A method for detecting the offset of parallel pipelines under a train.

Abstract

This invention relates to the field of image processing technology and discloses a method for detecting the offset of parallel pipelines under a train. The invention establishes a static reference information acquisition step, a dynamic information acquisition step, a trend judgment step, and a loosening warning step. Images are acquired using image processing methods during both stationary and moving train phases. The method then determines whether there is a high risk of pipeline loosening, facilitating early warning before loosening occurs and allowing maintenance personnel to intervene in advance, preventing pipeline damage caused by collisions. The trend judgment step consists of two parts: risk pre-screening and loosening trend score calculation. In the risk pre-screening step, a preliminary scoring threshold is calculated using partial data to determine if high-risk pipeline sections exist. If high-risk sections are found, a loosening trend score is calculated to quickly identify high-risk pipelines. The loosening trend score calculation step incorporates actual operating scenarios to improve the accuracy of risk assessment.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, specifically to an application for detecting the offset of parallel pipelines under a train. It is particularly suitable for complex scenarios, using image processing technology to extract image features and perform dynamic analysis to achieve early warning of pipeline loosening risks. Background Technology

[0002] The piping under the railcar is a crucial component of the train, responsible for transporting fluids. To prevent collisions between pipes, cable ties, pipe clamps, and other fasteners are typically used to secure them side-by-side to the train's underside. During high-speed train operation, these fasteners may loosen, affecting the effectiveness of the pipe fixation. Therefore, regular inspections of the pipe fastener installation are necessary.

[0003] With the application of image processing technology in industrial inspection, image-based non-contact inspection has become the mainstream approach. Conventional inspection methods typically involve acquiring images of the location of anti-loosening markers and then using image detection methods to determine if the markers have shifted or deviated after installation. If the markers have shifted or deviated, it is considered loose and requires re-fixing. However, this image processing method focuses only on the single feature of the anti-loosening marker, simply capturing changes in its position within the image. It cannot establish a correlation between continuous changes in features in a dynamic image, making it difficult to reflect the potential trend of pipeline loosening. Furthermore, this image detection method can only detect problems after loosening has occurred, failing to provide early warnings or reliable basis for risk prediction. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the present invention aims to provide a method for detecting the offset of parallel pipelines under a train. Through steps such as static reference information acquisition, dynamic information extraction, trend judgment, and loosening warning, the method achieves early warning and accurate positioning of the risk of loosening of pipelines under the train, thereby improving detection efficiency and accuracy.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] The static reference information acquisition step involves acquiring an image of the pipeline under the train in a stationary state, defining it as static reference image information. From this static reference image information, basic pipeline information is obtained, including pipeline specifications, clamping morphology data, clamping length, and the medium flowing through it.

[0007] The dynamic information acquisition step involves acquiring images of the pipelines under the train during operation, defining them as dynamic image information, and simultaneously collecting train operation data. Dynamic information, reflecting the degree of pipeline offset, is then extracted from this dynamic image information.

[0008] The trend assessment step, based on basic information, train operation data, and dynamic information, predicts the loosening risk index for each pipe section. This loosening risk index reflects the likelihood of the corresponding pipe section becoming loose in the future.

[0009] The loosening warning step is configured with preset loosening warning conditions. If the loosening warning conditions are met, a loosening warning signal is generated.

[0010] In this invention, preferably, the static reference information acquisition step is configured with a pipe segment identification strategy, including:

[0011] The pipeline region and clamping region are identified in the static reference image information. The pipeline region is specifically the area of ​​the pipeline in the static reference image information, and the clamping region is specifically the area of ​​the pipeline covered by pipe fasteners. The clamping region divides the pipeline region into at least two pipe segment regions.

[0012] Identify the long side boundaries of the pipe segment region and the centerline of the pipe segment. The long side boundaries represent the two boundaries of the pipe segment region along the length of the pipe segment. Any point on the centerline of the pipe segment is equidistant from the two long side boundaries.

[0013] The length of the pipe segment is obtained by acquiring the length of the centerline of the pipe segment, and the length of the pipe covered by the clamping area is obtained to acquire the clamping length.

[0014] The pipe segment curvature distribution data and pipe segment width distribution data are obtained as the clamping morphology data. The pipe segment curvature distribution data represents the distribution of the curvature of the pipe segment centerline in the length direction, and the pipe segment width distribution data specifically represents the width distribution of the pipe segment region in the length direction of the pipeline.

[0015] In this invention, preferably, the dynamic information acquisition step is configured with a dynamic information extraction strategy, including:

[0016] Acquire continuous dynamic image information, and identify each pipe segment in each dynamic image information.

[0017] The system acquires the variation curve of the pipe segment spacing and the position offset curve of each clamping point. The pipe segment spacing is specifically the distance between adjacent pipe segments, and the clamping point represents the center point of the clamping area. The position offset curve reflects the change in the position offset of the clamping point in continuous dynamic image information.

[0018] The vibration amplitude and frequency of the pipe segment are calculated based on the variation curve of the pipe segment spacing.

[0019] In this invention, preferably, the loosening risk indicator includes a loosening trend score, and the loosening warning condition includes the loosening trend score exceeding a preset loosening score threshold within a preset future time period, or the rate of change of the loosening trend score exceeding a preset loosening trend change rate threshold.

[0020] In this invention, preferably, the loosening trend score includes a basic score item, a dynamic score item, and an environmental amplification item. The basic score item reflects the degree of influence of the basic data on the loosening trend score, the dynamic score item reflects the degree of influence of the dynamic information on the loosening trend score, and the environmental amplification item reflects the degree of influence of train operation data on the loosening trend score.

[0021] In this invention, preferably, the offset detection method is configured with a risk pre-screening strategy, including:

[0022] A preliminary risk score for the pipe segment is calculated based on the clamping length, pipe segment spacing, and pipe vibration amplitude.

[0023] The initial risk score is compared with the initial score threshold. If the initial risk score exceeds the preset initial score threshold, the pipe section is defined as a high-risk pipe section and the loosening trend score is calculated based on the preset early warning score calculation strategy.

[0024] In this invention, preferably, the risk pre-screening strategy further includes a group comparison sub-strategy, comprising:

[0025] Preliminary risk scores were obtained for multiple pipe segments arranged side-by-side under the same vehicle chassis and used as a group comparison group.

[0026] Calculate the regional mean and regional variance among the preliminary risk scores within the group comparison group. If the regional variance exceeds a preset variance threshold, it is determined that there is a differential pipe section, and the differential pipe section is defined as a high-risk pipe section and a loosening trend score is calculated.

[0027] In this invention, preferably, the pipe segment curvature distribution data is configured with a curvature correction strategy, including correcting the pipe segment curvature distribution data based on the elastic modulus of the pipe material and the type of the flowing medium. The corrected pipe segment curvature distribution data is used to optimize the influence weight of the clamping shape on the loosening trend in the basic scoring item.

[0028] In this invention, preferably, the calculation of the loosening trend score is configured with a time decay correction strategy, including:

[0029] Record the cumulative runtime and historical maintenance records of the pipe section. The historical maintenance records include the time points for fastener replacement and pipe adjustment.

[0030] Based on the cumulative running time and maintenance interval, a time decay coefficient is set, which increases with the increase of the cumulative running time and the increase of the time after maintenance.

[0031] The loosening trend score is corrected using a time decay coefficient. The corrected score is used to adjust the loosening score threshold in the early warning conditions. The loosening score threshold is negatively correlated with the time decay coefficient.

[0032] In this invention, preferably, the offset detection method is configured with a graded early warning output strategy, including:

[0033] Based on the numerical range of the loosening trend score, the warning signals are divided into Level 1, Level 2, and Level 3 warning signals. The Level 1 warning signal indicates that intervention and maintenance are required. The Level 2 warning signal indicates that the loosening standard has not been met but continuous monitoring is required. The Level 3 warning signal indicates that the loosening standard has not been met and continuous monitoring is not required.

[0034] The beneficial effects of this invention are:

[0035] 1. This invention, through the setting of static reference information acquisition steps, dynamic information acquisition steps, trend judgment steps and loosening warning steps, uses image processing methods to collect images in two stages: when the train is stationary and when the train is running. Then, it judges whether there is a high risk of pipe section loosening, which facilitates early warning before loosening occurs, so that maintenance personnel can intervene in maintenance in advance and avoid pipe damage caused by collision between pipes.

[0036] 2. Trend judgment is divided into two stages: risk pre-screening and loosening trend score calculation. In the risk pre-screening stage, a preliminary score threshold is calculated using some data to determine whether there are high-risk pipe sections. If high-risk pipe sections are found, the loosening trend score is calculated. This solution simplifies the calculation process. When calculating the preliminary score threshold, there is no need to call environmental parameters, which effectively reduces the amount of calculation and makes it easier to quickly screen out high-risk pipes, reducing the workload of subsequent inspections. The loosening trend score calculation stage is combined with the actual operation scenario to improve the accuracy of risk assessment. Attached Figure Description

[0037] Figure 1 This is a schematic diagram of the overall process of the present invention;

[0038] Figure 2 This is a flowchart illustrating the pipe segment identification strategy in this invention;

[0039] Figure 3 This is a schematic diagram of the clamping area and the pipe segment area in this invention;

[0040] Figure 4 This is a flowchart illustrating the dynamic information extraction strategy in this invention;

[0041] Figure 5 This is a flowchart illustrating the risk pre-screening strategy in this invention;

[0042] Figure 6 This is a flowchart illustrating the group comparison sub-strategy in this invention. Detailed Implementation

[0043] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0044] It should be noted that when a component is described as "fixed to" another component, it can be directly on the other component or may have a component in between. When a component is considered "connected to" another component, it can be directly connected to the other component or may have a component in between. When a component is considered "set on" another component, it can be directly set on the other component or may have a component in between. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0045] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0046] Please also see Figures 1 to 6 This embodiment provides a method for detecting the offset of parallel pipelines under a train, including a static reference information acquisition step, a dynamic information acquisition step, and a trend judgment step.

[0047] The static reference information acquisition step involves acquiring images of the pipelines under the train in a stationary state, defining these as static reference image information. Basic pipeline information is then extracted from this static reference image information, including pipeline specifications, clamping morphology data, clamping length, and the medium flowing through it. Static reference image information is acquired using a camera carried by an inspection robot. When the train enters the station and is stationary, the pipeline system is shut off to prevent the flow of the medium from affecting the pipeline morphology. The inspection robot enters the underside of the train via the tracks to take pictures, and the captured images are defined as static reference image information. The static reference image information must clearly show the overall shape of the pipelines, the position of the clamping components, and the relative positions between pipelines. This basic information serves as a benchmark for subsequent dynamic offset comparisons and risk assessments.

[0048] In the static reference image information, basic pipeline information is obtained through a preset pipeline segment identification strategy. This basic information includes pipeline specifications, clamping morphology data, clamping length, pipeline segment length, and the circulating medium. The pipeline specifications can be determined by combining features from the static reference image information with information about the pipelines actually used on the current train. Pipeline specifications include outer diameter, wall thickness, and material. The circulating medium includes medium type and density; medium types include brake fluid, hydraulic oil, and coolant. The clamping morphology data reflects the morphological characteristics of the pipeline under the constraint of the clamping area, including pipeline segment curvature distribution data and pipeline segment width distribution data. Specifically, the pipeline segment curvature distribution data refers to the distribution of the curvature of the pipeline centerline along the length of the pipeline segment. The pipe segment centerline refers to a straight line or curve equidistant from any point within the pipe segment area to the two long side boundaries along the pipe segment's length. It is obtained through image fitting. The pipe segment width distribution data specifically reflects the width variation of the pipe segment area along the pipe's length. Width refers to the dimension of the pipe segment area perpendicular to its length, while curvature is a physical quantity describing the degree of curvature of a curve. The clamping length refers to the constraint length of the clamping area on the pipe, i.e., the length of the pipe segment centerline within the clamping area. The pipe segment length is the length of the pipe segment centerline. These basic information serve as the static basis for loosening risk indicators, and their impact on risk essentially determines the inherent resistance of the pipeline to loosening. This basic information will have varying degrees of influence on the loosening risk indicators. Clamping length is negatively correlated with the loosening risk indicator; the longer the clamping length, the lower the loosening risk indicator, and vice versa. Pipe segment curvature distribution data and pipe segment width distribution data can reflect the stability of the pipeline's shape under the constraint of the clamping area. The greater the curvature and the greater the variation in pipe segment width, the higher the loosening risk indicator. The higher the density of the circulating medium, the greater the inertial force of the liquid column during pipeline operation, and the higher the risk of loosening, making it easier for loosening to occur.

[0049] The specific implementation process of the pipe segment identification strategy includes: identifying the pipe area and clamping area in the static reference image information. The pipe area is specifically the pixel area occupied by the pipe in the static reference image information, and the clamping area is specifically the pixel area of ​​the pipe covered by pipe fasteners. Since parallel pipes under a train are usually fixed at intervals by multiple fasteners, the clamping area divides each pipe area into at least two pipe segment areas. Then, an edge detection algorithm is used to identify the long side boundary of each pipe segment area. The long side boundary represents the two boundaries of the pipe segment area along the pipe segment length direction, i.e., the pixel boundaries corresponding to the two sides of the pipe segment. The pipe segment centerline is then fitted to ensure that the pixel distance from any point on the pipe segment centerline to the two long side boundaries is equal. The pixel distance is converted into an actual distance based on the calibration parameters of the industrial camera. Finally, the actual length of the pipe segment centerline within the pipe area covered by the clamping area is measured; this length is the clamping length, and the length of the pipe segment centerline is taken as the pipe segment length. Finally, along the length of the pipe segment's centerline, several measurement points are taken at equal intervals. The curvature at each measurement point is calculated by fitting an arc to the coordinates of three adjacent measurement points. The curvature is the reciprocal of the arc radius, thus obtaining the pipe segment curvature distribution data. At the same time, the width of the pipe segment area at each measurement point is measured. The width is the actual dimension perpendicular to the centerline direction, thus obtaining the pipe segment width distribution data.

[0050] It should be noted that all lengths acquired in the static and dynamic image information in this application are pixel lengths. Therefore, these length data need to be converted to ensure a consistent length benchmark. For example, the outer diameter of a pipe can be used as the benchmark for conversion. When a certain type of pipe is identified, its outer diameter can be obtained, and the corresponding pixel length of the pipe in the image can also be acquired. Dividing the outer diameter by the pixel length gives the length benchmark in the image.

[0051] To improve the accuracy of clamping shape data, a curvature correction strategy is implemented for the pipe segment curvature distribution data. This strategy corrects the pipe segment curvature distribution data based on the elastic modulus of the pipe material and the type of flowing medium. Different pipe materials have different elastic moduli; for example, stainless steel has an elastic modulus of 200 GPa, and aluminum alloy has an elastic modulus of 70 GPa. The smaller the elastic modulus, the more easily the pipe deforms due to its own weight or medium pressure, leading to a deviation between the measured curvature and the actual curvature under working conditions. Simultaneously, different types of flowing media have different effects on pipe pressure. For example, brake fluid has a static pressure of approximately 0.1 MPa, while hydraulic oil has a static pressure of approximately 0.3 MPa. The higher the pressure, the more pronounced the pipe deformation. The specific implementation process of the curvature correction strategy is as follows: First, obtain the elastic modulus E of the pipe material and the pressure coefficient P corresponding to the type of flowing medium. For brake fluid, P is set to 1.0, and for hydraulic oil, P is set to 1.2. Then, correct the curvature of each measurement point according to the correction formula:

[0052] ,

[0053] in The corrected curvature. To measure the curvature, For reference, the elastic modulus is usually the elastic modulus of stainless steel; P is the pressure coefficient of the flowing medium, which is preset. The corrected pipe curvature distribution data is used to optimize the influence weight of the clamping form on the loosening trend in the basic scoring item. For example, the larger the corrected curvature, the more obvious the deformation of the pipeline under actual working conditions, and the higher the influence weight of the clamping form on the loosening trend.

[0054] The dynamic information acquisition step is used to acquire dynamic information of the pipeline and synchronous train operation data during train operation. The dynamic information reflects the offset state of the pipeline under actual working conditions. This step is completed by an industrial camera pre-installed on the bottom of the train. When the train is running on the normal operating line, the industrial camera continuously acquires images of the pipeline under the train. These images are defined as dynamic image information. At the same time, train operation data is acquired synchronously, including train speed v, track smoothness R, ambient temperature T, and ambient humidity H. Track smoothness R reflects the unevenness of the track surface and is acquired and converted by the train wheel-rail vibration sensor. Dynamic information is obtained from the dynamic image information through a preset dynamic information extraction strategy. The dynamic information includes the change curve of the pipe segment spacing, the position offset curve of each clamping point, the vibration amplitude of the pipe segment, and the vibration frequency of the pipe segment.

[0055] In the specific implementation of the dynamic information extraction strategy, the same target detection algorithm as the static reference information acquisition step is used. In each frame of dynamic image information, each pipe segment region and clamping point are identified. The clamping point is the center point of the clamping region, ensuring a one-to-one correspondence between the dynamically identified pipe segments and the statically identified pipe segments. Then, for each pair of adjacent pipe segments, the pipe segment spacing in each frame of dynamic image is measured. The pipe segment spacing is specifically the vertical distance between the center lines of adjacent pipe segments. The pipe segment spacing data of consecutive frames are arranged in chronological order to obtain the pipe segment spacing variation curve. Simultaneously, the coordinates of each clamping point in each frame of dynamic image are measured. A two-dimensional coordinate system is established with the coordinates of the clamping point in the static reference image as the origin. The x-axis is parallel to the train length direction, and the y-axis is perpendicular to the train length direction. The offset of the clamping point relative to the origin in each frame is calculated. The offset data of consecutive frames are arranged in chronological order to obtain the position offset curve of each clamping point. The horizontal axis of this curve is time t. Finally, the vibration amplitude and vibration frequency of the pipe segment are calculated based on the pipe segment spacing variation curve. ,in This represents the maximum value of the pipe segment spacing. The minimum value of the pipe segment spacing is given. The unit of the pipe segment vibration frequency f is Hz. It is calculated by performing a Fourier transform on the pipe segment spacing variation curve and extracting the frequency component with the highest power spectral density. This frequency component is the main frequency generated by the pipe segment due to vibration.

[0056] The trend judgment step is used to predict the loosening risk index of each pipe section based on basic information, train operation data, and dynamic information. The loosening risk index is a quantitative indicator reflecting the probability of the corresponding pipe section becoming loose in the future, including a loosening trend score. The loosening trend score is a value calculated by comprehensively considering basic information, dynamic information, and train operation data. The higher the score, the greater the probability of the pipe section becoming loose in the future. In this embodiment, considering that the loosening risk is affected by multiple factors, the loosening trend score includes a basic score item, a dynamic score item, and an environmental amplification item. The basic score item reflects the degree of influence of basic information on the loosening trend score, the dynamic score item reflects the degree of influence of dynamic information on the loosening trend score, and the environmental amplification item reflects the degree of influence of train operation data on the loosening trend score. At the same time, to improve the accuracy of the loosening trend score, the calculation of the loosening trend score is configured with a time decay correction strategy. The time decay correction strategy corrects the loosening trend score by introducing a time decay coefficient to reflect the aging effect caused by the cumulative running time of the pipe section and the maintenance effect decay effect caused by the length of time after maintenance.

[0057] Before calculating the loosening trend score, considering the large number of pipe segments, calculating the loosening risk score for each segment would result in an excessively large amount of data. Therefore, a risk pre-screening strategy was implemented to improve detection efficiency. This strategy first identifies potentially high-risk pipe segments and calculates the loosening trend score only for these high-risk segments. The risk pre-screening strategy includes individual judgment sub-strategies and group comparison sub-strategies.

[0058] The individual assessment sub-strategy includes calculating a preliminary risk score for the pipe segment based on clamping length, pipe segment length, pipe segment spacing, and pipe vibration amplitude. The preliminary risk score uses a weighted calculation method, and the calculation formula is as follows:

[0059] ,

[0060] in This indicates a preliminary risk score. The weighting coefficient represents the clamping length. This represents the normalized value of the clamping length. The weighting coefficient representing the pipe segment length. This represents the normalized value of the pipe segment length. The weighting coefficient represents the spacing between pipe sections. This represents the normalized value of the pipe segment spacing. The weighting coefficient representing the vibration amplitude of the pipe section. This represents the normalized value of the pipe segment vibration amplitude. Normalized values ​​are used here to eliminate the influence of inconsistent dimensions among parameters. Specifically, the normalized value of the clamping length is the ratio of the pipe segment's clamping length to the preset maximum allowable clamping length. Since a longer clamping length reduces the risk of pipe segment loosening, a value of (1- ) is used. The clamping length contributes to the initial risk score. The normalized pipe segment length is specifically the ratio of the pipe segment length to the standard length of a pipeline of the same specification. The normalized pipe segment spacing is calculated using the minimum allowable safety distance as the high-risk threshold and the maximum safety distance as the low-risk threshold, transforming the actual average pipe segment spacing into a value between 0 and 1 through linear mapping. The normalized pipe vibration amplitude is the ratio of the pipe segment's vibration amplitude to the maximum allowable vibration amplitude for that type of pipeline. Weighting coefficients are set according to the degree of influence of each parameter on the loosening risk; if the vibration amplitude has the greatest impact on the loosening risk, then the weighting coefficient is the largest.

[0061] Finally, the initial risk score is compared with the initial score threshold. The initial score threshold is set based on operation and maintenance experience, such as 60 points. If the initial risk score exceeds the preset initial score threshold, the pipe section is defined as a high-risk pipe section.

[0062] The specific sub-strategy of group comparison includes obtaining the preliminary risk scores of multiple pipe segments arranged side-by-side under the same vehicle undercarriage and using them as a group comparison group. These pipe segments are typically in the same operating environment and vibration conditions, and their loosening risks should be consistent, thus allowing for horizontal comparison. If the preliminary risk score of a certain pipe segment differs significantly from other pipe segments, that segment may have an anomaly. First, the regional mean μ and regional variance σ of the preliminary risk scores of the pipe segments within this group comparison group are calculated. 2 Regional variance reflects the dispersion of the initial risk scores for each pipe segment. The variance threshold is set based on operational experience, such as 25. If the regional variance exceeds the preset threshold, a differential pipe segment is identified. A differential pipe segment is defined as a segment where the absolute value of the difference between the initial risk score and the regional mean exceeds twice the preset standard deviation; these segments are then defined as high-risk segments. For high-risk segments selected through individual judgment sub-strategies or group comparison sub-strategies, a loosening trend score is calculated based on a preset early warning score calculation strategy. In this application, both the group comparison sub-strategy and the individual judgment sub-strategy effectively reduce computational load, focusing the core calculations on high-risk pipe segments.

[0063] The early warning score calculation strategy includes first calculating the basic score items. These basic score items are calculated based on the clamping length, corrected pipe segment curvature distribution data, pipe segment width distribution data, and flow medium density from the basic information. The calculation formula for the basic score items is as follows:

[0064] ,

[0065] in Where D is the clamping length; D is the outer diameter of the pipe section; and k is the wall thickness correction factor. The density of the medium flowing inside the pipeline. The value ranges from 0 to 1, and the larger the value, the higher the risk of loosening due to the basic information.

[0066] This represents the basic scoring item, with a value range of 0-1. The larger the value, the higher the risk of loosening due to the basic information. , , , The weighting coefficients of each basic information parameter represent the degree of influence of the parameter on the basic risk, and the sum of the four factors equals the sum of the four factors. This represents the normalized value of the clamping length, ranging from 0 to 1, reflecting the clamping constraint capability. This represents the normalized value of the maximum curvature of the pipe segment after correction, ranging from 0 to 1. Specifically, it is the ratio of the maximum curvature of the pipe segment after correction to the maximum curvature allowed by the pipe material, reflecting the degree of deformation of the pipe segment. Indicates the coefficient of variation of pipe section width. = (maximum width - minimum width) / average width, reflecting the uniformity of pipe segment shape; This represents the normalized density value of the circulating medium, ranging from 0 to 1. Specifically, it is the ratio of the actual medium density to the preset maximum fluid medium density, reflecting the magnitude of the dynamic load on the pipeline.

[0067] Then, the dynamic scoring item is calculated. The dynamic scoring item is calculated based on the vibration amplitude, vibration frequency, clamping point offset, and pipe segment spacing variation coefficient in the dynamic information. The calculation formula for the dynamic scoring item is:

[0068] ,

[0069] represents the dynamic scoring item, with a value ranging from 0 to 1. The larger the value, the higher the risk of loosening due to dynamic information; e represents the natural constant, which reflects the cumulative amplification effect of dynamic risk through an exponential function. , , , This represents the weighting coefficient of each dynamic parameter, reflecting the priority of the parameter's impact on dynamic risk; This represents the normalized value of the vibration amplitude of the pipe section, ranging from 0 to 1, reflecting the severity of the vibration. This represents the normalized value of the pipe section vibration frequency, ranging from 0 to 1, reflecting the fatigue damage rate. Specifically, it is the ratio of the actual vibration frequency to the preset pipe resonance frequency. This represents the normalized value of the maximum offset of the clamping point, ranging from 0 to 1. Specifically, it is the ratio of the actual total offset to the maximum allowable offset, directly reflecting the loosening of the fastener. This represents the coefficient of variation of the pipe segment spacing. = (maximum pipe segment spacing - minimum pipe segment spacing) / average pipe segment spacing, reflecting the risk of collision / pull between adjacent pipe segments.

[0070] Next, the environmental amplification term is calculated. This term is based on track smoothness and train speed data from the train operation data. The formula for calculating the environmental amplification term is:

[0071] ,

[0072] in Indicates environmental amplification. This is a normalized value for track smoothness, specifically the ratio of the actual track irregularity amplitude to the preset maximum allowable irregularity amplitude. This is the normalized value of the train speed, specifically the ratio of the actual operating speed to the preset maximum permissible operating speed. , These are weighting coefficients, set according to the degree of impact of environmental factors on risk. =0.6, =0.4, the amplification effect of track smoothness on vibration is more significant, therefore Larger; The value range is greater than 1, and the larger the value, the stronger the amplification effect of environmental factors on the risk of loosening.

[0073] ,

[0074] Finally, a time decay coefficient is introduced. The product of the above three items is adjusted to obtain the loosening trend score. Time decay coefficient This is a coefficient based on the cumulative operating time of the pipe section and the time after maintenance. It is used to reflect the impact of pipe aging and the decline in maintenance effectiveness on the risk of loosening. The formula for calculating the time decay coefficient is:

[0075] ,

[0076] Where T represents the cumulative running time of the pipe section. The time is reset after the pipe section is reinstalled, reflecting the aging degree of the pipe section. The maintenance intervals for this type of piping are set according to industry standards, such as for brake piping. =5000 hours; t is the duration after maintenance of the pipe section, representing the time interval from the completion of the last maintenance to the current inspection; , The contribution coefficient is set according to the degree of influence of aging effect and maintenance effect degradation, such as =0.2, =0.1, the cumulative runtime has a more significant impact on aging, therefore Larger; The value is greater than 1, and increases with the cumulative running time. The longer the cumulative running time, the more severe the aging of the pipeline. The larger the value, the greater it becomes, and the longer the maintenance period, the more significant the decrease in maintenance effectiveness. The larger the value, the greater the loosening trend score. The final calculation formula is as follows: =S 基 ×S 动 ×S 环 ×K t ×100, the calculation result is limited to 0-100 points. If the calculation result exceeds 100 points, it is taken as 100 points.

[0077] The time decay correction strategy also requires recording the cumulative runtime and historical maintenance records of the pipeline section. Historical maintenance records include the time points for fastener replacement and pipeline adjustments, which are automatically obtained through the train operation and maintenance management system or manually entered. After correcting the loosening trend score using the time decay coefficient, the corrected score is used to adjust the loosening score threshold in the early warning conditions. The loosening score threshold is negatively correlated with the time decay coefficient, i.e. The larger the value, the lower the loosening score threshold. The specific adjustment method is calculated using the formula: Threshold Correction Value = Initial Threshold Value / The initial threshold value is the basic warning threshold for this type of pipeline, set according to industry standards, such as 80 points. When the value is 1.36, the threshold correction value is approximately 80 / 1.36 ≈ 58.8 points, to reflect that aging pipe sections require stricter early warning standards.

[0078] The loosening warning step is used to determine whether to generate a warning signal based on loosening risk indicators. It is configured with preset loosening warning conditions, which include the loosening trend score exceeding a preset loosening score threshold within a preset future time period, or the rate of change of the loosening trend score exceeding a preset loosening trend change rate threshold. The preset future time period is set according to the operation and maintenance response capability, such as 72 hours, which means judging whether the pipe section may loosen within the next 3 days. The loosening trend score within the future time period can be obtained by linearly fitting the current and the past 3 loosening trend scores. For example, if the past 3 scores are 70, 85 and 100 points respectively, with a time interval of 24 hours, and the fitting slope is 15 points / 24 hours, the predicted score for the next 72 hours is 100 + 15 × 3 = 145 points, which is taken as 100 points.

[0079] The formula for calculating the rate of change of loosening trend is: Rate of change = (Current score - Historical score) / Historical score × (24 / Time interval).

[0080] The time interval is the time difference between the current detection and the historical detection, in hours. The change rate is in % / day. The preset loosening trend change rate threshold is set based on operation and maintenance experience, such as 5% / day. That is, if the score increases by more than 5% every day, it is judged as a rapid increase in risk.

[0081] The offset detection method also features a graded early warning output strategy. Based on the numerical range of the loosening trend score, early warning signals are divided into Level 1, Level 2, and Level 3 warning signals. A Level 1 warning signal indicates that intervention and maintenance are required, corresponding to a loosening trend score ≥ 80. At this point, the risk of pipe section loosening is extremely high, and maintenance personnel must be arranged within 24 hours for inspection and maintenance, such as retightening fasteners and replacing worn clamping components. A Level 2 warning signal indicates that the loosening standard has not been met but continuous monitoring is required, corresponding to a loosening trend score 60 ≤ score < 80. At this point, the pipe section has a potential risk of loosening and requires further monitoring. Shorter inspection intervals, such as reducing from once every 7 days to once every 3 days, continuously track score changes; Level 3 warning signals indicate that the loosening standard has not been met and continuous monitoring is not required, corresponding to a loosening trend score of <60 points. At this time, the risk of pipe section loosening is low, and inspection can be carried out according to the normal inspection cycle, such as once every 7 days; After the warning signal is generated, the warning signal and the corresponding pipe section information, including pipe section location, static reference image, dynamic image clips, and scoring calculation process, are pushed to the train operation and maintenance terminal. At the same time, the historical maintenance records of the pipe section associated with the Level 1 warning signal are provided so that operation and maintenance personnel can quickly understand the pipe section status and formulate maintenance plans.

[0082] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principle of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for detecting the offset of parallel pipelines under a train, characterized in that, The method comprises: a static reference information acquisition step, which acquires a pipe image of a train bottom in a static state of the train to define static reference image information, acquires basic information of the pipe in the static reference image information, and the basic information comprises specifications of the pipe, clamping mode data, clamping length, and flow medium, a dynamic information acquisition step, which acquires a pipe image of the train bottom in a running process of the train to define dynamic image information, synchronously acquires train running data, acquires dynamic information in the dynamic image information, and the dynamic information reflects a degree of deviation of the pipe, a trend judgment step, which predicts a loosening risk index of each pipe section based on the basic information, the train running data, and the dynamic information, and the loosening risk index reflects a possibility of loosening of the corresponding pipe section in future time, a loosening early warning step, which is configured with a preset loosening early warning condition, and generates a loosening early warning signal if the loosening early warning condition is met. The static reference information acquisition step is configured with a pipe section identification strategy, which comprises: identifying a pipe region and a clamping region in the static reference image information, the pipe region is specifically a region of the pipe in the static reference image information, and the clamping region is specifically a region in which a pipe fastener covers the pipe, and the clamping region divides the pipe region into at least two pipe section regions, identifying a long-side boundary of the pipe section region and a pipe section center line, the long-side boundary represents two boundaries of the pipe section region in a pipe section length direction, and a distance between any point on the pipe section center line and the two long-side boundaries is equal, acquiring a length of the pipe section center line to acquire a pipe section length, and acquiring a length of the clamping region covering the pipe to acquire the clamping length, acquiring pipe section curvature distribution data and pipe section width distribution data as the clamping mode data, the pipe section curvature distribution data represents a distribution of a curvature of the pipe section center line in the length direction, and the pipe section width distribution data is specifically a distribution of a width of the pipe section region in the pipe length direction.

2. The method of claim 1, wherein, The dynamic information acquisition step is configured with a dynamic information extraction strategy, which comprises: acquiring continuous dynamic image information, and identifying each pipe section in each dynamic image information, acquiring a change curve of a pipe section spacing and a position deviation curve of each clamping point, the pipe section spacing is specifically a distance between adjacent pipe sections, the clamping point represents a center point of the clamping region, and the position deviation curve reflects a change of a position deviation of the clamping point in the continuous dynamic image information, calculating a pipe section vibration amplitude and a vibration frequency based on the change curve of the pipe section spacing.

3. The method of claim 2, wherein, The loosening risk index comprises a loosening trend score, and the loosening early warning condition comprises that the loosening trend score exceeds a preset loosening score threshold or a change rate of the loosening trend score exceeds a preset loosening trend change rate threshold within a preset future time length.

4. The method of offset detection of side-by-side pipes of a train consist according to claim 3, characterized in that, The loosening trend score comprises a basic score item, a dynamic score item, and an environmental amplification item, the basic score item reflects an influence degree of the basic data on the loosening trend score, the dynamic score item reflects an influence degree of the dynamic information on the loosening trend score, and the environmental amplification item reflects an influence degree of the train running data on the loosening trend score.

5. The method of claim 3, wherein, The offset detection method configures a risk pre-screening strategy, comprising: calculating a preliminary risk score of the pipe segment based on the clamping length, the pipe segment length, the pipe segment spacing, and the pipe vibration amplitude, judging the size of the preliminary risk score and the preliminary score threshold value, if the preliminary risk score exceeds the preset preliminary score threshold value, defining the pipe segment as a high-risk pipe segment and calculating the loosening trend score based on the preset early warning score calculation strategy.

6. The method of offset detection of side-by-side pipes of a train consist of claim 5, wherein, The risk pre-screening strategy further comprises a group comparison sub-strategy, comprising: obtaining the preliminary risk scores of multiple pipe segments arranged side by side on the same car bottom as a group comparison group, calculating the regional mean and regional variance between each preliminary risk score in the group comparison group, if the regional variance exceeds the preset variance threshold value, judging that there is a difference pipe segment, and defining the difference pipe segment as a high-risk pipe segment and calculating the loosening trend score.

7. The method of claim 1, wherein, The pipe segment curvature distribution data is configured with a curvature correction strategy, comprising correcting the pipe segment curvature distribution data based on the elastic modulus of the pipe material and the type of the flow medium, and the corrected pipe segment curvature distribution data is used to optimize the influence weight of the clamping shape on the loosening trend in the basic score item.

8. The method of claim 4, wherein, The calculation of the loosening trend score is configured with a time decay correction strategy, comprising: recording the cumulative running time of the pipe segment and the historical maintenance record, the historical maintenance record including the time node of fastener replacement and pipe adjustment, based on the cumulative running time and the maintenance interval, setting a time decay coefficient, the time decay coefficient increases with the increase of the cumulative running time and the time length after maintenance; correcting the loosening trend score using the time decay coefficient, and the corrected score is used to adjust the loosening score threshold value in the early warning condition, the loosening score threshold value is negatively correlated with the time decay coefficient.

9. The method of claim 6, wherein, The offset detection method is configured with a pre-warning grading output strategy, comprising: dividing the early warning signal into a first-level early warning signal, a second-level early warning signal and a third-level early warning signal based on the numerical range of the loosening trend score, the first-level early warning signal indicating that intervention maintenance is needed, the second-level early warning signal indicating that the loosening standard has not been reached but needs to be continuously monitored, and the third-level early warning signal indicating that the loosening standard has not been reached and does not need to be continuously monitored.

Images