A turnout operation state rapid detection method based on traction force envelope

Abstract

The present application belongs to the field of intelligent operation and maintenance of rail transit, and discloses a method for quickly detecting the operation state of a turnout based on a traction envelope. The method adopted by the present application is as follows: if the traction curve to be detected is within the upper and lower envelopes of the normal reference curve, the detection result is normal; if the traction curve to be detected exceeds the envelope, the detection result is the matching curve information obtained by using the DTW algorithm, and for the case where the matching result is normal, the upper and lower envelopes of the normal reference curve are further updated and written into the curve sample library. According to the method adopted by the present application, it can be quickly determined whether the traction curve is normal, thereby avoiding a large number of invalid DTW distance calculations and improving the detection efficiency.

Description

A rapid detection method for turnout operating status based on traction envelope Technical Field

[0001] This invention mainly relates to the field of intelligent operation and maintenance technology for rail transit, and in particular to a method for rapid detection of turnout operating status based on traction envelope. Background Technology

[0002] Turnouts are one of the fundamental pieces of equipment in railway signaling systems, directly affecting train operation efficiency and safety. Utilizing collected data to monitor the operational status of turnouts has significant practical value. Currently, the main sources of data for turnout switching machines include gap images, operating current, power, and traction force. Compared to other monitoring data, the traction force curve directly reflects the resistance and mechanical performance of the turnout during operation.

[0003] Dynamic Time Warping (DTW) is an algorithm based on dynamic programming that non-linearly adjusts two time series in the time domain to accurately calculate their similarity. It does not require strict alignment of the lengths of each series, allowing stretching or compression along the time axis. The traditional steps for detecting turnout status using the DTW algorithm are: calculating the DTW distance between the curve to be detected and each sample curve in the sample curve library; the detection result is the state information carried by the sample curve corresponding to the smallest DTW distance. However, the DTW algorithm has high time complexity, and in reality, turnouts are in normal operation most of the time, leading to the calculation of a large number of invalid results and long processing times. Summary of the Invention

[0004] The purpose of this invention is to provide a rapid diagnostic method for turnout operation status based on traction envelope, which addresses the shortcomings of existing turnout fault diagnosis methods by quickly determining whether the traction curve is normal, thereby avoiding a large amount of invalid DTW distance calculation and improving detection efficiency.

[0005] To achieve the above objectives, the present invention provides a rapid diagnostic method for turnout operating status based on traction envelope, comprising the following steps:

[0006] Step 1: Extract the upper and lower envelopes of the normal reference curve from the traction force curve sample library to obtain the threshold of the normal reference curve;

[0007] Step 2: Determine whether the number of data points of the traction force curve to be tested exceeds the threshold of the normal reference curve. If yes, proceed to step 4; otherwise, proceed to step 3.

[0008] Step 3: Determine whether the traction force curve to be tested is within the upper and lower envelopes of the normal reference curve. If so, the test result is that the traction force curve to be tested is normal and the process ends; otherwise, proceed to step 4.

[0009] Step 4: Use the optimized DTW algorithm to match the traction force curve to be detected with the sample curves in the traction force curve sample library;

[0010] Step 5: Determine whether the matching result between the traction force curve to be tested and the sample curve in the traction force curve sample library is that the traction force curve to be tested is normal. If so, the detection result is that the traction force curve to be tested is normal. Then update the upper and lower envelopes of the normal reference curve in the traction force curve sample library and end the process.

[0011] Otherwise, the detection result matches the state information carried by the traction force curve to be detected, and the process ends.

[0012] Furthermore, in step 1, the ability of the upper and lower envelope lines to screen normal reference curves is positively correlated with the threshold size;

[0013] The normal reference curve is the standard traction force curve F0 of length n under normal turnout operation. Its upper and lower envelopes are initialized by considering amplitude and time shift deviations, and are dynamically updated during turnout operation. The initialization method for its upper and lower envelopes is as follows: Where U0 and L0 are the initialization constructions of the upper and lower envelopes, and N U N C The constants that control the amplitude of the upper and lower envelopes are used to increase the margin in the amplitude of the envelopes. r is a constant that controls the time offset window and is used to increase the margin in the time shift deviation of the envelopes.

[0014] Furthermore, in step 2, the threshold is set based on the fluctuation range of the number of data points of the normal reference curve, while adding a margin.

[0015] The ability to prune curves under abnormal operating conditions by comparing the number of data points is negatively correlated with the threshold value. If the number of data points of the traction force curve to be tested significantly exceeds the fluctuation range of the number of data points of the normal reference curve, then the traction force curve to be tested is in an abnormal operating state and does not need to be compared with the envelope.

[0016] Furthermore, in step 3, if the length of the traction force curve to be tested is less than or equal to the length of the envelope, then each data point of the traction force curve to be tested is compared to see if it is between the corresponding points of the upper and lower envelopes.

[0017] If the length of the traction force curve to be tested is greater than the length of the envelope, then the data points that the traction force curve exceeds are compared with the last data points of the upper and lower envelopes.

[0018] Furthermore, in step 3, for an upper envelope U and a lower envelope L of length n, and a traction force curve F of length m to be tested, if the traction force curve F to be tested lies between the upper envelope U and the lower envelope L, then the following equation is satisfied: .

[0019] Furthermore, in step 4, optimizing the DTW algorithm includes methods such as early termination, using a lower bound function, and adjusting the calculation order of lower bound function data points.

[0020] The process of matching the traction force curve F to be detected with the sample curves in the traction force curve sample library using the optimized DTW algorithm includes:

[0021] S1: The descending index value of the absolute value of the traction force curve to be tested after Z-score standardization;

[0022] S2: Initialize the matching curve to the first sample curve in the curve library, and initialize the minimum distance of DTW to infinity;

[0023] S3: If all the sample curves in the curve sample library have been traversed, the process ends; otherwise, each sample curve is retrieved sequentially.

[0024] S4: Calculate the lower bound function of the sample curve. If the length of the traction force curve to be detected and the sample curve are the same, the lower bound function type is LB_Keogh; otherwise, the type is LB_Yi.

[0025] S5: Calculate the DTW lower bound distance between the curve to be detected and the lower bound function in descending index order. If the distance is greater than the minimum DTW distance during the calculation process, jump to step S3; otherwise, go to step S6.

[0026] S6: Calculate the DTW distance between the curve to be detected and the sample curve. If the distance is greater than the minimum DTW value, then jump to step S3. Otherwise, update the minimum value to the DTW distance, update the matching curve to the sample curve, and jump to step S3.

[0027] Furthermore, in step S1, for a sequence x1, x2, ..., x with mean m and variance s... n Its Z-score standardized sequence y1,y2,…,y n The calculation formula is: .

[0028] Furthermore, in step S2, the sample curves in the sample curve library include historical fault curves, which include six categories: fault curves caused by turnout tightness leading to unlocking difficulties, fault curves caused by switch rail deformation leading to switch rail rebound, fault curves caused by foreign object intrusion leading to switching obstruction, fault curves caused by overtight switch rail bolts leading to switching difficulties, fault curves caused by spring aging leading to unbalanced switching force, and fault curves caused by loose operating nuts leading to locking obstruction. It is determined whether there are any new fault types in the historical fault curves that do not belong to the curve sample library. If so, the sample curve library is updated.

[0029] Furthermore, in step S4, LB_Keogh and LB_Yi are the two existing types of lower bound functions:

[0030] If the length of the traction force curve to be detected is the same as the length of the envelope, the LB_Keogh function constructs a sequence of the sliding time window that closely follows the upper and lower envelopes to realize the scaling calculation of the DTW distance.

[0031] If the length of the traction force curve to be detected is different from the length of the envelope, the LB_Yi function selects the maximum and minimum values ​​of one of the sequences as a loose envelope to perform scaling calculation.

[0032] Furthermore, in step S5, the update of the upper and lower envelopes of the normal reference curve, combined with amplitude and time shift deviation, includes the following steps:

[0033] S7: For the normal traction force F of the upper envelope U and lower envelope L, which exceed the original length n, first apply the formula... Update amplitude information; where n is the length of U and L, and m is the length of F;

[0034] S8: Based on the amplitude update, according to the formula Perform time-shift bias update; denote the set of indices of updated data points in the original U and L as index. U index L , where r is a constant that controls the time offset window, and its size is set in the same way as in step 1.

[0035] Beneficial Effects: This invention provides a rapid detection method for turnout operating status based on traction force envelope. Specifically, if the traction force curve to be detected is within the upper and lower envelopes of the normal reference curve, the detection result is normal; if the traction force curve to be detected exceeds the envelope, the detection result is the matching curve information obtained using the DTW algorithm. Furthermore, for cases where the matching result is normal, the upper and lower envelopes of the normal reference curve are further updated and written into the curve sample library. According to the method disclosed in this invention, whether the traction force curve is normal can be quickly determined, thereby avoiding a large number of invalid DTW distance calculations and improving detection efficiency. Attached Figure Description

[0036] Figure 1 is a flowchart of the turnout status detection process according to an embodiment of the present invention;

[0037] Figure 2 is a flowchart of the optimized DTW algorithm involved in the embodiment of the present invention;

[0038] Figure 3 is a schematic diagram of the lower bound function of LB_Yi involved in the embodiment of the present invention;

[0039] Figure 4 is a schematic diagram of the LB_Keogh lower bound function involved in the embodiment of the present invention;

[0040] Figure 5 is a normal reference curve and its initial upper and lower envelope diagram involved in the embodiments of the present invention;

[0041] Figure 6 is a schematic diagram of the detection result of the embodiment of the present invention, in which the traction force curve to be detected is normal but exceeds the initial envelope;

[0042] Figure 7 is a schematic diagram of the initial envelope amplitude update according to an embodiment of the present invention;

[0043] Figure 8 is a schematic diagram of the initial envelope time offset update according to an embodiment of the present invention;

[0044] Figure 9 is a curve sample library diagram related to an embodiment of the present invention. Detailed Implementation

[0045] The preferred mechanisms and implementation methods of the present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0046] As shown in Figures 1 to 9, this invention discloses a technical solution for a rapid detection method of turnout operating status based on traction force envelope. Example 1

[0047] This invention provides a rapid detection method for turnout operating status based on traction force envelope. The overall flowchart of turnout status detection is shown in Figure 1, including the following steps:

[0048] Step 1: Extract the upper and lower envelopes of the normal reference curve from the traction force curve sample library to obtain the threshold of the normal reference curve;

[0049] Step 2: Determine whether the number of data points of the traction force curve to be tested exceeds the threshold of the normal reference curve. If yes, proceed to step 4; otherwise, proceed to step 3.

[0050] Step 3: Determine whether the traction force curve to be tested is within the upper and lower envelopes of the normal reference curve. If so, the test result is that the traction force curve to be tested is normal and the process ends; otherwise, proceed to step 4.

[0051] Step 4: Use the optimized DTW algorithm to match the traction force curve to be detected with the sample curves in the traction force curve sample library;

[0052] Step 5: Determine whether the matching result between the traction force curve to be tested and the sample curve in the traction force curve sample library is that the traction force curve to be tested is normal. If so, the detection result is that the traction force curve to be tested is normal. Then update the upper and lower envelopes of the normal reference curve in the traction force curve sample library and end the process.

[0053] Otherwise, the detection result matches the state information carried by the traction force curve to be detected, and the process ends. Example 2

[0054] The steps in this embodiment are basically the same as those in Embodiment 1, except that in step 1, the normal reference curve is the standard traction force curve F0 of length n under normal turnout operation. The initialization construction of its upper and lower envelopes combines amplitude and time shift deviation, and is dynamically updated during turnout operation. The initialization construction method of its upper and lower envelopes is as follows: Where U0 and L0 are the initialization constructions of the upper and lower envelopes, and N U N C Constants controlling the amplitudes of the upper and lower envelopes are used to increase the margin in the envelope amplitude. `r` is a constant controlling the time offset window, used to increase the margin in the time shift deviation of the envelope. It should be noted that the envelope's ability to filter normal curves is positively correlated with the threshold, but an excessively large threshold may lead to abnormal curves being falsely classified as normal. Since the envelope is dynamically updated during operation, a more robust, slightly smaller value can be selected during initialization.

[0055] In this embodiment, the sampling interval of the normal reference curve F0 is 10ms, and its number of points is n. The method for initializing the upper and lower envelopes is as follows: Here, the amplitude of both the upper and lower envelopes is set to a traction force of 900N, and the time window is set to 20 points before and after, i.e., 0.2s before and after.

[0056] In step 3, for the upper envelope U and lower envelope L of length n, and the traction force curve F to be tested of length m, F is considered to be between the envelopes when the following equation is satisfied: ;

[0057] In step 4, optimizing the DTW algorithm includes methods such as early termination, using a lower bound function, and adjusting the calculation order of lower bound function data points.

[0058] The process of matching the traction force curve F to be detected with the sample curves in the traction force curve sample library using the optimized DTW algorithm is shown in Figure 2, including the following steps:

[0059] S1: The descending index value of the absolute value of the traction force curve to be tested after Z-score standardization;

[0060] S2: Initialize the matching curve to the first sample curve in the curve library, and initialize the minimum distance of DTW to infinity;

[0061] S3: If all the sample curves in the curve sample library have been traversed, the process ends; otherwise, each sample curve is retrieved sequentially.

[0062] S4: Calculate the lower bound function of the sample curve. If the length of the traction force curve to be detected and the sample curve are the same, the lower bound function type is LB_Keogh; otherwise, the type is LB_Yi.

[0063] S5: Calculate the DTW lower bound distance between the curve to be detected and the lower bound function in descending index order. If the distance is greater than the minimum DTW distance during the calculation process, jump to step S3; otherwise, go to step S6.

[0064] S6: Calculate the DTW distance between the curve to be detected and the sample curve. If the distance is greater than the minimum DTW value, then jump to step S3. Otherwise, update the minimum value to the DTW distance, update the matching curve to the sample curve, and jump to step S3.

[0065] Specifically, in step S1, for the sequence x1,x2,…,x… n Given a sequence y1, y2, ..., y with mean m and variance s, then the Z-score standardized sequence is... n The calculation formula is: ;

[0066] Specifically, in step S2, the sample curves in the sample curve library include historical fault curves, which include six categories: fault curves caused by turnout tightness leading to unlocking difficulties, fault curves caused by switch rail deformation leading to switch rail rebound, fault curves caused by foreign object intrusion leading to switching difficulties, fault curves caused by overtight switch rail bolts leading to switching force imbalance due to spring aging, and fault curves caused by loose operating nuts leading to locking blockage. Since there are many types of turnout faults, the sample curves in the curve sample library cannot cover all fault situations and need to be updated during operation. For example, through regular inspections by professional maintenance personnel or cluster analysis, it can be determined whether there are new fault types in the historical fault curves that do not belong to the curve sample library. If so, the sample curve library is updated.

[0067] Specifically, in step S4, LB_Keogh and LB_Yi are two existing types of lower bound functions, as shown in Figures 3 and 4. The LB_Keogh function constructs the upper and lower envelopes of one sequence based on the sliding time window to realize the scaling calculation of the DTW distance. The calculation result is more accurate, but it requires the two sequences to have the same length. The LB_Yi function selects the maximum and minimum values ​​of one sequence as a loose envelope to realize the scaling calculation. The calculation result has lower accuracy, but it does not require the sequence length.

[0068] Specifically, in step S5, the traditional lower bound distance calculation order is from left to right. However, the actual turnout has relatively small data in the initial unlocking stage, which has a low impact on the lower bound distance. The data points with a high impact on the lower bound distance are concentrated in the middle conversion stage. Therefore, the descending index obtained in step S1 is used to correct the calculation process to reduce the amount of calculation.

[0069] Furthermore, in step 5, for the normal traction force F exceeding the original upper envelope U and lower envelope L, the amplitude information is first updated according to the following formula: ;

[0070] Where n is the length of U and L, and m is the length of F. Let the set of indices of updated data points in the original U and L be denoted as index. U index L Then, the formula for updating the time shift bias based on the amplitude update is: ;

[0071] Where r is a constant that controls the time offset window, and its size is set in the same way as in step 1.

[0072] Figures 5 to 8 illustrate the initial envelope update process of the normal reference curve in a specific embodiment. Figure 5 shows the normal reference curve and its initial upper and lower envelopes. Figure 6 shows a schematic diagram where the detection result is that the traction force curve to be detected is normal but exceeds the initial envelope. Figure 7 is a schematic diagram of the initial envelope amplitude update. Figure 8 is a schematic diagram of the initial envelope time offset update involved in the embodiment of the present invention. Example 3

[0073] The steps in this embodiment are basically the same as those in Embodiment 2. The difference is that the traction force data of a certain subway line, a certain station, and a certain turnout switch machine in December 2023 and January 2024 are used as examples to further explain and illustrate the method proposed in this patent.

[0074] The turnout switch machine had 3 sets of abnormal data and 1070 sets of normal data in December 2023, and 3 sets of abnormal data and 2371 sets of normal data in January 2024. A sample curve library and the upper and lower envelopes of the normal reference curve were constructed using the data from December 2023, as shown in Figure 9.

[0075] A Python program was used to compare the efficiency of the method proposed in this patent with the traditional DTW method. The hardware environment was as follows: CPU R7-5800H, GPU RTX3060, 16GB memory, and Python version 3.10. The traditional method of traversing and calculating the DTW distance between all sample curves took 130.09 seconds, while the envelope-based method proposed in this patent took 3.41 seconds. Of the 2371 normal data sets, 2327 sets were within the envelope, and only 44 sets required DTW distance calculation, significantly improving detection efficiency. Furthermore, all three sets of abnormal data used in the test matched the faulty sample curve 3. The matching curves carried information indicating conversion difficulties, consistent with reality. Simultaneously, no normal data were misclassified as abnormal curves, indicating a detection accuracy of 100%.

[0076] This invention provides a rapid detection method for turnout operating status based on traction force envelope. Specifically, if the traction force curve to be detected is within the upper and lower envelopes of a normal reference curve, the detection result is normal. If the traction force curve to be detected exceeds the envelope, the detection result is the matching curve information obtained using the DTW algorithm. Furthermore, for cases where the matching result is normal, the upper and lower envelopes of the normal reference curve are updated and written into the curve sample library. According to the method disclosed in this invention, whether the traction force curve is normal can be quickly determined, thereby avoiding a large number of invalid DTW distance calculations and improving detection efficiency.

[0077] Finally, it should be noted that the above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. However, any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for rapid detection of turnout operating status based on traction force envelope, characterized in that, Includes the following steps: Step 1: Extract the upper and lower envelopes of normal reference curves from the traction curve sample library to obtain the threshold of normal reference curves; in Step 1, the ability of the upper and lower envelopes to screen normal reference curves is positively correlated with the threshold size; The normal reference curve is the standard traction force curve F0 of length n under normal turnout operation. Its upper and lower envelopes are initialized by considering amplitude and time shift deviations, and are dynamically updated during turnout operation. The initialization method for its upper and lower envelopes is as follows: ,in, 、 These represent the initialization construction of the upper and lower envelopes, respectively. This is a constant representing the control time offset window, used to increase the margin in the envelope time shift deviation. 、 Step 1: Control the constants of the upper and lower envelope amplitudes to increase the margin in the envelope amplitude; Step 2: Determine whether the number of data points of the traction force curve to be tested exceeds the threshold of the normal reference curve. If yes, proceed to Step 4; otherwise, proceed to Step 3. In Step 2, the threshold setting is based on the fluctuation range of the number of data points of the normal reference curve, while increasing the margin. The ability of the curve to prune abnormal operation states by comparing the number of data points is negatively correlated with the threshold size. If the number of data points of the traction force curve to be tested significantly exceeds the fluctuation range of the number of data points of the normal reference curve, then the traction force curve to be tested is in an abnormal operation state and does not need to be compared with the envelope; Step 3: Determine whether the traction force curve to be tested is located within the upper and lower envelopes of the normal reference curve. If yes, the detection result is that the traction force curve to be tested is normal and the process ends; otherwise, proceed to Step 4. In step 3, if the length of the traction force curve to be tested is less than or equal to the length of the envelope, then each data point of the traction force curve to be tested is compared to see if it is between the corresponding points of the upper and lower envelopes; if the length of the traction force curve to be tested is greater than the length of the envelope, then all data points of the traction force curve to be tested that exceed the envelope are compared with the last data points of the upper and lower envelopes; step 4: use the optimized DTW algorithm to match the traction force curve to be tested with the sample curves in the traction force curve sample library; step 5: determine whether the matching result between the traction force curve to be tested and the sample curves in the traction force curve sample library indicates that the traction force curve to be tested is normal. If so, the detection result indicates that the traction force curve to be tested is normal, then the upper and lower envelopes of the normal reference curves in the traction force curve sample library are updated and the process ends; otherwise, the detection result indicates that the state information carried by the traction force curve to be tested is matched and the process ends.

2. The method for rapid detection of turnout operating status based on traction envelope according to claim 1, characterized in that, In step 3, for an upper envelope U and a lower envelope L of length n, and a traction force curve F of length m to be tested, if the traction force curve F to be tested lies between the upper envelope U and the lower envelope L, then the following equation is satisfied: 。 3. The method for rapid detection of turnout operating status based on traction envelope according to claim 1, characterized in that, In step 4, optimizing the DTW algorithm includes methods such as early termination, using a lower bound function, and adjusting the calculation order of lower bound function data points. The process of matching the traction force curve F to be detected with the sample curves in the traction force curve sample library using the optimized DTW algorithm includes: S1: After Z-score standardization of the traction force curve to be detected, calculate the descending index value of the absolute value; S2: Initialize the matching curve as the first sample curve in the curve library, and initialize the DTW minimum distance to infinity; S3: If all sample curves in the curve sample library have been traversed, the process ends; otherwise, retrieve the sample curves sequentially; S4: Calculate the lower bound function of the sample curves. If the traction force curve to be detected... If the length of the lower bound function is the same as that of the sample curve, then the type of the lower bound function is LB_Keogh; otherwise, the type is LB_Yi. S5: Calculate the DTW lower bound distance between the curve to be detected and the lower bound function in descending index order. If the distance is greater than the minimum DTW distance during the calculation, then jump to step S3; otherwise, go to step S6. S6: Calculate the DTW distance between the curve to be detected and the sample curve. If the distance is greater than the minimum DTW value, then jump to step S3; otherwise, update the minimum value to the DTW distance, update the matching curve to the sample curve, and jump to step S3.

4. The method for rapid detection of turnout operating status based on traction envelope according to claim 3, characterized in that, In step S1, for a sequence x1,x2,…,x with mean m and variance s... n Its Z-score standardized sequence y1,y2,…,y n The calculation formula is: 。 5. The method for rapid detection of turnout operating status based on traction envelope according to claim 3, characterized in that, In step S2, the sample curves in the sample curve library include historical fault curves, which include six categories: fault curves caused by turnout tightness leading to unlocking difficulties, fault curves caused by switch rail deformation leading to switch rail rebound, fault curves caused by foreign object intrusion leading to switching obstruction, fault curves caused by overtight switch rail bolts leading to switching difficulties, fault curves caused by spring aging leading to unbalanced switching force, and fault curves caused by loose operating nuts leading to locking obstruction. It is determined whether there are any new fault types in the historical fault curves that do not belong to the curve sample library. If so, the sample curve library is updated.

6. The method for rapid detection of turnout operating status based on traction envelope according to claim 3, characterized in that, In step S4, LB_Keogh and LB_Yi are two existing types of lower bound functions: if the length of the traction force curve to be detected is the same as the length of the envelope, the LB_Keogh function constructs a sequence that closely follows the upper and lower envelopes according to the sliding time window to realize the scaling calculation of the DTW distance; if the length of the traction force curve to be detected is different from the length of the envelope, the LB_Yi function selects the maximum and minimum values ​​of a sequence as a loose envelope to realize the scaling calculation.

7. The method for rapid detection of turnout operating status based on traction envelope according to claim 3, characterized in that, In step S5, the updates of the upper and lower envelopes of the normal reference curve are combined with the amplitude and time shift deviation. Includes the following steps: S7: For the traction force curve F to be tested, which exceeds the original length n of the upper envelope U and lower envelope L, first use the formula... Update amplitude information, where m is the length of the traction force curve F to be detected; S8: Based on the amplitude update, according to the formula Perform time shift bias updates.

Images