A method and system for detecting loose pipes of rail trains

Abstract

The present invention provides a pipeline loosening detection method and system for rail trains, comprising a pipeline image acquisition step, acquiring a pipeline image collected by a visual camera; a tie abnormality analysis step, determining whether tie features can be extracted from the pipeline image and determining whether the current tie is normal or abnormal through shape judgment after extraction; a pipeline loosening calculation step, when the current tie is abnormal, obtaining pipeline vibration frequency and pipeline vibration amplitude through shooting frequency analysis and relative position analysis of the pipeline in the pipeline image based on continuous pipeline images, and obtaining the pipeline loosening rate through a loosening rate formula with a fluid data set and a train running speed; a loosening upper limit feedback step, recording the duration under the pipeline loosening rate, and comparing it with a time-consuming upper limit threshold, and outputting a pipeline interface leakage instruction or a pipeline interface crisis instruction according to the comparison result; the present invention has the advantage of being able to determine the looseness of the pipeline when the train is running, and to estimate the leakage moment of the pipeline interface.

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 loose pipes of rail trains. 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, due to factors such as vibration, pressure changes or abnormal cable ties, pipelines may become loose and other abnormal conditions may occur. Loose pipelines may cause safety hazards such as leakage and bursting. Therefore, it is necessary to detect abnormal conditions such as loose pipelines in a timely and accurate manner to ensure the safe operation of trains.

[0003] At present, the detection methods for loose and leaking pipeline interfaces mainly include manual inspection, robot inspection or internal pipeline alarm decompression caused by leakage. For intelligent needs, it still mainly relies on robot inspection, which adopts multi-sensor fusion detection through sound waves or vision, but this type of detection is only applicable when the train is parked statically. In addition, this type of detection cannot predict in advance the looseness of the pipeline under various data when the cable tie is abnormal. 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 loose pipes in a rail train.

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

[0006] A method for detecting loose pipes 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;

[0008] A cable tie abnormality analysis step, extracting cable tie features through a preset target detection model according to the pipeline image, and determining whether the current cable tie is normal or abnormal according to the shape of the cable tie features if the cable tie features exist and if so, determining whether the current cable tie is normal or abnormal.

[0009] A pipeline looseness calculation step, when the current cable tie is abnormal, a plurality of continuous pipeline images are obtained by shooting with a certain visual camera, and the pipeline vibration frequency and pipeline vibration amplitude are obtained by analyzing the shooting frequency and the relative position of the pipeline in the pipeline image, and the pipeline vibration frequency, pipeline vibration amplitude, fluid data set and train speed are calculated by a looseness rate formula to obtain the pipeline looseness rate;

[0010] The loose upper limit feedback step indexes the pipeline loose rate into the preset pipeline loose table and the corresponding time upper limit threshold, wherein the time upper limit threshold reflects the upper limit of the duration of the pipeline at the loose rate, records the duration at the pipeline loose rate, compares the duration with the time upper limit threshold, and outputs a pipeline interface leakage instruction or a pipeline interface crisis instruction according to the comparison result.

[0011] Furthermore, it also includes a pipeline looseness verification step, in which a pipeline interface area map is obtained through feature extraction based on the pipeline image, and the corresponding pipeline interface type is indexed in the pipeline interface area map in a preset pipeline library, and the pipeline interface types include docking type and plug-in type. Then, the pipeline interface line of the pipeline connection is obtained from the pipeline interface area map through edge detection and identification, and the pipeline interface line is selected according to the pipeline interface type to output an interface positive position instruction or an interface misalignment instruction through a quantity judgment strategy or a position judgment strategy, and then the interface positive position instruction or the interface misalignment instruction is compared with the comparison result of the loose upper limit feedback step. If they are the same, an ignore instruction is output; if they are different, a prompt instruction is output.

[0012] Furthermore, when the pipeline interface type is a docking type, a quantity judgment strategy is selected, and the quantity judgment strategy includes judging whether the number of pipeline interface lines is the same as the number of pipeline interface lines when the pipeline interface is not loose at different times. If they are the same, an interface positive position instruction is output; if they are different, an interface misalignment instruction is output.

[0013] Furthermore, when the pipeline interface type is a plug-in type, a position judgment strategy is selected, and the position judgment strategy includes judging whether the position of the pipeline interface line at different times is the same as the position of the pipeline interface line when the pipeline interface is not loose. If they are the same, an interface positive position instruction is output; if they are different, an interface misalignment instruction is output.

[0014] Furthermore, the looseness rate calculation formula is configured as:

[0015] ,

[0016] in, is the integral function of the train speed, indicating the effect of the fluid pressure in the pipeline on the loosening rate as the train speed increases; It is the sum function of the vibration frequency and amplitude of the pipeline after the cable tie is detached or loosened, indicating the influence of different vibration frequencies and amplitudes on the loosening rate; is the train speed; 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; is the pipeline vibration frequency; is the pipeline vibration amplitude; It is the pipeline looseness rate, and its value range is between 0 and 1.

[0017] Furthermore, it also includes a pipeline leakage calculation step, obtaining the fluid pressure values ​​at both ends of the pipeline interface, subtracting the fluid pressure values ​​at both ends to obtain the pressure difference at both ends of the pipeline interface, and calculating the pipeline vibration frequency, pipeline vibration amplitude, pressure difference at both ends of the pipeline interface and fluid flow rate value through a leakage rate formula to obtain the leakage rate.

[0018] Furthermore, the leakage rate formula is configured as:

[0019] ,

[0020] in, is the pipeline interface leakage rate; is the pipeline vibration frequency; It is the vibration reference frequency when the pipeline interface is not loose; is the pipeline vibration amplitude; It is the vibration reference amplitude when the pipeline interface is not loose; is the pressure difference between the two ends of the pipeline interface; It is the flow rate value of the fluid entering the pipeline in the fluid data set.

[0021] Further, the current cable tie abnormality includes the current cable tie being lost, the current cable tie being displaced, and the current cable tie being loose.

[0022] Furthermore, the pipeline library stores several pipeline models;

[0023] The cable tie abnormality analysis step also includes a pipeline matching strategy, which includes selecting any pipeline image, indexing the corresponding pipeline model of the pipeline image in the pipeline library, and obtaining the corresponding fluid data set according to the pipeline model.

[0024] A pipeline loosening 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;

[0025] A cable tie abnormality analysis module is used to extract cable tie features through a preset target detection model according to the pipeline image, and determine whether the current cable tie is normal or abnormal according to the shape of the cable tie features if the cable tie features exist or not;

[0026] A pipeline looseness calculation module, when the current cable tie is abnormal, obtains a plurality of continuous pipeline images by shooting with a certain visual camera, obtains pipeline vibration frequency and pipeline vibration amplitude by analyzing the shooting frequency and the relative position of the pipeline in the pipeline image, and calculates the pipeline vibration frequency, pipeline vibration amplitude, fluid data set and train speed by a looseness rate formula to obtain the pipeline looseness rate;

[0027] The loose upper limit feedback module indexes the pipeline loose rate into the preset pipeline loose table to obtain the corresponding upper limit time threshold, the upper limit time threshold reflects the upper limit time that the pipeline lasts at the loose rate, records the duration at the pipeline loose rate, compares the duration with the upper limit time threshold, and outputs a pipeline interface leakage instruction or a pipeline interface crisis instruction according to the comparison result.

[0028] The beneficial effects of the present invention are as follows: by monitoring and identifying the cable ties on the pipeline and the pipeline itself through visual images, and then processing the data of the fluid in the pipeline, it is possible to determine the looseness of the pipeline under the above circumstances, and then estimate the leakage moment of the pipeline interface by predicting the time upper limit corresponding to the loosening of the pipeline, so as to judge the criticality of the pipeline in advance. Specifically, when an abnormality of the cable tie is detected, the vibration frequency and vibration amplitude of the pipeline are further analyzed, and the loosening rate of the pipeline is calculated in combination with the fluid data set and the train speed, so that the possible loosening problem of the pipeline can be discovered in time, and the time-consuming upper limit threshold is obtained by indexing the pipeline loosening rate with the preset pipeline loosening table, and the duration is monitored in real time. Once the threshold is exceeded, the corresponding instruction is immediately output, which is helpful to take timely measures to avoid the deterioration of the problem, thereby improving the intelligence level and automation level of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is the overall flow chart of the present invention;

[0030] Figure 2 It is a flow chart of pipeline looseness verification in the present invention;

[0031] Figure 3 It is a flow chart for analyzing pipeline leakage in the present invention. DETAILED DESCRIPTION

[0032] 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.

[0033] At present, the detection methods for loose and leaking pipeline interfaces mainly include manual inspection, robot inspection or internal pipeline alarm decompression caused by leakage. For intelligent needs, it mainly relies on robot inspection, and the technical means adopted is multi-sensor fusion detection through sound waves or vision, but this kind of detection is only applicable to when the train is parked statically. In addition, this kind of detection cannot predict in advance the looseness of the pipeline under various data when the cable tie is abnormal. 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 a pipeline loosening detection method for such rail trains.

[0034] like Figure 1 As shown, the following steps are included:

[0035] In the pipeline image acquisition step, a high-resolution, low-noise visual camera is used to ensure that a clear and complete image of the pipeline system at the bottom of the train (pipeline image) can be captured even in a complex light environment. The pipeline system includes several pipelines. The collected pipeline images are then pre-processed by denoising and contrast enhancement to facilitate the recognition accuracy of the subsequent target detection model.

[0036] A pipeline library is preset, which stores several pipeline models. Each pipeline model corresponds to the name of the fluid in the pipeline, the fluid properties, the fluid data set and the interface type of the pipeline;

[0037] The cable tie abnormality analysis step includes a pipeline matching strategy, which is: select any pipeline image, index the corresponding pipeline model of the pipeline image in the pipeline library (the indexing method applies a target detection model, and the pipeline image is substituted into the preset target detection model to output the corresponding pipeline model), and obtain the corresponding fluid data set and other corresponding information according to the pipeline model;

[0038] Cable tie object detection model training: A cable tie feature recognition target detection model is trained in advance through a convolutional neural network. A large number of cable tie shapes, sizes, bundling positions, and other cable tie features are divided into a certain proportion. Training, verification, and testing are completed so that the model can accurately distinguish cable ties from other pipeline components and extract area maps with only cable ties.

[0039] Specific judgment of cable ties: compare the extracted cable tie features with the shape of the cable tie during normal bundling, and analyze the shape deviation of the cable tie features to determine whether the current cable tie is normal or abnormal. Once the current cable tie is abnormal, an abnormal cable tie instruction is immediately sent to the train driving system so that the background or maintenance department can record and carry out targeted repairs;

[0040] The current cable tie abnormality includes the current cable tie loss, the current cable tie displacement and the current cable tie looseness. The specific matching is as follows: cable tie loss includes: ①, template matching, creating a template image containing the normal cable tie shape when tied, and judging through template matching; ②, color or contour analysis, comparing the image of the cable tie that is not lost with the real-time image, judging whether the color of the cable tie binding position is the same as the corresponding color in the image when the cable tie is not lost, and the contour comparison is the same;

[0041] Cable tie displacement includes: 1. edge detection, using edge detection algorithm to identify the edge of the cable tie, and judging whether the cable tie is displaced by calculating the position deviation between the edge position and the initial normal cable tie binding position; 2. shape matching, using shape matching algorithm (such as Hu moment, shape context, etc.) to match the cable tie shape in the real-time image with the preset normal shape. If the shape matching degree is lower than the preset threshold, it is judged that the cable tie is displaced;

[0042] Loose cable ties include: 1. Tension analysis. If the cable tie is loose and the pipeline will vibrate, the tension information of the cable tie (such as bending degree, stretching degree, etc.) is extracted through image processing, and the looseness of the cable tie is judged based on the tension information, or by judging the shape change of the cable tie under pipeline vibration; 2. Pipeline vibration analysis. If the cable tie is loose, the amplitude of pipeline vibration will increase, but the pipeline vibration will be restricted by the cable tie and vibrate within a specified range. Therefore, the judgment is made based on the vibration law and vibration range of the cable tie.

[0043] Calculation steps for pipeline looseness and specific judgment of pipeline vibration: When the current cable tie is abnormal, use a high-speed visual camera to continuously shoot multiple images of the pipeline from the same perspective, which can ensure that the vibration or swing of the pipeline is captured, and then align the continuous images to ensure that the baseline of the pipeline background in each image remains unchanged, and then construct the two-dimensional coordinates of the background baseline or construct the three-dimensional coordinates of the background baseline through images from two perspectives. According to the edge detection algorithm, the edge contour of the pipeline is identified, and the edge contour is mapped to the coordinate system. Then, the coordinate changes of the edge contour in different images are judged, or any multiple reference points on the pipeline surface are selected and mapped to the coordinate system. Then, the coordinate changes of each reference point in different images are judged. The vibration frequency and vibration amplitude of the pipeline can be calculated by combining the time difference between the continuous image shooting;

[0044] Calculation of pipeline looseness rate: The pipeline vibration frequency and pipeline vibration amplitude are combined and extracted into the current fluid data set of the pipeline and the train running speed, and the pipeline looseness rate is calculated through the looseness rate formula, so that the possible looseness problem of the pipeline can be discovered in time;

[0045] Specifically, the loose rate calculation formula is configured as follows:

[0046] ,

[0047] in, is the integral function of the train speed, indicating the effect of the fluid pressure in the pipeline on the loosening rate as the train speed increases; It is the sum function of the vibration frequency and amplitude of the pipeline after the cable tie is detached or loosened, indicating the influence of different vibration frequencies and amplitudes on the loosening rate; is the train speed; 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; is the pipeline vibration frequency; is the pipeline vibration amplitude; It is the pipeline looseness rate, and its value range is between 0 and 1;

[0048] Normally, the pressure and flow rate of the fluid in the pipeline remain unchanged. When the train is stopped, the train speed is 0, at this time, the pipeline is no longer affected by the movement of the train and vibrates, so the calculated looseness rate is zero. At this time, this scheme does not consider the detection of pipeline looseness when the train is stationary, but considers the detection of pipeline looseness when the train is moving. If the value is 0, the calculated looseness rate is between 0 and 1 in percentage. When the train speed is higher, the vibration of the pipeline is relatively more obvious, so the looseness rate value is larger.

[0049] The loose upper limit feedback step is to index the pipeline loose rate into the preset pipeline loose table and the corresponding time consumption upper limit threshold value, the time consumption upper limit threshold value reflects the upper limit of the duration of the pipeline under the loose rate, and the duration under the pipeline loose rate is recorded, and the duration is compared with the time consumption upper limit threshold value, and the pipeline interface leakage instruction or pipeline interface crisis instruction is output according to the comparison result;

[0050] For example: if the pipeline looseness rate is 60%, the upper limit time of leakage in the corresponding matching pipeline interface under the pipeline looseness rate is t1, and the timing starts when the looseness rate reaches 60%. When the train travel time reaches t1, the pipeline interface leakage instruction is output. Within the range of t1, the pipeline interface crisis instruction is output. In addition, in order to consider the changes in the train travel speed and the data changes of the fluid in the pipeline, the pipeline looseness rate will change. Therefore, it is necessary to associate the looseness rate, time and pipeline interface leakage. This scheme only considers the association between the looseness rate and leakage when the train is at a constant speed and the fluid data in the pipeline remains unchanged.

[0051] The advantages of the present invention are: by monitoring and identifying the cable ties on the pipeline and the pipeline itself through visual images, and then processing the data of the fluid in the pipeline, it is possible to determine the looseness of the pipeline under the above circumstances, and then estimate the leakage moment of the pipeline interface by predicting the time upper limit corresponding to the loosening of the pipeline, so as to judge the criticality of the pipeline in advance. Specifically, when the cable tie abnormality is detected, the pipeline vibration frequency and vibration amplitude are further analyzed, and the pipeline loosening rate is calculated in combination with the fluid data set and the train speed, so that the possible loosening problem of the pipeline can be discovered in time, and the time-consuming upper limit threshold is obtained by indexing the pipeline loosening rate with the preset pipeline loosening table, and the duration is monitored in real time. Once the threshold is exceeded, the corresponding instruction is immediately output, which is helpful to take measures in time to avoid the deterioration of the problem, thereby improving the intelligence level and automation level of the system.

[0052] like Figure 2 As shown, it also includes a pipeline looseness verification step, in which a pipeline interface area map is obtained through feature extraction based on the pipeline image, and the corresponding pipeline interface type is indexed in the pipeline library in the pipeline interface area map. The pipeline interface types include docking type and plug-in type, wherein the indexing of the pipeline interface type in the pipeline library mainly relies on feature matching of the image, and the sizes of the two pipeline interfaces of the docking type are almost the same, while the two pipeline interfaces of the plug-in type are generally one-in-one, so the pipeline interface line of the pipeline connection is obtained from the pipeline interface area map through edge detection and identification, and the pipeline interface line is selected according to the pipeline interface type, and the interface positive position instruction or the interface misalignment instruction is output through the quantity judgment strategy or the position judgment strategy, and then the interface positive position instruction or the interface misalignment instruction is compared with the comparison result of the loose upper limit feedback step, if the two are the same, the ignore instruction is output, and if they are different, the prompt instruction is output.

[0053] Specifically, when the pipeline interface type is a docking type, a quantity judgment strategy is selected, and the quantity judgment strategy includes judging whether the number of pipeline interface lines is the same as the number of pipeline interface lines when the pipeline interface is not loose at different times, and if they are the same, outputting an interface positive position instruction, and if they are different, outputting an interface misalignment instruction, wherein when the pipeline interface is not loose, the number of pipeline interface lines is recorded as a baseline, and secondly, at different times, judging whether the number of pipeline interface lines is the same as the number of the baseline;

[0054] When the pipeline interface type is a plug-in type, a position judgment strategy is selected. The position judgment strategy includes judging whether the position of the pipeline interface line is the same as the position of the pipeline interface line when the pipeline interface is not loose at different times. If they are the same, an interface positive position instruction is output; if they are different, an interface misalignment instruction is output; the main method is to judge the real-time position of the pipeline interface line and the position of the baseline. Under normal circumstances, if the pipeline interface is loose or offset, the pipeline interface line and the baseline will be separated or the pipeline interface line will be skewed relative to the baseline. Both of the above situations can be used to judge and output the interface misalignment instruction; the above-mentioned detection results, instruction output and other information are recorded in the log for subsequent tracing. When the interface is misaligned or loosened for multiple consecutive times, the fault warning mechanism is triggered, and a warning is issued to the train control system for subsequent targeted maintenance.

[0055] Since the vibration of the pipeline will cause the pipeline interface to loosen, the loosening of the pipeline interface will cause the leakage of the fluid in the pipeline. Figure 3 As shown, it also includes a pipeline leakage calculation step, obtaining the fluid pressure values ​​at both ends of the pipeline interface, and obtaining the pressure difference at both ends of the pipeline interface by subtracting the fluid pressure values ​​at both ends. The factors causing the pipeline to vibrate and leak are the vibration frequency and amplitude of the pipeline, which will cause the pipeline interface to be squeezed or stretched, and easily misplaced or shifted. The pressure difference at both ends of the pipeline interface and the flow rate value of the fluid will cause the fluid to impact in the pipeline, which will easily cause the pipeline interface to be misplaced or shifted. Therefore, the pipeline vibration frequency, pipeline vibration amplitude, the pressure difference at both ends of the pipeline interface and the flow rate value of the fluid are calculated through the leakage rate formula to obtain the leakage rate.

[0056] Specifically, the leakage rate calculation formula is configured as:

[0057] ,

[0058] in, is the pipeline interface leakage rate; is the pipeline vibration frequency; It is the vibration reference frequency when the pipeline interface is not loose; is the pipeline vibration amplitude; It is the vibration reference amplitude when the pipeline interface is not loose; is the pressure difference between the two ends of the pipeline interface; is the flow rate value of the fluid entering the pipeline in the fluid data set;

[0059] By comparing the difference between the current pipeline vibration frequency and the reference frequency, as well as the difference between the current pipeline vibration amplitude and the reference amplitude, we can evaluate whether the pipeline may leak. At the same time, considering that the increase in pressure difference and flow rate will also increase the possibility of leakage when the degree of pipeline looseness remains unchanged, The range of is [0,∞), when When it is 0, it means there is no leakage. When >0, it means there are leaks of different possibilities. If the value exceeds the threshold, it means that the pipeline interface has 100% leakage, and it can be verified by the pressure difference at both ends of the pipeline interface. When it is completely leaked, the pressure value at one end (fluid outlet end) is almost 0, and the pressure value at the other end (fluid inlet end) is normal.

[0060] Correspondingly, a system for monitoring the looseness of pipelines of rail trains is provided, including 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;

[0061] The cable tie abnormality analysis module extracts the cable tie features through the preset target detection model according to the pipeline image, and determines whether the current cable tie is normal or abnormal according to the shape of the cable tie features if they exist;

[0062] The pipeline looseness calculation module, when the current cable tie is abnormal, obtains a number of continuous pipeline images based on a visual camera, and obtains the pipeline vibration frequency and pipeline vibration amplitude through shooting frequency analysis and relative position analysis of the pipeline in the pipeline image. The pipeline vibration frequency, pipeline vibration amplitude, fluid data set and train speed are calculated through the looseness rate formula to obtain the pipeline looseness rate;

[0063] The loose upper limit feedback module indexes the corresponding time upper limit threshold of the pipeline loose rate in the preset pipeline loose table. The time upper limit threshold reflects the upper limit of the time that the pipeline lasts at the loose rate. The duration at the pipeline loose rate is recorded, and the duration is compared with the time upper limit threshold. According to the comparison result, the pipeline interface leakage instruction or the pipeline interface crisis instruction is output.

[0064] 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 loose pipes of a rail 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; A cable tie abnormality analysis step, extracting cable tie features through a preset target detection model according to the pipeline image, and determining whether the current cable tie is normal or abnormal according to the shape of the cable tie features if the cable tie features exist and if so, determining whether the current cable tie is normal or abnormal. A pipeline looseness calculation step, when the current cable tie is abnormal, a plurality of continuous pipeline images are obtained by shooting with a certain visual camera, and the pipeline vibration frequency and pipeline vibration amplitude are obtained by analyzing the shooting frequency and the relative position of the pipeline in the pipeline image, and the pipeline vibration frequency, pipeline vibration amplitude, fluid data set and train speed are calculated by a looseness rate formula to obtain the pipeline looseness rate; The looseness rate calculation formula is configured as: ， in, is the integral function of the train speed, indicating the effect of the fluid pressure in the pipeline on the loosening rate as the train speed increases; It is the sum function of the vibration frequency and amplitude of the pipeline after the cable tie is detached or loosened, indicating the influence of different vibration frequencies and amplitudes on the loosening rate; is the train speed; 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; is the pipeline vibration frequency; is the pipeline vibration amplitude; It is the pipeline looseness rate, and its value range is between 0 and 1; The loose upper limit feedback step indexes the corresponding upper limit time threshold of the pipeline loose rate in the preset pipeline loose table, the upper limit time threshold reflects the upper limit time that the pipeline continues at the loose rate, records the duration at the pipeline loose rate, compares the duration with the upper limit time threshold, and outputs the pipeline interface leakage instruction or the pipeline interface crisis instruction according to the comparison result.

2. A method for detecting loose pipelines of a railway train according to claim 1, characterized in that: The method also includes a pipeline looseness verification step, in which a pipeline interface area map is obtained through feature extraction based on the pipeline image, and the corresponding pipeline interface type is indexed in the preset pipeline library in the pipeline interface area map, and the pipeline interface types include docking type and plug-in type. Then, the pipeline interface line of the pipeline connection is obtained from the pipeline interface area map through edge detection and identification, and the pipeline interface line is selected according to the pipeline interface type to output an interface correct position instruction or an interface misalignment instruction through a quantity judgment strategy or a position judgment strategy. Then, the interface correct position instruction or the interface misalignment instruction is compared with the comparison result of the loose upper limit feedback step. If they are the same, an ignore instruction is output; if they are different, a prompt instruction is output.

3. A method for detecting loose pipes in a railway train according to claim 2, characterized in that: When the pipeline interface type is a docking type, a quantity judgment strategy is selected, and the quantity judgment strategy includes judging at different times whether the number of pipeline interface lines is the same as the number of pipeline interface lines when the pipeline interface is not loose. If they are the same, an interface positive position instruction is output; if they are different, an interface misalignment instruction is output.

4. A method for detecting loose pipes in a railway train according to claim 2, characterized in that: When the pipeline interface type is a plug-in type, a position judgment strategy is selected. The position judgment strategy includes judging whether the position of the pipeline interface line is the same as the position of the pipeline interface line when the pipeline interface is not loose at different times. If they are the same, an interface positive position instruction is output; if they are different, an interface misalignment instruction is output.

5. The method for detecting loose pipelines of a railway train according to claim 1, characterized in that: The method also includes a pipeline leakage calculation step, obtaining the fluid pressure values ​​at both ends of the pipeline interface, subtracting the fluid pressure values ​​at both ends to obtain the pressure difference at both ends of the pipeline interface, and calculating the pipeline vibration frequency, pipeline vibration amplitude, pressure difference at both ends of the pipeline interface and flow rate value of the fluid through a leakage rate formula to obtain the leakage rate; The leakage rate formula is configured as: ， in, is the pipeline interface leakage rate; is the pipeline vibration frequency; It is the vibration reference frequency when the pipeline interface is not loose; is the pipeline vibration amplitude; It is the vibration reference amplitude when the pipeline interface is not loose; is the pressure difference between the two ends of the pipeline interface; It is the flow rate value of the fluid entering the pipeline in the fluid data set.

6. A method for detecting loose pipes in a railway train according to claim 1, characterized in that: The current cable tie abnormality includes the current cable tie being lost, the current cable tie being displaced, and the current cable tie being loose.

7. A method for detecting loose pipelines of a railway train according to claim 2, characterized in that: The pipeline library stores a number of pipeline models; The cable tie abnormality analysis step also includes a pipeline matching strategy, which includes selecting any pipeline image, indexing the corresponding pipeline model of the pipeline image in the pipeline library, and obtaining the corresponding fluid data set according to the pipeline model.

8. A pipeline loosening 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; A cable tie abnormality analysis module is used to extract cable tie features through a preset target detection model according to the pipeline image, and determine whether the current cable tie is normal or abnormal according to the shape of the cable tie features if the cable tie features exist or not; A pipeline looseness calculation module, when the current cable tie is abnormal, obtains a plurality of continuous pipeline images by shooting with a certain visual camera, obtains pipeline vibration frequency and pipeline vibration amplitude by analyzing the shooting frequency and the relative position of the pipeline in the pipeline image, and calculates the pipeline vibration frequency, pipeline vibration amplitude, fluid data set and train speed by a looseness rate formula to obtain the pipeline looseness rate; The looseness rate calculation formula is configured as: ， in, is the integral function of the train speed, indicating the effect of the fluid pressure in the pipeline on the loosening rate as the train speed increases; It is the sum function of the vibration frequency and amplitude of the pipeline after the cable tie is detached or loosened, indicating the influence of different vibration frequencies and amplitudes on the loosening rate; is the train speed; 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; is the pipeline vibration frequency; is the pipeline vibration amplitude; It is the pipeline looseness rate, and its value range is between 0 and 1; The loose upper limit feedback module indexes the pipeline loose rate into the preset pipeline loose table and the corresponding time upper limit threshold, the time upper limit threshold reflects the upper limit of the duration of the pipeline at the loose rate, records the duration at the pipeline loose rate, compares the duration with the time upper limit threshold, and outputs a pipeline interface leakage instruction or a pipeline interface crisis instruction according to the comparison result.

Images