A method and system for detecting bulges on the pipeline surface of a rail train

Abstract

The present invention provides a pipeline surface bulge detection method and system for rail trains, comprising a pipeline image acquisition step, acquiring a pipeline image of a pipeline system at the bottom of a train by a visual camera; a pipeline abnormality analysis step, analyzing a bulge area on the pipeline surface according to the pipeline image after fluid is introduced, and synchronously mapping the bulge area in the pipeline image after the fluid is extracted; a bulge tightening calculation step, analyzing the difference in pipeline size change after fluid is introduced into the pipeline and after fluid is extracted, and obtaining a pipeline surface tightening value by calculation with a fluid data set; a pipeline crack matching step, obtaining a matching value by calculation of the tightening value of the bulge area on the pipeline surface, the tightening value of the normal area on the pipeline surface, and the fluid data set, comparing the matching value with a preset matching threshold N1, and outputting the crack type according to the comparison result; the present invention has the advantage of avoiding recognition difficulties caused by visual errors or environmental errors, quickly providing accurate and reliable crack conditions, and facilitating subsequent processing solutions.

Description

Technical Field

[0001] The present invention relates to the technical field of rail vehicle wheel pair detection, and more particularly to a method and system for detecting bulges on the surface of a pipeline of a rail train. Background Art

[0002] With the rapid development of rail transit, the safety and reliability of rail trains have become increasingly important issues. As an important part of rail trains, the status of pipelines directly affects the normal operation and safety of trains. During the operation of rail trains, pipelines may bulge due to factors such as vibration, pressure changes, and material aging. These bulges not only affect the appearance of the pipelines, but more importantly, they may cause safety hazards such as leakage and bursting. Therefore, it is necessary to detect abnormal conditions such as pipeline bulges in a timely and accurate manner to ensure the safe operation of trains.

[0003] At present, the detection of pipeline bulges includes traditional manual inspection and computer vision inspection. Among them, visual inspection is to extract bulge features by acquiring visual images. However, due to the different pressures and flow rates of the fluid in the pipeline, the filling state of the pipeline itself is different, and the expansion of the pipeline is different. Therefore, there is still uncertainty in the single visual judgment of bulges, and the existing detection method cannot further judge the size of the bulge. Therefore, there is no good judgment on the criticality of the pipeline, which is difficult to meet the current intelligent safety detection needs of rail trains. Summary of the invention

[0004] In view of the deficiencies in the prior art, an object of the present invention is to provide a method and system for detecting bulges on the pipeline surface of a railway train.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A method for detecting bulges on the pipeline surface of a rail train comprises the following steps:

[0007] a pipeline image acquisition step, acquiring an image of the pipeline system at the bottom of the train captured by a visual camera as a pipeline image, wherein the pipeline image includes an image after fluid is introduced into the pipeline and an image after fluid is extracted from the pipeline;

[0008] a pipeline abnormality analysis step, analyzing the pipeline image after the fluid is passed into the pipeline through a preset target detection model to obtain a bulge area on the pipeline surface, and synchronously mapping the bulge area on the pipeline surface in the pipeline image after the fluid is extracted;

[0009] a bulge tightening calculation step, obtaining a difference in pipeline size change after fluid is introduced into the pipeline and after fluid is extracted from the pipeline according to the pipeline image analysis, and calculating the pipeline size change difference and the fluid data set through a tightening value formula to obtain a pipeline surface tightening value, wherein the pipeline surface tightening value includes a tightening value of a bulge area on the pipeline surface and a tightening value of a normal area on the pipeline surface;

[0010] The pipeline crack matching step calculates the matching value of the pipeline surface bulge area tightening value, the pipeline surface normal area tightening value and the fluid data set through the bulge and crack relationship formula, compares the matching value with several preset matching threshold ranges, and outputs the crack type according to the comparison result.

[0011] Furthermore, the pipeline abnormality analysis step analyzes the position of the pipeline surface bulge area in the pipeline image, and outputs the pipeline front bulge type or pipeline side bulge type corresponding to the pipeline image under the current visual camera;

[0012] The bulge tightening calculation step is as follows: when the bulge area on the pipeline surface is a bulge type on the front of the pipeline, the bulge height is calculated by the item point pixels of the bulge area on the pipeline surface, and the pipeline size change difference is calculated based on the bulge height; when the bulge area on the pipeline surface is a bulge type on the side of the pipeline, the pipeline image is used to plan the pipeline edge trajectory through an edge detection algorithm, the center point of the bulge area on the pipeline surface is found, and the pipe diameter normal connecting the center point and the pipe perpendicular to the pipe is drawn, and the pipe size change difference is calculated through the change of the pipe diameter normal.

[0013] Furthermore, the tightening value formula is configured as:

[0014] ,

[0015] in is the tightening value of the pipeline surface at time t; , , , They are respectively the pressure data, flow data, temperature data and pipeline size change in the pipeline at time t in the fluid data set; is the i-th data in the history; is the preset first adjustment parameter; It is the preset second adjustment parameter used to balance the impact of historical data; The number of historical data.

[0016] Furthermore, the relationship between the bulge and the crack is configured as follows:

[0017] ,

[0018] ,

[0019] ,

[0020] in, is a multivariable function used to describe the effects of temperature, pressure and flow rate on pipeline status in the fluid data set; is a two-variable function used to compare the difference in tightening values ​​between the normal area on the pipe surface and the bulging area on the pipe surface; is the matching value; It is the tightening value of the normal area on the pipe surface; It is the tightening value of the bulging area on the pipeline surface; The temperature value of the fluid entering the pipeline in the fluid data set; is the pressure value of the fluid entering the pipeline in the fluid data set; is the flow rate value of the fluid entering the pipeline in the fluid data set; They are the first constant, the second constant, the third constant, the fourth constant and the fifth constant, which are used to adjust the influence of each factor respectively.

[0021] Furthermore, the crack types include longitudinal cracks, transverse cracks, short shallow cracks and long deep cracks.

[0022] Furthermore, it also includes a crack type verification step, in which the edge contour of the bulge area is obtained by edge detection based on the bulge area on the pipeline surface, and the bulge area in the two-dimensional view is obtained by area calculation of the edge contour of the bulge area, the crack type is matched according to the bulge area, and compared with the short shallow cracks or long deep cracks matched in the pipeline crack matching step, and the matching normal instructions and matching abnormal instructions are output according to the comparison results.

[0023] Furthermore, in the crack type verification step, the center point of the edge contour of the bulge area is found, and a line is drawn through the center point between any two points of the edge contour of the bulge area, and the bulge type is determined according to the length of the line. The bulge type includes circular or elliptical types. The bulge type is compared with the longitudinal crack or transverse crack matched in the pipeline crack matching step, and a normal matching instruction and an abnormal matching instruction are output according to the comparison result.

[0024] Furthermore, a pipeline system library is preset, and a fluid upper limit data set is stored in the pipeline system library. The fluid upper limit data set corresponds to the crack type one by one, and the fluid upper limit data set includes a pressure threshold N2, a flow rate threshold N3, and a temperature threshold N4;

[0025] The method also includes a pipeline fluid feedback step, in which a corresponding fluid upper limit data set is indexed in a pipeline system library according to the crack type as a safety data set of the pipeline fluid.

[0026] Furthermore, the pipeline system library stores several pipeline models;

[0027] The pipeline abnormality analysis step selects any pipeline image, indexes the pipeline model corresponding to the pipeline image in the pipeline system library, and obtains the corresponding fluid data set according to the pipeline model.

[0028] A pipeline surface bulge detection system for a rail train includes a pipeline image acquisition module, which acquires an image of a pipeline system at the bottom of the train captured by a visual camera as a pipeline image, wherein the pipeline image includes an image after a fluid is introduced into the pipeline and an image after the fluid is extracted;

[0029] The pipeline abnormality analysis module analyzes the pipeline image after the fluid is passed into the pipeline through a preset target detection model to obtain the bulge area on the pipeline surface, and synchronously maps the bulge area on the pipeline surface into the pipeline image after the fluid is extracted;

[0030] A bulge tightening calculation module, which obtains the difference in pipeline size change after the fluid is introduced into the pipeline and after the fluid is extracted according to the pipeline image analysis, and calculates the pipeline size change difference and the fluid data set through a tightening value formula to obtain a pipeline surface tightening value, wherein the pipeline surface tightening value includes a tightening value of a bulge area on the pipeline surface and a tightening value of a normal area on the pipeline surface;

[0031] The pipeline crack matching module calculates the matching value of the pipeline surface bulge area tightening value, the pipeline surface normal area tightening value and the fluid data set through the bulge and crack relationship formula, compares the matching value with the preset matching threshold N1, and outputs the crack type according to the comparison result.

[0032] The beneficial effects of the present invention include: monitoring in real time through images whether there are bulges on the pipeline surface; if so, screening out the bulged area according to target detection, and then deflate the bulge of the pipeline through a one-through-two-row method; calculating the tightening value of the bulge and the tightening value of the normal part by comparing the difference in the size change of the pipeline and combining various fluid data; finally, calculating the type of crack on the pipeline surface by the difference in the tightening values; compared with traditional machine vision learning and crack analysis, the bulge-crack deduction of the present invention is more accurate and can be applied to various types of pipelines, avoiding recognition difficulties due to visual errors or environmental errors, and quickly providing accurate and reliable crack conditions, which is convenient for subsequent treatment plans. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is the overall flow chart of the present invention;

[0034] Figure 2 It is a flow chart of the bulge analysis in the present invention;

[0035] Figure 3 It is a flow chart of crack type verification in the present invention. DETAILED DESCRIPTION

[0036] The present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. The same parts are represented by the same reference numerals. It should be noted that the words "front", "rear", "left", "right", "upper" and "lower" used in the following description refer to directions in the accompanying drawings, and the words "bottom surface" and "top surface", "inner" and "outer" refer to directions toward or away from the geometric center of a specific component, respectively.

[0037] At present, the detection of pipeline bulges includes traditional manual inspection, computer vision detection and unsupervised learning detection, wherein visual detection is to extract bulge features through acquired visual images. However, due to the influence of different pressures and flow rates of the fluid in the pipeline, the filling state of the pipeline itself is different, and the expansion of the pipeline is different. Therefore, there is still uncertainty in the single visual judgment of bulges, and the existing detection method cannot further judge the size of the bulge. Therefore, there is no good judgment on the criticality of the pipeline, which is difficult to meet the current intelligent safety detection needs of rail trains; therefore, the present invention designs such a pipeline surface bulge detection method for rail trains, such as Figure 1 As shown, the following steps are included:

[0038] In the pipeline image acquisition step, at least one set of high-resolution visual cameras is arranged on the side of each set of pipelines, and the visual cameras collect images of the target pipeline system (hydraulic pipes, which are rubber pipes) at the bottom of the train as pipeline images, wherein the pipeline images include images after fluid is introduced into the pipeline (expansion) and images after fluid is extracted (contraction).

[0039] The pipeline abnormality analysis step defines that when the train is running at high speed, the pipeline is fixed at a certain position. After the pipeline is filled with fluid, the surface of the pipeline swells. At this time, if there is a defect on the pipeline surface, a bulge will appear on the surface of the pipeline. Then, the target detection model that has been fully trained and verified by deep learning is used to identify and extract the pipeline image with bulges, and the bulge features can be accurately extracted, that is, the bulge area on the pipeline surface. Correspondingly, when the fluid is extracted or closed, the bulge will tighten and deflate. Therefore, the bulge area on the pipeline surface in the pipeline image after the fluid is introduced is synchronously mapped in the pipeline image after the fluid is extracted, so as to facilitate further analysis of the changes in the bulge features. A pipeline system library is constructed, which stores several pipeline models and pipeline setting positions. Any pipeline image is selected, and the corresponding pipeline model of the pipeline image is indexed in the pipeline system library, and the data set of the fluid introduced into the pipeline at this time is obtained according to the pipeline model.

[0040] Bulge tightening calculation steps: If a bulge area has been identified during abnormal analysis, the pipeline can be divided into a bulge part and a normal part. According to the pipeline image analysis, the difference in pipeline size change after the fluid is introduced into the pipeline and after the fluid is extracted is obtained. The pipeline size change difference includes the size change difference from the bulge part to the tightening and the size change difference from the normal part to the tightening. Then, the pipeline size change difference and the fluid data set are calculated through the tightening value formula to obtain the pipeline surface tightening value. Correspondingly, the pipeline surface tightening value includes the tightening value of the bulge area on the pipeline surface and the tightening value of the normal area on the pipeline surface. The fluid data set includes the pressure value of the fluid entering the pipeline, the temperature value of the fluid entering the pipeline, and the flow rate value of the fluid entering the pipeline.

[0041] In the pipeline crack matching step, the tightening value of the bulge area on the pipeline surface, the tightening value of the normal area on the pipeline surface and the fluid data set are calculated through the bulge and crack relationship formula to obtain a matching value, and the matching value is compared with several preset matching threshold ranges, where several matching threshold ranges are several stage thresholds, such as the matching threshold of a certain stage is N1. The matching value is compared with the threshold of each stage, and the crack type is output according to the comparison result, where the crack type includes longitudinal cracks, transverse cracks, short shallow cracks and long deep cracks. The longitudinal cracks and transverse cracks are the same group of classifications, and the short shallow cracks and long deep cracks are the same group of classifications.

[0042] The beneficial effects of the present invention are: real-time monitoring of whether there is a bulge on the pipeline surface through images, if so, the bulged area is screened out according to target detection, and then the bulge of the pipeline is deflated through a one-through-two-row method, and the tightening value of the bulge and the tightening value of the normal part are calculated by the difference in the size change of the pipeline and in combination with various fluid data, and finally the type of crack on the pipeline surface is calculated by the difference in the tightening values. Compared with traditional machine vision learning and crack analysis, the bulge-crack derivation of the present invention is more accurate and can be applied to various types of pipelines, avoiding recognition difficulties due to visual errors or environmental errors, and quickly giving accurate and reliable crack conditions, which is convenient for subsequent treatment plans.

[0043] like Figure 2 As shown, the pipeline abnormality analysis steps, assuming that a pipeline is provided with two sets of visual cameras, which are symmetrically arranged on both sides of the pipeline, then the two sets of visual cameras can capture the front and back sides of the pipeline. Taking a set of cameras on one side as an example, the captured pipeline image is A, then the way in which bulges appear in image A includes bulge features that can be directly displayed in the 2D image and bulge features located on both sides of the pipeline. The bulge features on both sides can only be displayed as bulges on the edge of the pipeline. Therefore, when identifying bulge features, there should be two preset target detection models, corresponding to the above two bulge forms respectively. By analyzing the position of the bulge area on the pipeline surface in the pipeline image A, the pipeline front bulge type or pipeline side bulge type corresponding to the pipeline image A under the current visual camera is output;

[0044] Bulge tightening calculation step: when the bulge area on the pipeline surface is a bulge type on the front of the pipeline, the bulge height is calculated by the item point pixels of the bulge area on the pipeline surface, and then the pipeline size change difference is calculated according to the bulge height;

[0045] When the bulge area on the pipeline surface is the type of bulge on the side of the pipeline, the pipeline image is used to plan the pipeline edge trajectory through the edge detection algorithm. The center point of the edge trajectory line of the bulge area is found through geometric analysis of the bulge area or image segmentation technology. The pipe diameter normal connecting the center point and the pipeline is drawn. By comparing the length or angle changes of the normal at different positions (before and after the bulge), the change difference of the pipeline size is calculated.

[0046] The analysis logic of the change difference matching different pipeline sizes for different bulge types can improve the accuracy and efficiency of detection, ensure the monitoring of the size of pipeline bulges, and accurately infer the situation of cracks on the pipeline surface, so as to facilitate subsequent correct and timely processing decisions.

[0047] In the present invention, the tightening value formula is configured as:

[0048] ,

[0049] in is the tightening value of the pipeline surface at time t; , , , They are respectively the pressure data, flow data, temperature data and pipeline size change in the pipeline at time t in the fluid data set; is the i-th data in the history; is the preset first adjustment parameter; It is the preset second adjustment parameter used to balance the impact of historical data; is the number of historical data;

[0050] In the tightened value formula, the range is the set of real numbers, that is, ,when A high value indicates that there is a large tightening of the pipeline surface. The change of can effectively distinguish the normal part and the bulging part of the pipeline. In addition, the tightening value formula can match the corresponding relationship between the tightening value of the pipeline and various parameters. When the fluid data parameters are different, the size of the pipeline changes differently, that is, the tightening value There are changes. In abnormal analysis, the normal part and the bulge part can be distinguished by visual analysis, but the size and shape of the bulge cannot be accurately analyzed by vision. Therefore, the tightening value of the normal part is calculated. Tightening value of the bulge part The comparative relationship between them.

[0051] The crack type is deduced through the bulge. Specifically, the relationship between the bulge and the crack is configured as follows:

[0052] ,

[0053] ,

[0054] ,

[0055] in, is a multivariable function used to describe the effects of temperature, pressure and flow rate on pipeline status in the fluid data set; is a two-variable function used to compare the difference in tightening values ​​between the normal area on the pipe surface and the bulging area on the pipe surface; is the matching value; It is the tightening value of the normal area on the pipe surface; It is the tightening value of the bulging area on the pipeline surface; The temperature value of the fluid entering the pipeline in the fluid data set; is the pressure value of the fluid entering the pipeline in the fluid data set; is the flow rate value of the fluid entering the pipeline in the fluid data set; They are the first constant, the second constant, the third constant, the fourth constant and the fifth constant, which are used to adjust the influence of each factor respectively;

[0056] The tightening value of the bulge is different when the pipeline surface is a longitudinal crack or a transverse crack, and a pipeline system library is preset, and a corresponding table of matching values ​​and cracks is stored in the pipeline system library, that is, each crack type has a stage threshold, longitudinal cracks and transverse cracks are a group of the same type, short shallow cracks and long deep cracks are a group of the same type, for example, in the corresponding table, Value less than the matching threshold When , it indicates that there are short and shallow longitudinal cracks. Values ​​matching threshold - When it is between , it indicates that there are short and shallow horizontal cracks. Values ​​matching threshold - When it is between , it indicates that there is a long and deep longitudinal crack. Values ​​matching threshold - When it is between , it indicates that there are long and deep transverse cracks.

[0057] like Figure 3As shown, it also includes a crack type verification step, preprocessing, denoising, contrast enhancement and other image processing of the pipeline image of the bulge area on the pipeline surface, so as to improve the distinctiveness of the bulge features; edge detection, edge detection of the bulge area using an edge detection algorithm, obtaining the edge contour of the bulge area, and extracting a clear edge contour of the bulge area; contour tracking, using a contour tracking algorithm in image processing technology to track the extracted edge contour of the bulge area; area calculation, according to the tracked contour, calculating the area of ​​the bulge area in a two-dimensional view (plane image), wherein the calculation method can be a geometric calculation or counting the pixels inside the contour and converting it into an actual physical area;

[0058] Establish a database containing different crack types (such as short shallow cracks, long deep cracks, etc.) and their corresponding bulge areas, match the calculated bulge area with the data in the crack type database, find the crack type closest to the bulge area, set a matching threshold, if the difference between the calculated bulge area and the area of ​​a crack type in the database is within the threshold range, then the match is considered successful, compare the crack type matched in the crack type verification step with the short shallow crack or long deep crack matched in the pipeline crack matching step, check whether the crack types matched in the two steps are consistent, if they are consistent, output a normal match instruction; if they are inconsistent, output an abnormal match instruction, then it is necessary to correct the database or configure the bulge and crack relationship formula. Adjust the constants.

[0059] The crack type verification step is to calculate the centroid of the edge contour of the bulge area as the center point through the centroid calculation method in the image processing technology (it also includes center point verification, judging whether the contour points distributed around the center point are evenly distributed according to the distance from each point in the contour to the center point), and draw a line through the center point between any two points of the edge contour of the bulge area according to the center point (in order to more comprehensively analyze the bulge shape, multiple such lines can be drawn), calculate the length of each line, and count the distribution of these lengths, and then judge the bulge type according to the distribution of the line length. The bulge type includes circular or elliptical. If the line lengths are basically equal and evenly distributed, the bulge may be close to a circle. If the line lengths are different and show a certain regularity (such as the major axis and the minor axis), the bulge may be close to an ellipse.

[0060] Under normal circumstances, when the bulge is elliptical, the crack can be judged to be longitudinal or transverse according to the corresponding relationship between the long axis position of the bulge and the length of the pipeline, and then compared with the longitudinal crack or transverse crack matched in the pipeline crack matching step. If they are consistent, a normal matching instruction is output; if they are inconsistent, an abnormal matching instruction is output;

[0061] In addition, according to the bulge type (circular or elliptical), the association rules with the pipeline crack type (such as longitudinal cracks and transverse cracks) are established, and the corresponding matching logic is designed to compare the determined bulge type with the crack type matched in the pipeline crack matching step. For example, a circular bulge may be associated with a transverse crack, while an elliptical bulge may be associated with a longitudinal crack or an oblique crack in a specific direction. A matching threshold is set to judge the accuracy of the matching result. If the degree of association between the bulge type and the crack type is within the threshold range, the match is considered successful; otherwise, the match is considered failed.

[0062] A pipeline system library is preset, and a fluid upper limit data set is stored in the pipeline system library (the fluid upper limit data set is a set of safety thresholds preset for different crack types, which is used to evaluate the safety of pipeline fluids). The fluid upper limit data set corresponds to the crack type one by one, and the fluid upper limit data set includes a pressure threshold N2, a flow rate threshold N3, and a temperature threshold N4;

[0063] It also includes a pipeline fluid feedback step, indexing a corresponding fluid upper limit data set in a pipeline system library according to the crack type as a safety data set of the pipeline fluid;

[0064] In addition, by establishing association rules between crack types and fluid upper limit data sets, after analyzing the crack types, the pressure threshold N2, flow rate threshold N3 and temperature threshold N4 in the fluid data set can be directly obtained according to the association algorithm.

[0065] A pipeline surface bulge detection system is designed corresponding to the bulge detection method. Specifically, it includes a pipeline image acquisition module, which acquires the image of the pipeline system at the bottom of the train captured by a visual camera as a pipeline image. The pipeline image includes an image after the fluid is introduced into the pipeline and an image after the fluid is extracted;

[0066] The pipeline abnormality analysis module uses a preset target detection model to analyze the pipeline image after the fluid is passed into the pipeline to obtain the bulge area on the pipeline surface, and synchronously maps the bulge area on the pipeline surface to the pipeline image after the fluid is extracted;

[0067] The bulge tightening calculation module obtains the difference in pipeline size change after the fluid is introduced into the pipeline and after the fluid is extracted according to the pipeline image analysis, and calculates the pipeline size change difference and the fluid data set through the tightening value formula to obtain the pipeline surface tightening value. The pipeline surface tightening value includes the tightening value of the bulge area on the pipeline surface and the tightening value of the normal area on the pipeline surface;

[0068] The pipeline crack matching module calculates the matching value of the bulge area on the pipeline surface, the tightening value of the normal area on the pipeline surface and the fluid data set through the bulge and crack relationship formula, compares the matching value with the preset matching threshold N1, and outputs the crack type according to the comparison result.

[0069] The above are only preferred embodiments of the present invention. The protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A method for detecting bulges on the pipeline surface of a railway train, characterized in that: The steps include: a pipeline image acquisition step, acquiring an image of the pipeline system at the bottom of the train captured by a visual camera as a pipeline image, wherein the pipeline image includes an image after fluid is introduced into the pipeline and an image after fluid is extracted from the pipeline; a pipeline abnormality analysis step, analyzing the pipeline image after the fluid is passed into the pipeline through a preset target detection model to obtain a bulge area on the pipeline surface, and synchronously mapping the bulge area on the pipeline surface in the pipeline image after the fluid is extracted; The bulge tightening calculation step is to obtain the difference in pipeline size change after the fluid is introduced into the pipeline and after the fluid is extracted according to the pipeline image analysis, and calculate the pipeline size change difference and the fluid data set through a tightening value formula to obtain a pipeline surface tightening value, wherein the pipeline surface tightening value includes a tightening value of a bulge area on the pipeline surface and a tightening value of a normal area on the pipeline surface; the tightening value formula is configured as: ， in is the tightening value of the pipeline surface at time t; , , , They are respectively the pressure data, flow data, temperature data and pipeline size change in the pipeline at time t in the fluid data set; is the i-th data in the history; is the preset first adjustment parameter; It is the preset second adjustment parameter used to balance the impact of historical data; is the number of historical data; The pipeline crack matching step calculates the tightening value of the pipeline surface bulge area, the tightening value of the pipeline surface normal area and the fluid data set through the bulge and crack relationship formula to obtain a matching value, compares the matching value with a plurality of preset matching threshold ranges, and outputs the crack type according to the comparison result; the bulge and crack relationship formula is configured as: ， ， ， in, is a multivariable function; is a two-variable function used to compare the difference in tightening values ​​between the normal area on the pipe surface and the bulging area on the pipe surface; is the matching value; It is the tightening value of the normal area on the pipe surface; It is the tightening value of the bulging area on the pipeline surface; The temperature value of the fluid entering the pipeline in the fluid data set; is the pressure value of the fluid entering the pipeline in the fluid data set; is the flow rate value of the fluid entering the pipeline in the fluid data set; They are the first constant, the second constant, the third constant, the fourth constant and the fifth constant, which are used to adjust the influence of each factor respectively.

2. According to claim 1, a method for detecting bulges on the pipeline surface of a railway train is characterized in that: The pipeline abnormality analysis step analyzes the position of the pipeline surface bulge area in the pipeline image, and outputs the pipeline front bulge type or pipeline side bulge type corresponding to the pipeline image under the current visual camera; The bulge tightening calculation step is as follows: when the bulge area on the pipeline surface is a bulge type on the front of the pipeline, the bulge height is calculated by the item point pixels of the bulge area on the pipeline surface, and the pipeline size change difference is calculated based on the bulge height; when the bulge area on the pipeline surface is a bulge type on the side of the pipeline, the pipeline image is used to plan the pipeline edge trajectory through an edge detection algorithm, the center point of the bulge area on the pipeline surface is found, and the pipe diameter normal connecting the center point and the pipe perpendicular to the pipe is drawn, and the pipe size change difference is calculated through the change of the pipe diameter normal.

3. The method for detecting bulges on the pipeline surface of a railway train according to claim 1, characterized in that: The crack types include longitudinal cracks, transverse cracks, short shallow cracks and long deep cracks.

4. The method for detecting bulges on the pipeline surface of a railway train according to claim 3, characterized in that: The method also includes a crack type verification step, in which the edge contour of the bulge area is obtained by edge detection based on the bulge area on the pipeline surface, and the bulge area in the two-dimensional view is obtained by area calculation of the edge contour of the bulge area. The crack type is matched according to the bulge area and compared with the short shallow crack or long deep crack matched in the pipeline crack matching step, and a normal matching instruction and an abnormal matching instruction are output according to the comparison result.

5. A method for detecting bulges on the pipeline surface of a railway train according to claim 4, characterized in that: The crack type verification step finds the center point of the edge contour of the bulge area, draws a line connecting any two points of the edge contour of the bulge area through the center point based on the center point, and determines the bulge type based on the length of the line. The bulge type includes circular or elliptical types. The bulge type is compared with the longitudinal crack or transverse crack matched in the pipeline crack matching step, and a normal matching instruction and an abnormal matching instruction are output based on the comparison result.

6. A method for detecting bulges on the pipeline surface of a railway train according to claim 3, characterized in that: A pipeline system library is preset, and a fluid upper limit data set is stored in the pipeline system library. The fluid upper limit data set corresponds to the crack type one by one, and the fluid upper limit data set includes a pressure threshold N2, a flow rate threshold N3, and a temperature threshold N4; The method also includes a pipeline fluid feedback step, in which a corresponding fluid upper limit data set is indexed in a pipeline system library according to the crack type as a safety data set of the pipeline fluid.

7. A method for detecting bulges on the pipeline surface of a railway train according to claim 6, characterized in that: The pipeline system library stores a number of pipeline models; The pipeline abnormality analysis step selects any pipeline image, indexes the pipeline model corresponding to the pipeline image in the pipeline system library, and obtains the corresponding fluid data set according to the pipeline model.

8. A pipeline surface bulge detection system for a rail train, characterized in that: It includes a pipeline image acquisition module, which acquires an image of the pipeline system at the bottom of the train captured by a visual camera as a pipeline image, wherein the pipeline image includes an image after fluid is introduced into the pipeline and an image after fluid is extracted from the pipeline; The pipeline abnormality analysis module analyzes the pipeline image after the fluid is passed into the pipeline through a preset target detection model to obtain the bulge area on the pipeline surface, and synchronously maps the bulge area on the pipeline surface into the pipeline image after the fluid is extracted; The bulge tightening calculation module obtains the difference in pipeline size change after the fluid is introduced into the pipeline and after the fluid is extracted according to the pipeline image analysis, and calculates the pipeline size change difference and the fluid data set through a tightening value formula to obtain a pipeline surface tightening value, wherein the pipeline surface tightening value includes a tightening value of a bulge area on the pipeline surface and a tightening value of a normal area on the pipeline surface; the tightening value formula is configured as: ， in is the tightening value of the pipeline surface at time t; , , , They are respectively the pressure data, flow data, temperature data and pipeline size change in the pipeline at time t in the fluid data set; is the i-th data in the history; is the preset first adjustment parameter; It is the preset second adjustment parameter used to balance the impact of historical data; is the number of historical data; The pipeline crack matching module calculates the tightening value of the pipeline surface bulge area, the tightening value of the pipeline surface normal area and the fluid data set through the bulge and crack relationship formula to obtain a matching value, compares the matching value with a preset matching threshold N1, and outputs the crack type according to the comparison result; the bulge and crack relationship formula is configured as: ， ， ， in, is a multivariable function; is a two-variable function used to compare the difference in tightening values ​​between the normal area on the pipe surface and the bulging area on the pipe surface; is the matching value; It is the tightening value of the normal area on the pipe surface; It is the tightening value of the bulging area on the pipeline surface; The temperature value of the fluid entering the pipeline in the fluid data set; is the pressure value of the fluid entering the pipeline in the fluid data set; is the flow rate value of the fluid entering the pipeline in the fluid data set; They are the first constant, the second constant, the third constant, the fourth constant and the fifth constant, which are used to adjust the influence of each factor respectively.

Images