments of the present application, when splitting the image data set according to the comparison results to obtain a defective image data chain and a normal image data chain, it includes: obtaining a first initial blank data chain and a second initial blank data chain, wherein each initial blank data chain is provided with a number of data chain nodes and connection nodes.

[0073] Specifically, the data volume of the real-time image data in the image data set is compared with the data volume of the standard image. When the difference between the data volume of the real-time image data and the data volume of the standard image is greater than a times the data volume of the standard image, a chain change mark is generated. All real-time image data carrying the chain change mark are transferred to the first initial blank data chain to obtain a defective image data chain. The unmarked real-time image data in the image data set is transferred to the second initial blank data chain to obtain a normal image data chain.

[0074] It can be understood that before splitting the image data set, two initial blank data chains are first generated, namely the first initial blank data chain and the second initial blank data chain. These blank data chains are containers for storing different types of image data, and a number of data chain nodes and connection nodes are set on each data chain. The data chain nodes are used to store image data, while the connection nodes ensure the continuity and orderliness of the data chain. This structure provides a basis for the classification and transfer of subsequent image data. By comparing the data volume of real-time image data with that of standard images, it is determined whether the image data is abnormal (i.e., defective). The data volume difference between the real-time image data and the standard image is calculated, and a threshold value (i.e., "a times the data volume of the standard image") is set. When the difference exceeds this threshold, it means that the real-time image data is greatly different from the standard image and there is a defect, and a "chain change mark" will be generated for the image data. According to the chain change mark, the image data in the image data set is divided into two categories. The data volume of the above-mentioned a times the standard image is a reference value, and the specific size of the a value can be determined according to actual monitoring needs.

[0075] It can be understood that by comparing the image data volume difference, the image data can be quickly divided into two categories: defective and normal, achieving efficient data classification and management. Compared with traditional image processing methods, using data volume difference as the standard for defect identification can still work effectively when the subtle differences in the image are difficult to detect through vision or traditional algorithms. In addition, by constructing and classifying and storing the initial blank data chain, different types of data chains can be effectively managed, making subsequent defect analysis and processing more efficient and orderly.

[0076] In some embodiments of the present application, when the defect image data chain is split into a plurality of sub-defect image data chains according to the defect type, it includes:

[0077] Convert each real-time image data in the defect image data chain into a grayscale image.

[0078]

[0079] Use Gaussian filtering to smooth the grayscale image.

[0080]

[0081] The Sobel operator is used to calculate the horizontal and vertical gradients of the smoothed image, and the Canny algorithm is applied to extract the image edge.

[0082]

[0083]

[0084] The contour of the edge image is obtained based on contour detection, the contour area, contour perimeter and contour aspect ratio are extracted from each contour, and the defect type is determined according to the contour area, contour perimeter and contour aspect ratio.

[0085] in, represents the gray value at position (x, y), , , Represent the pixel values ​​of the red, green, and blue channels at position (x, y), respectively. represents the pixel value of the smoothed image at position (x, y), k represents the radius of the Gaussian kernel, represents the value of the Gaussian kernel, represents the processed grayscale image, represents the horizontal gradient of the image at position (x, y), Represents the vertical gradient of the image at position (x, y).

[0086] In some embodiments of the present application, when the defect type is determined based on the contour area, contour perimeter, and contour aspect ratio, it includes: when the contour area is less than or equal to the area threshold and the contour perimeter is greater than the perimeter threshold, it is determined to be a scratch. When the contour is curved and the contour area is greater than the area threshold and the perimeter is greater than the perimeter threshold, it is determined to be corrosion. When the contour aspect ratio is greater than 1, it is determined to be a crack.

[0087] Specifically, each real-time image in the defect image data chain is converted into a grayscale image. Each pixel value of the grayscale image is calculated by weighting the pixel values ​​of the red, green, and blue channels in a certain proportion, which can simplify image processing, reduce the amount of calculation, and retain the structural information of the image. After conversion to a grayscale image, the image is smoothed using a Gaussian filter. The Gaussian filter performs a convolution operation on the image through a Gaussian kernel, which can effectively remove noise from the image, reduce subtle textures or random pixel fluctuations, and thus present the overall structure of the image more clearly. It helps to reduce the false alarm rate in subsequent edge detection. The smoothed image uses the Sobel operator to calculate the horizontal and vertical gradients of the image. The Sobel operator can detect the area with the largest grayscale change in the image, which usually corresponds to the edge in the image. Next, the Canny algorithm is applied to further extract the edges of the image. The Canny algorithm can detect the main edges in the image and ignore noise or tiny pseudo edges through two thresholding and edge connection operations. Based on the extracted image edges, contour detection is used to obtain closed areas in the image, which are where the defects appear. For each contour area, the geometric features such as the area, perimeter and aspect ratio of the contour are extracted. These geometric features reflect the key information of the defect such as the shape and size. When the contour area is less than or equal to the set area threshold, but the contour perimeter is greater than the set perimeter threshold, the defect is judged to be a scratch. Because although the area is small, it presents a long and thin shape. When the contour area is greater than the area threshold and the perimeter is also greater than the perimeter threshold, and the contour is curved, it is judged to be corrosion. This means that the defect range in this area is large and the shape is irregular, which meets the characteristics of corrosion. When the aspect ratio of the contour is greater than 1, it is judged to be a crack. It shows that the defect is long and narrow, which meets the characteristics of a crack.

[0088] It is understandable that by combining image processing techniques, including grayscale conversion, Gaussian filtering, Sobel operator, Canny algorithm and contour detection, key defect information can be extracted from complex image data, and accurate classification can be achieved based on the geometric features of the defects. Compared with traditional manual detection or simple image processing methods, this improves the accuracy and automation of defect identification, especially for the identification and classification of subtle defects. In addition, by extracting multiple features such as area, perimeter and aspect ratio, it can adapt to different types of defects and ensure effective operation under various complex working conditions.

[0089] In some embodiments of the present application, when analyzing and processing each sub-defect image data chain to obtain a defect evaluation value of each sub-defect image data chain, the defect evaluation value is calculated by the following formula:

[0090] ;

[0091] Among them, E represents the defect evaluation value, Ai represents the area of ​​the i-th contour in the sub-defect image data chain, Pi represents the circumference of the i-th contour in the sub-defect image data chain, Wi represents the length of the i-th contour in the sub-defect image data chain, Hi represents the width of the i-th contour in the sub-defect image data chain, ARi represents the aspect ratio of the i-th contour in the sub-defect image data chain, and t represents the number of real-time image data in the sub-defect image data chain.

[0092] In some embodiments of the present application, when determining the allocation weight according to the sub-defect image data chain, it includes:

[0093] The defect assessment mean is obtained according to the defect assessment values ​​of all sub-defect image data chains, and the difference between the defect assessment value of each sub-defect image data chain and the defect assessment mean is calculated by the following formula.

[0094]

[0095] Among them, Qa represents the difference value, Ea represents the defect assessment value of the a-th sub-defect image data chain, and E0 represents the defect assessment mean value.

[0096] Determine the weight of each sub-defect image data chain according to the difference value:

[0097]

[0098] Wherein, Za represents the weight of the a-th sub-defect image data chain, and w represents the standard deviation.

[0099] In some embodiments of the present application, when the overall defect assessment value of the area to be monitored is obtained according to the assigned weights and all defect assessment values, the overall defect assessment value is calculated by the following formula:

[0100] ;

[0101] Wherein, Ej represents the defect evaluation value of the j-th sub-defect image data chain, Zj represents the weight of the j-th sub-defect image data chain, and X represents the overall defect evaluation value.

[0102] It can be understood that by quantitatively analyzing the geometric features of each sub-defect image data chain, the severity of each defect is accurately assessed. By calculating the defect assessment value, mean difference and weight, and combining these factors to evaluate the entire area, a comprehensive and accurate quantitative assessment of the defect situation in the monitored area is achieved. Compared with traditional manual detection or simple threshold determination methods, it can process complex and multi-dimensional defect data and dynamically assign weights according to their relative severity, effectively improving the accuracy and reliability of the assessment.

[0103] In some embodiments of the present application, when determining the risk assessment value of the area to be monitored according to the dynamic parameters and the overall defect assessment value, it includes:

[0104] Dynamic parameters include changes in population density and building density.

[0105] The risk estimate is calculated using the following formula.

[0106]

[0107] Among them, F represents the risk valuation, X represents the overall defect assessment value, represents the change in population density, It represents the change of building density, w1 and w2 represent weight coefficients, and w1+w2=1. Dp represents the population density mitigation factor, and Db represents the building density mitigation factor.

[0108] In some embodiments of the present application, when an early warning is issued for a monitored area according to risk assessment, it includes:

[0109] A risk excess value is obtained based on the risk valuation and the risk threshold. The risk excess value is the difference between the risk valuation and the risk threshold. The risk excess value is compared with a first preset risk excess value and a second preset risk excess value, respectively. An early warning is issued for the monitored area based on the comparison result.

[0110] When the risk excess value is less than or equal to the first preset risk excess value, a third-level warning is issued to the monitored area. When the risk excess value is greater than the first preset risk excess value and less than or equal to the second preset risk excess value, a second-level warning is issued to the monitored area. When the risk excess value is greater than the second preset risk excess value, a first-level warning is issued to the monitored area. The third-level warning is less than the second-level warning, and the second-level warning is less than the first-level warning.

[0111] It is understandable that by combining the dynamic parameters of the area to be monitored and the overall defect assessment value, a more accurate risk valuation is calculated, reflecting the fluctuation of risk under the change of external environment. Especially in the scenario of dynamic changes in population density and building density, risks can be predicted more accurately. The hierarchical design of the early warning mechanism sends out corresponding early warning signals according to different risk levels to ensure timely response when potential dangers occur. It improves the predictability and response efficiency of pipeline monitoring and reduces the possibility of accidents caused by defects.

[0112] In the above embodiment, multiple information collection points are deployed in the area to be monitored, and real-time image data is collected using a CCD camera to construct an image data set, thereby achieving real-time monitoring of the pipeline and its operating status. By comparing the collected real-time image with the standard image, the defect image data of the pipeline can be accurately identified, and the defects can be further classified and analyzed based on edge recognition technology, thereby improving the accuracy and efficiency of defect identification. The defect image data is subdivided and weighted according to the defect type to obtain a more accurate defect assessment value. In addition, by combining the dynamic parameters of the pipeline, the overall risk level of the area is comprehensively assessed, and ultimately an intelligent early warning of potential risks is achieved. The intelligence level of pipeline monitoring and the reliability of early warning are improved, and the problems of lag, limitations and manual inspection errors existing in traditional monitoring methods are solved, ensuring the safe operation of the pipeline.

[0113] In another preferred embodiment based on the above embodiment, refer to Figure 2 As shown, this embodiment provides a high-consequence area assessment system for oil and gas pipelines based on machine vision, which is used to apply the above-mentioned high-consequence area assessment method for oil and gas pipelines based on machine vision, including:

[0114] There are several CCD cameras, which are deployed at information collection points. The CCD cameras are used to collect real-time image data at each collection point and construct an image data set based on all the real-time image data.

[0115] A judgment unit is configured to compare each real-time image data in the image data set with the standard image, split the image data set according to the comparison result, and obtain a defective image data chain and a normal image data chain;

[0116] A classification unit is configured to analyze each real-time image data in the defect image data chain based on edge recognition, and split the defect image data chain into a plurality of sub-defect image data chains according to defect types;

[0117] A processing unit is configured to analyze and process each sub-defect image data chain, obtain a defect assessment value of each sub-defect image data chain, determine an allocation weight according to the sub-defect image data chain, and obtain an overall defect assessment value of the area to be monitored according to the allocation weight and all defect assessment values;

[0118] The early warning unit is configured to collect dynamic parameters of the area to be monitored, determine the risk valuation of the area to be monitored based on the dynamic parameters and the overall defect assessment value, and issue an early warning to the area to be monitored based on the risk valuation.

[0119] It is understandable that by deploying multiple information collection points in the area to be monitored, using CCD cameras to collect real-time image data, and constructing an image data set, real-time monitoring of the pipeline and its operating status is achieved. By comparing the collected real-time images with standard images, the defect image data of the pipeline can be accurately identified, and the defects can be further classified and analyzed based on edge recognition technology, thereby improving the accuracy and efficiency of defect identification. The defect image data is subdivided and weighted according to the defect type to obtain a more accurate defect assessment value. By combining the dynamic parameters of the pipeline, the overall risk level of the area is comprehensively assessed, and ultimately an intelligent early warning of potential risks is achieved. The intelligence level of pipeline monitoring and the reliability of early warning are improved, and the problems of lag, limitations, and manual inspection errors in traditional monitoring methods are solved, ensuring the safe operation of the pipeline.

[0120] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application 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 codes.

[0121] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes 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.

[0122] 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 including 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.

[0123] 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.

[0124] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for assessing high-consequence areas of oil and gas pipelines based on machine vision, characterized in that: include: Set up several information collection points in the area to be monitored, deploy a CCD camera at each collection point, collect real-time image data at each collection point, and build an image data set based on all real-time image data; Compare each real-time image data in the image data set with the standard image respectively, split the image data set according to the comparison result, and obtain a defect image data chain and a normal image data chain; Analyze each real-time image data in the defect image data chain based on edge recognition, and split the defect image data chain into a plurality of sub-defect image data chains according to defect types; Analyze and process each sub-defect image data chain to obtain a defect assessment value of each sub-defect image data chain, determine an allocation weight according to the sub-defect image data chain, and obtain an overall defect assessment value of the area to be monitored according to the allocation weight and all defect assessment values; Collecting dynamic parameters of the area to be monitored, determining a risk estimate of the area to be monitored based on the dynamic parameters and an overall defect assessment value, and issuing an early warning to the area to be monitored based on the risk estimate; When analyzing and processing each sub-defect image data chain to obtain the defect evaluation value of each sub-defect image data chain, the defect evaluation value is calculated by the following formula: ； Wherein, E represents the defect evaluation value, Ai represents the area of ​​the i-th contour in the sub-defect image data chain, Pi represents the perimeter of the i-th contour in the sub-defect image data chain, Wi represents the length of the i-th contour in the sub-defect image data chain, Hi represents the width of the i-th contour in the sub-defect image data chain, ARi represents the aspect ratio of the i-th contour in the sub-defect image data chain, and t represents the number of real-time image data in the sub-defect image data chain; When determining the allocation weight according to the sub-defect image data chain, it includes: The defect assessment mean is obtained according to the defect assessment values ​​of all sub-defect image data chains, and the difference between the defect assessment value of each sub-defect image data chain and the defect assessment mean is calculated by the following formula; ； Wherein, Qa represents the difference value, Ea represents the defect evaluation value of the a-th sub-defect image data chain, and E0 represents the defect evaluation mean value; The weight of each sub-defect image data chain is determined according to the difference value: ； Wherein, Za represents the weight of the a-th sub-defect image data chain, and w represents the standard deviation; When the overall defect assessment value of the area to be monitored is obtained according to the allocated weights and all defect assessment values, the overall defect assessment value is calculated by the following formula: ； Wherein, Ej represents the defect evaluation value of the j-th sub-defect image data chain, Zj represents the weight of the j-th sub-defect image data chain, and X represents the overall defect evaluation value; When determining the risk assessment value of the area to be monitored according to the dynamic parameters and the overall defect assessment value, it includes: The dynamic parameters include the change in population density and the change in building density; The risk valuation is obtained by calculating the following formula: ； Among them, F represents the risk valuation, X represents the overall defect assessment value, represents the change in population density, It represents the change of building density, w1 and w2 represent weight coefficients, and w1+w2=1; Dp represents the population density mitigation factor, and Db represents the building density mitigation factor.

2. The method for evaluating high-consequence areas of oil and gas pipelines based on machine vision according to claim 1 is characterized in that: The image data set is split according to the comparison result to obtain a defect image data chain and a normal image data chain, including: Acquire a first initial blank data chain and a second initial blank data chain, wherein each initial blank data chain is provided with a plurality of data chain nodes and connection nodes; Comparing the data volume of the real-time image data in the image data set with the data volume of the standard image, and generating a link switching mark when the difference between the data volume of the real-time image data and the data volume of the standard image is greater than a times the data volume of the standard image; Transferring all real-time image data carrying the chain change mark to the first initial blank data chain to obtain the defective image data chain; The unlabeled real-time image data in the image data set is transferred to the second initial blank data chain to obtain the normal image data chain.

3. The method for evaluating high-consequence areas of oil and gas pipelines based on machine vision according to claim 1 is characterized in that: When the defect image data chain is split into a plurality of sub-defect image data chains according to defect types, it includes: Convert each real-time image data in the defect image data chain into a grayscale image; ； Using Gaussian filtering to smooth the grayscale image; ； The Sobel operator is used to calculate the horizontal and vertical gradients of the smoothed image, and the Canny algorithm is used to extract the image edge; ； ； Acquire the contour of the edge image based on contour detection, extract the contour area, contour perimeter and contour aspect ratio from each contour, and determine the defect type according to the contour area, contour perimeter and contour aspect ratio; in, represents the gray value at position (x, y), , , Represent the pixel values ​​of the red, green, and blue channels at position (x, y), respectively. represents the pixel value of the smoothed image at position (x, y), k represents the radius of the Gaussian kernel, represents the value of the Gaussian kernel, represents the processed grayscale image, represents the horizontal gradient of the image at position (x, y), Represents the vertical gradient of the image at position (x, y).

4. The method for evaluating high-consequence areas of oil and gas pipelines based on machine vision according to claim 3 is characterized in that: When determining the defect type according to the contour area, contour perimeter and contour aspect ratio, it includes: When the contour area is less than or equal to the area threshold and the contour perimeter is greater than the perimeter threshold, it is determined to be a scratch; When the contour is curved and the contour area is greater than the area threshold, and the perimeter is greater than the perimeter threshold, it is judged as corrosion; When the aspect ratio of the outline is greater than 1, it is determined to be a crack.

5. The method for evaluating high-consequence areas of oil and gas pipelines based on machine vision according to claim 1, characterized in that: When issuing an early warning for the area to be monitored according to the risk assessment, it includes: Obtain a risk excess value according to the risk estimation and the risk threshold, wherein the risk excess value is the difference between the risk estimation and the risk threshold, and compare the risk excess value with a first preset risk excess value and a second preset risk excess value respectively, and issue an early warning to the area to be monitored according to the comparison result; When the risk excess value is less than or equal to the first preset risk excess value, a third-level warning is issued for the area to be monitored; when the risk excess value is greater than the first preset risk excess value and less than or equal to the second preset risk excess value, a second-level warning is issued for the area to be monitored; when the risk excess value is greater than the second preset risk excess value, a first-level warning is issued for the area to be monitored; the third-level warning is less than the second-level warning, and the second-level warning is less than the first-level warning.

6. A machine vision-based oil and gas pipeline high consequence area assessment system, used for applying the machine vision-based oil and gas pipeline high consequence area assessment method according to any one of claims 1 to 5, characterized in that: include: A plurality of CCD cameras are provided, wherein the CCD cameras are deployed at information collection points, and the CCD cameras are used to collect real-time image data of each collection point, and to construct an image data set based on all the real-time image data; A judgment unit is configured to compare each real-time image data in the image data set with a standard image, and split the image data set according to the comparison result to obtain a defective image data chain and a normal image data chain; a classification unit, configured to analyze each real-time image data in the defect image data chain based on edge recognition, and split the defect image data chain into a plurality of sub-defect image data chains according to defect types; a processing unit configured to analyze and process each sub-defect image data chain to obtain a defect assessment value of each sub-defect image data chain, determine an allocation weight according to the sub-defect image data chain, and obtain an overall defect assessment value of the to-be-monitored area according to the allocation weight and all defect assessment values; The early warning unit is configured to collect dynamic parameters of the area to be monitored, determine the risk valuation of the area to be monitored according to the dynamic parameters and the overall defect assessment value, and issue an early warning to the area to be monitored according to the risk valuation.

Images