A method for evaluating the health status of bow-net matching

Abstract

The present invention discloses a method for assessing the health status of bow-net matching, and relates to the technical field of contact network detection. The present invention introduces a method for data conversion, which converts discrete data into ordered data by filtering outliers and assigning ranks, thereby providing guarantees for waveform alignment, key feature identification, and multi-dimensional data association analysis, and making the analysis results more accurate. The present invention introduces methods for data segmentation processing, calculating correlation coefficients, calculating Euclidean distance for rough positioning, and precise matching based on feature points during waveform alignment, which solves the problem that waveforms cannot be aligned in the context of long-distance railways. The present invention introduces methods for template matching and calculating similarity to identify key waveform features, which can solve the problems of false alarms in special sections and inaccurate defect detection that may exist in the line, and improve the accuracy and reliability of contact network detection.

Description

Technical Field

[0001] The present invention relates to the field of contact network detection technology, and more specifically to a method for evaluating the health status of a catenary matching based on an artificial intelligence algorithm. Background Art

[0002] The pantograph-catenary integrated detection device is an important component of the traction power supply safety detection and monitoring system. We call it the 1C detection device, which is installed on the comprehensive detection train. As the comprehensive detection train runs on the line, the 1C detection device measures the current parameters between the contact network and the pantograph and the geometric parameters of the contact network, and outputs the detection waveform curve graph at the same time, and then automatically statistically analyzes the detection data. The detection results are used to guide the "condition repair" and "precision measurement and repair" of the contact network. After using the 1C detection device to dynamically detect and evaluate the contact network, the detection data is used as the basic data, combined with the contact network operation quality index CQI, section quality evaluation, contact network dynamic performance evaluation CDI and detection data correlation analysis, to evaluate the contact network operation status and provide guidance for contact network maintenance.

[0003] my country's contact network detection technology has moved from the independent exploration stage to the independent innovation research stage. However, while the contact network detection technology is developing rapidly, there are also some problems, such as: there are deviations in the positioning of the detection vehicle, errors between some defect kilometer marks and on-site kilometer marks, and the defect location cannot be quickly found when the defect is reviewed, resulting in low maintenance efficiency; the accuracy of the detection data needs to be further improved, and there are cases where the detection vehicle detects non-working branch data when passing through special sections such as anchor joints, phase splitters, and line switches, causing false alarms; in addition, a large amount of bow net data detected by the detection vehicle in different cycles has not been fully used. How to improve the accuracy of contact network detection technology, make full use of data, and accurately identify and locate defect types is an important problem in contact network detection technology. Summary of the invention

[0004] In order to overcome the defects existing in the above-mentioned prior art, the present invention discloses a method for assessing the health status of bow-net matching. The present invention conducts a comprehensive analysis of different types of waveform graphs from aspects such as pull-out value defects, height defects, height difference defects, bow-net contact pressure defects, contact line hard points, arcing defects, and historical detection data comparison, change trends, and repeatability comparison, and analyzes the potential risks of the contact network in combination with positioning data information, providing a basis for judging contact network defects and achieving the best contact network detection effect. The waveform graph analysis technology can lock the defect at a certain point or within a certain span, effectively enhancing the targeting of defect review and greatly improving the efficiency of on-site defect rectification. The present invention has the advantages of comprehensive analysis, accurate positioning, and high detection accuracy.

[0005] In order to achieve the above objectives, the technical solution adopted by the present invention is:

[0006] A method for evaluating the health status of a bow-net matching comprises the following steps:

[0007] 1. Data Conversion

[0008] S1, obtain the discrete data of the bow net detection and filter out the abnormal values, and then assign ranks to the discrete data to transform them into ordered data;

[0009] Preferably, step S1 comprises the following steps:

[0010] S11, collecting discrete data of bow-net detection;

[0011] In step S11, the discrete data is ,in, For each discrete data.

[0012] S12. Calculate each point in discrete data The abnormal coefficient , and the anomaly coefficient With abnormal threshold Compare, if Points outliers and filter them out;

[0013] Preferably, in step S12, the abnormal coefficient for:

[0014]

[0015] in, is the abnormal coefficient, is a discrete data point, is the sample mean, is the sample standard deviation.

[0016] S13, after filtering out the outliers, assign ranks to the discrete data according to the kilometer mark. , whose rank is , and output an ordered dataset ,in is the point and rank number.

[0017] 2. Waveform Alignment Algorithm for Long-Distance Railway Scenarios

[0018] S2, segmenting the ordered data and removing outliers, and then aligning the segmented data to be matched with the standard data waveform;

[0019] Preferably, step S2 comprises the following steps:

[0020] S21, input standard data and data to be matched, wherein the data to be matched is the ordered data obtained in step S1;

[0021] Preferably, in step S21, the standard data is recorded as , the data to be matched is recorded as ,in and are the number of data points of standard data and data to be matched respectively.

[0022] S22, segmenting the standard data and the data to be matched;

[0023] Preferably, step S22 comprises:

[0024]

[0025] in, Represents the segmented data, Indicates The first data points, Represents the number of data points per paragraph.

[0026] S23, screening outliers in the segmented data to be matched and removing the outliers;

[0027] In the present invention, both step S12 and step S23 remove outliers, but the difference is that the range of outlier screening is different. In step S12, outlier screening is performed on the entire discrete data. In step S23, outlier screening is performed on each segment of data after segmentation, ensuring that there are no outliers in the data finally used for waveform alignment.

[0028] Preferably, step S23 includes: setting a data point threshold for the data to be matched, and if a data point in the data to be matched exceeds the data point threshold, filtering out the data point.

[0029] S24, calculating the correlation coefficient between the data to be matched and the standard data after the outliers are removed;

[0030] Preferably, step S24 comprises: for the first The correlation coefficient of the segment data is:

[0031]

[0032] in, and Respectively The standard data in the segment and the data to be matched data points, and Respectively The mean of the standard data and the data to be matched in the segment, For paragraphs, for The correlation coefficient of the segment, For the data points, is the total number of data points.

[0033] S25, using the correlation coefficient to calculate the Euclidean distance and perform rough positioning;

[0034] Preferably, step S25 includes: selecting several segments of standard data and data to be matched whose correlation coefficients exceed a correlation coefficient threshold, and calculating the Euclidean distance between the selected several segments of standard data and the data to be matched, and using the Euclidean distance to roughly locate the standard data and the data to be matched. During the rough positioning, the standard data and the data segment to be matched with the smallest Euclidean distance among the several segments of standard data and the data to be matched are matched as the same part.

[0035] Preferably, in step S25, for the correlation coefficient Segments exceeding the correlation coefficient threshold , calculate its Euclidean distance :

[0036]

[0037] Find the Euclidean distance The smallest segment , as the same position of the standard data and the data to be matched, the rough positioning is completed.

[0038] S26, based on the characteristic inflection point search method, search for the characteristic extreme value points of the standard data and the extreme value points of the data to be matched after rough positioning;

[0039] S27, performing a second scaling correction on the data to be matched after the rough positioning according to the relationship between the extreme value points of the data to be matched after the rough positioning and the extreme value points of the characteristic of the standard data, so as to complete the precise alignment;

[0040] S28. Output the detection data after precise alignment.

[0041] 3. Accurate identification of key waveform features

[0042] S3, performing template matching on the waveform alignment data and the waveform template to obtain key features of the waveform;

[0043] Preferably, step S3 comprises the following steps:

[0044] S31, analyzing and extracting the waveform data to obtain feature point information and constructing a waveform template of a special section or defect using the feature point information;

[0045] S32, performing template matching on the waveform alignment data and the waveform template to obtain a special section waveform or a defect waveform;

[0046] Preferably, the template matching in step S32 includes:

[0047]

[0048] in, Indicates the matching degree between the current waveform data and the special section or defect waveform template. Indicates the first feature points, Indicates the first feature points, Indicates The similarity of feature points, represents the weight of the corresponding feature point, Indicates the number of feature points.

[0049] S33. Mark the special section waveform or defect waveform.

[0050] Preferably, step S3 includes determining whether the suspected positioner is not under force, including the following steps:

[0051] A1. Window size , slide on the pull-out value waveform to calculate the slope of each positioning point After calculating the slopes of all positioning points, update the slope threshold ;

[0052] Preferably, the slope of step A1 is for:

[0053]

[0054] in, To pull out the value, For point, is the total number of waveform points, For the feature points.

[0055] A2. Compare the slope of the current positioning point with the slope threshold. If the slope of the current positioning point is , then perform template matching calculation on the positioning point waveform to obtain the matching degree ;

[0056] Preferably, the template matching calculation in step A2 includes:

[0057]

[0058] in, Indicates the matching degree between the current waveform data and the waveform template. Indicates the first feature points, Indicates the first feature points, Indicates The similarity of feature points, represents the weight of the corresponding feature point, Indicates the number of feature points.

[0059] A3. When the matching degree Greater than the matching threshold When , it is determined that the locator in this waveform is suspected to be under no force, and the waveform is bolded and marked on the waveform.

[0060] 4. Multidimensional Data Correlation Analysis

[0061] S4. Use waveform alignment data and waveform key feature data to conduct historical data correlation analysis and historical defect data comparison analysis respectively to evaluate the status and defects of the contact network.

[0062] Preferably, the historical data association analysis in step S4 includes the following steps:

[0063] S411, establishing a historical detection database, and storing the waveform alignment data of different parameters of the contact network in the historical detection database;

[0064] In step S411, the waveform alignment data of different parameters of the contact network include the contact wire pull-out value, height, and smoothness.

[0065] S412, calculating the correlation coefficients between different parameters of the contact network in the historical detection database;

[0066] Preferably, the correlation coefficient of step S412 includes:

[0067]

[0068] in, for and The correlation coefficient of the parameters, and For data at different time points, and for and The mean of is the number of data points.

[0069] S413. Based on the correlation coefficients between different parameters of the contact network, a historical data analysis report is generated to display the status of the contact network.

[0070] Preferably, the historical defect data comparative analysis in step S4 includes the following steps:

[0071] S421, establishing a historical defect database, and storing defect waveform data and positioning point information in the waveform key features in the historical defect database;

[0072] S422, comparing the currently detected defect waveforms with the defect information in the historical defect database to identify the commonalities and particularities of the defects;

[0073] S423. Generate a defect analysis report based on the defect comparison analysis results to describe the defect conditions and trends of each part of the contact network.

[0074] Preferably, the defect waveform data in step S421 includes pull-out value defect, height defect, height difference defect, bow-net contact pressure defect, contact line hard point, and arcing defect.

[0075] Beneficial effects of the present invention:

[0076] The present invention introduces a data conversion method, which converts discrete data into ordered data by filtering outliers and assigning ranks, thereby providing guarantees for waveform alignment, key feature identification, and multidimensional data association analysis, making the analysis results more accurate.

[0077] The present invention introduces a method of data segmentation processing, calculation of correlation coefficients, calculation of Euclidean distance for rough positioning, and precise matching based on feature points during waveform alignment, thereby solving the problem that waveforms cannot be aligned in the context of long-distance railways.

[0078] The present invention introduces a method of template matching and similarity calculation to identify key waveform features, which can solve the problems of false alarms in special sections and inaccurate defect detection that may exist in the line, and improve the accuracy and reliability of contact network detection.

[0079] The present invention introduces a multi-dimensional data association analysis method to comprehensively evaluate the state of the contact network, and comprehensively evaluates the state and defects of the contact network through analysis and comparison of multiple parameters.

[0080] The invention has low cost and can work continuously for 24 hours; it has high speed and stable operation; it has consistent detection standards and good stability; it has multi-dimensional data comparison and analysis, high precision and strong reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] Figure 1 The calculation process of the bow-net matching health status assessment method of the present invention;

[0082] Figure 2 This is a schematic diagram showing the completion of data conversion in the present invention;

[0083] Figure 3 This is a schematic diagram for comparing the input data of the present invention with the standard data;

[0084] Figure 4 This is a schematic diagram of the inflection point feature point search effect of the present invention;

[0085] Figure 5 It is a schematic diagram of the alignment result of the present invention;

[0086] Figure 6 It is a schematic diagram of the data annotation effect of the present invention. DETAILED DESCRIPTION

[0087] The concept, specific structure and technical effects of the present invention will be clearly and completely described below in conjunction with the embodiments and drawings to fully understand the purpose, characteristics and effects of the present invention.

[0088] Example 1

[0089] A method for evaluating the health status of bow-net matching, such as Figure 1 As shown, the following steps are included:

[0090] S1, obtain the discrete data of the bow net detection and filter out the abnormal values, and then assign ranks to the discrete data to transform them into ordered data;

[0091] S2, segmenting the ordered data and removing outliers, and then aligning the segmented data to be matched with the standard data waveform;

[0092] S3, performing template matching on the waveform alignment data and the waveform template to obtain key features of the waveform;

[0093] S4. Use waveform alignment data and waveform key feature data to conduct historical data correlation analysis and historical defect data comparison analysis respectively to evaluate the status and defects of the contact network.

[0094] Example 2

[0095] This embodiment further describes step S1 based on embodiment 1. Step S1 mainly performs data conversion, which is as follows:

[0096] Data conversion is to convert discrete data into ordered data and filter out outliers in the data to provide guarantee for subsequent waveform alignment and multi-dimensional data association analysis. Data conversion is divided into the following steps:

[0097] The first step is to collect the discrete data that needs to be converted. The data is represented as follows:

[0098]

[0099] The second step is to filter out the outliers in the data and set the data anomaly threshold as , used to identify and filter out outliers, and calculate the outlier coefficient for each point in the data , The calculation formula is as follows:

[0100]

[0101] in, represents the sample mean, represents the sample standard deviation, if ,but were identified as outliers and filtered out.

[0102] The third step is to assign ranks to the data after filtering out the outliers according to the kilometer mark. , whose rank is , and output the final result as an ordered data set, represented as follows:

[0103]

[0104] in is the point and rank number.

[0105] The final converted data is as follows Figure 2 shown.

[0106] Example 3

[0107] This embodiment further describes step S2 based on embodiment 2. Step S2 is a waveform alignment algorithm in a long-distance railway scenario, which is as follows:

[0108] The waveform alignment algorithm uses the line standard data as the alignment target, and aligns and matches the previous detection data with the standard template data to achieve the purpose of data verification and positioning correction. However, the amount of data for long-distance railways is too large, and there are many outliers in it, which often leads to the inability to align the waveforms. Therefore, a waveform alignment method for long-distance railway scenarios is proposed.

[0109] Waveform alignment is divided into the following steps: First, the input data is segmented to prevent the misalignment when the data is too large. The data is processed separately according to the divided segments. Then, the data to be matched is screened for outliers according to the established threshold to prevent outliers from causing the waveform to be misaligned. Then, the correlation coefficient between the test data and the standard data is calculated. A preliminary search is performed in the standard data based on the correlation coefficient. After the search is completed, the Euclidean distance between the standard data with a higher correlation coefficient and the data to be matched is calculated. Rough positioning is performed based on the Euclidean distance to preliminarily locate the test position in the standard data. Finally, based on the feature inflection point search method, the test data is accurately positioned and matched to achieve the effect of data alignment.

[0110] The first step is data input. The standard data is recorded as , the data to be matched is recorded as ,in and are the number of data points of standard data and data to be matched respectively.

[0111] like Figure 3 As shown, the two curves are standard data and detection data to be processed.

[0112] The second step is to process the data in segments. In the long-distance railway scenario, the data cannot be aligned due to the large amount of data. To solve this problem, the long-distance railway data needs to be processed in segments. The basic principle of data segmentation is to divide the long-distance railway data into several smaller segments, and then process each segment separately. This can effectively reduce the complexity of data processing, avoid processing too much data at one time, and better align the waveform. The formula is shown below.

[0113]

[0114] in, Represents the segmented data, Indicates The first data points, Represents the number of data points per paragraph.

[0115] The third step is to filter out outliers in the data to be matched. Outliers may interfere with the accuracy of waveform matching, so they need to be detected and processed before matching.

[0116] Set a data threshold. If a data point exceeds the threshold, filter out the data point to prevent it from affecting the waveform alignment.

[0117] The fourth step is to calculate the correlation coefficient between the standard data and the data to be matched.

[0118] The correlation coefficient is an indicator to measure the correlation between two sets of data. By calculating the correlation coefficients between different paragraphs, we can infer the correspondence between them.

[0119] For Segment data, calculate its correlation coefficient :

[0120]

[0121] in, and Respectively The standard data in the segment and the data to be matched data points, and Respectively The mean of the standard data and the data to be matched in the segment, For paragraphs, for The correlation coefficient of the segment, For the data points, is the total number of data points.

[0122] The fifth step is to calculate the Euclidean distance and perform rough positioning.

[0123] After calculating the correlation coefficient between the standard data and the data to be matched, select the standard data and the data to be matched with a high correlation coefficient to calculate the Euclidean distance between them, and roughly locate the standard data and the data to be matched according to the calculated Euclidean distance, that is, the standard data and the data to be matched with the smallest Euclidean distance can be matched as the same part.

[0124] For the correlation coefficient Higher segment , calculate its Euclidean distance :

[0125]

[0126] Find the The smallest segment , it is considered to be the same position and the rough positioning is completed.

[0127] The sixth step is precise positioning. After rough positioning, the new input data can be preliminarily aligned with the standard template. There will be some deviations in the details, so it is necessary to use the data's own characteristics for precise alignment. Based on the feature inflection point search method, find the extreme points of the standard data features and the extreme points of the data after rough positioning, such as Figure 4 shown.

[0128] The seventh step is precise matching. According to the relationship between the extreme points of the rough alignment data and the standard data, the rough alignment data is scaled and corrected for the second time to complete the precise alignment.

[0129] Step 8: Output the results, output the detection data after precise alignment, such as Figure 5 shown.

[0130] Example 4

[0131] Based on Example 3, this embodiment further elaborates on step S3, which is accurate identification of key waveform features, as follows:

[0132] The key feature accurate identification technology based on waveform is a technical method that realizes accurate identification of specific sections by analyzing the key feature points in the waveform data. This technology is proposed to solve the problem of false alarms that may occur when the inspection vehicle passes through special sections, and to judge the possible defects in the line by detecting the waveform, such as the positioner is not stressed, so as to improve the accuracy and reliability of contact network detection. This technology mainly includes the following key steps:

[0133] The first step is to construct a waveform template for a special section or defect. By analyzing and extracting the waveform data, the characteristic point information of the section can be obtained, and a waveform template can be constructed based on it. This waveform template will become the basis for subsequent waveform recognition to determine whether the waveform characteristic point belongs to a special section or a defect.

[0134] The second step is to achieve template matching of feature points. During the operation of the contact network inspection vehicle, the system will collect waveform data in real time and match it with the previously constructed special section waveform template. By comparing the similarity between the waveform data and the template, it can be determined whether the current waveform feature point matches the special section or defect. The calculation of template matching is shown in the formula:

[0135]

[0136] in, Indicates the matching degree between the current waveform data and the special section or defect waveform template. Indicates the first feature points, Indicates the first feature points, Indicates The similarity of feature points, represents the weight of the corresponding feature point, Indicates the number of feature points.

[0137] The third step is to mark and process the waveforms of special sections and defect waveforms. In the waveform software, the waveform data of special sections are marked and processed. By comparing and analyzing the waveform data of special sections with the normal sections, the detection accuracy can be further improved.

[0138] The following is the process of judging whether the positioner is suspected of not being stressed. There are several steps to judge whether the positioner is suspected of not being stressed.

[0139] The first step is to calculate the slope of the pulled value waveform in a sliding window manner, with the window size as , slide on the pull-out value waveform to calculate the slope of each positioning point. The calculation formula is as follows:

[0140]

[0141] in, To pull out the value, For point, is the total number of waveform points. After calculating the slopes of all positioning points, the slope threshold is updated adaptively. .

[0142] The second step is to compare the slope of the current positioning point with the slope threshold. , then further perform template matching calculation on the positioning point waveform, the formula is as follows:

[0143]

[0144] in, Indicates the matching degree between the current waveform data and the waveform template. Indicates the first feature points, Indicates the first feature points, Indicates The similarity of feature points, represents the weight of the corresponding feature point, Indicates the number of feature points.

[0145] When similarity Greater than the similarity threshold When the positioner is suspected to be under stress, the waveform segment is judged to have a positioner suspected to be under stress. In this case, the waveform segment is bolded and marked with "Positioner suspected to be under stress". Figure 6 shown.

[0146] Example 5

[0147] Based on Example 4, this example further describes step S4, which is multidimensional data association analysis, as follows:

[0148] The core goal of multidimensional data association analysis technology is to provide important support and guidance for the maintenance of the contact network by comprehensively evaluating the state of the contact network and improving the detection accuracy. Multidimensional data association analysis technology mainly includes two aspects: establishing historical data association analysis technology and historical defect data comparison analysis technology. Through the analysis and comparison of multiple parameters, the state and defects of the contact network are comprehensively evaluated.

[0149] a. Historical data association analysis technology

[0150] The first step is to establish a historical detection database.

[0151] In the long-distance railway scenario, data collection and database construction are required for the detection parameters of the contact network, such as the contact wire pull-out value, height, and smoothness. Through tools such as waveform analysis equipment, the data of the contact network is monitored and recorded in real time and stored in the historical detection database. These data can include the state parameters of the contact network in different time periods, forming a multi-dimensional historical data set.

[0152] The second step is to calculate the correlation coefficient of historical data.

[0153] Through historical data correlation analysis technology, the state of the contact network can be compared and analyzed. Using correlation analysis, by calculating the correlation coefficient between different parameters, the correlation relationship between them can be revealed. In this way, the change trend of the contact network can be deeply understood, such as whether the change of a certain parameter will affect other parameters. The calculation of the correlation coefficient is shown in the following formula.

[0154]

[0155] in, for and The correlation coefficient of the parameters, and For data at different time points, and for and The mean of is the number of data points.

[0156] The third step is to generate historical data analysis reports.

[0157] Based on the results of historical data correlation analysis, corresponding analysis reports are generated to display the status of the contact network.

[0158] b. Historical defect comparison analysis technology

[0159] The first step is to establish a historical defect database.

[0160] Use waveform analysis software to obtain the defect data and positioning point information of the locator such as suspected lack of force and arcing, and establish a historical defect database. By recording the defect data in different time periods, we can fully understand the defect situation of the contact network.

[0161] The second step is to conduct comparative analysis with the historical defect database.

[0162] Through the historical defect data comparison and analysis technology, the defects of the contact network are compared and analyzed. According to the defect information in the historical database, the currently detected defects are compared to identify the commonalities and particularities of the defects, thereby improving the detection accuracy.

[0163] The third step is to generate a defect analysis report.

[0164] Based on the defect comparison and analysis results, a defect report is generated, which describes in detail the defect conditions and trends of each part of the contact network. The report includes information such as defect type, location, size, etc., providing guidance for subsequent maintenance and improvement.

[0165] The above is a specific description of the implementation mode of the present invention, but the present invention is not limited to the described embodiments. Those skilled in the art may make various equivalent modifications or substitutions without violating the spirit of the present invention, and these equivalents or substitutions are all included in the scope defined by the claims of the present invention.

Claims

1. A method for evaluating the health status of a bow-net matching, characterized in that: The following steps are involved: S1, obtain the discrete data of the bow net detection and filter out the abnormal values, and then assign ranks to the discrete data to transform them into ordered data; S2, segmenting the ordered data and removing outliers, and then aligning the segmented data to be matched with the standard data waveform; S3, performing template matching on the waveform alignment data and the waveform template to obtain key features of the waveform; S4. Use waveform alignment data and waveform key feature data to conduct historical data association analysis and historical defect data comparison analysis to evaluate the state and defect conditions of the contact network; in: In step S1, the step of assigning ranks to discrete data and converting them into ordered data includes: assigning ranks to the discrete data after outliers are removed according to kilometer marks, and for each point , whose rank is , and output an ordered dataset ,in is the point and rank number; The S3 step includes the following steps: S31, analyzing and extracting the waveform data to obtain feature point information and constructing a waveform template of a special section or defect using the feature point information; S32, performing template matching on the waveform alignment data and the waveform template to obtain a special section waveform or a defect waveform, wherein the template matching includes: in, Indicates the matching degree between the current waveform data and the special section or defect waveform template. Indicates the first feature points, Indicates the first feature points, Indicates The similarity of feature points, represents the weight of the corresponding feature point, Indicates the number of feature points; S33, marking special section waveforms or defect waveforms; Step S3 includes determining whether the positioner is suspected of not being stressed, including the following steps: A1. Window size , slide on the pull-out value waveform to calculate the slope of each positioning point After calculating the slopes of all positioning points, update the slope threshold ; A2. Compare the slope of the current positioning point with the slope threshold. If the slope of the current positioning point is , then perform template matching calculation on the positioning point waveform to obtain the matching degree ; A3. When the matching degree Greater than the matching threshold When , it is determined that the locator in this waveform is suspected to be under no force, and the waveform is bolded and marked on the waveform.

2. A method for evaluating the health status of a bow-net matching according to claim 1, characterized in that: The S1 step includes the following steps: S11, collecting discrete data of bow-net detection; S12. Calculate each point in discrete data The abnormal coefficient , and the anomaly coefficient With abnormal threshold Compare, if Points outliers and filter them out; S13, after filtering out the outliers, assign ranks to the discrete data according to the kilometer mark. , whose rank is , and output an ordered dataset ,in is the point and rank number.

3. A method for evaluating the health status of a bow-net matching according to claim 1, characterized in that: The S2 step includes the following steps: S21, input standard data and data to be matched, wherein the data to be matched is the ordered data obtained in step S1; S22, segmenting the standard data and the data to be matched; S23, screening outliers in the segmented data to be matched and removing the outliers; S24, calculating the correlation coefficient between the data to be matched and the standard data after the outliers are removed; S25, using the correlation coefficient to calculate the Euclidean distance and perform rough positioning; S26, based on the characteristic inflection point search method, search for the characteristic extreme value points of the standard data and the extreme value points of the data to be matched after rough positioning; S27, performing a second scaling correction on the data to be matched after the rough positioning according to the relationship between the extreme value points of the data to be matched after the rough positioning and the extreme value points of the characteristic of the standard data, so as to complete the precise alignment; S28. Output the detection data after precise alignment.

4. A method for evaluating the health status of a bow-net matching according to claim 3, characterized in that: In step S21, the standard data is recorded as , the data to be matched is recorded as ,in and are the number of data points of standard data and data to be matched respectively; The S22 step segmentation includes: in, Represents the segmented data, Indicates The first data points, Indicates the number of data points per paragraph; Step S23 includes: setting a data point threshold of the data to be matched, and if a data point in the data to be matched exceeds the data point threshold, filtering out the data point.

5. A method for evaluating the health status of a bow-net matching according to claim 3, characterized in that: Step S24 includes: The correlation coefficient of the segment data is: in, and Respectively The standard data in the segment and the data to be matched data points, and Respectively The mean of the standard data and the data to be matched in the segment, For paragraphs, for The correlation coefficient of the segment, For the data points, is the total number of data points; Step S25 includes: selecting several segments of standard data and data to be matched whose correlation coefficients exceed a correlation coefficient threshold, calculating the Euclidean distance between the selected several segments of standard data and the data to be matched, and using the Euclidean distance to roughly locate the standard data and the data to be matched. During the rough location, the standard data and the data to be matched have the smallest Euclidean distance among the several segments of standard data and the data to be matched, and are matched as the same part; In step S25, for the correlation coefficient Segments exceeding the correlation coefficient threshold , calculate its Euclidean distance : Find the Euclidean distance The smallest segment , as the same position of the standard data and the data to be matched, the rough positioning is completed.

6. A method for evaluating the health status of a bow-net matching according to claim 1, characterized in that: The historical data association analysis in step S4 includes the following steps: S411, establishing a historical detection database, and storing the waveform alignment data of different parameters of the contact network in the historical detection database; S412, calculating the correlation coefficients between different parameters of the contact network in the historical detection database; S413. Based on the correlation coefficients between different parameters of the contact network, a historical data analysis report is generated to display the status of the contact network.

7. A method for evaluating the health status of a bow-net matching according to claim 1, characterized in that: The comparative analysis of historical defect data in step S4 includes the following steps: S421, establishing a historical defect database, and storing defect waveform data and positioning point information in the waveform key features in the historical defect database; S422, comparing the currently detected defect waveforms with the defect information in the historical defect database to identify the commonalities and particularities of the defects; S423. Generate a defect analysis report based on the defect comparison analysis results to describe the defect conditions and trends of each part of the contact network.

8. A method for evaluating the health status of a bow-net matching according to claim 6, characterized in that: The correlation coefficients of step S412 include: in, for and The correlation coefficient of the parameters, and For data at different time points, and for and The mean of is the number of data points.

Images