Track structure disease identification method and device based on fiber grating array

Abstract

The invention relates to a track structure disease recognition method and device based on a fiber bragg grating array, and belongs to the technical field of track disease recognizing.The method comprises the steps that effective information extraction is conducted on an initial vibration signal, an effective vibration signal is obtained, and therefore data of the fiber bragg grating array can be screened; the data volume of the to-be-processed data is reduced; and sequence calculation is carried out on the effective vibration signals based on a standard basic database to obtain a plurality of OSA sequences, so that error estimation of each sensing point is obtained, frequency characteristics of the signals can be reserved on the premise of reducing noise influence, and therefore, when region division is carried out on the plurality of sensing points according to disease data of error estimation, the accuracy of region division is improved. According to the method, the accuracy of area division is improved, curve fitting and disease category judgment can be carried out according to all error estimation in the sensing detection area, a disease recognition result is obtained, disease recognition of the track structure is achieved, and the accuracy of disease recognition is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of track disease identification, and particularly to a method and device for identifying track structure diseases based on an optical fiber grating array. Background Art

[0002] During the daily service of urban rail transit, due to the frequent running of trains, structural diseases inevitably occur. Common structural diseases include ballast bed cracks, sleeper looseness, fastener looseness, shear hinge looseness, steel spring looseness, etc. These structural diseases may all affect the safety of train operation. At present, the maintenance of urban rail transit is mainly carried out through manual track inspection and inspection vehicle inspection during the early morning empty window period. Relying on these measures, the workload is large, the timeliness is poor, and the subjectivity of manual work is strong. When the rail transit is in the operation period, it is impossible to obtain the structural state information. Although traditional electrical sensors can be used for daily monitoring, the detection range of electrical sensors is limited. If the sensor density is increased, it will increase potential safety hazards.

[0003] In order to alarm and give early warning of existing structural damages and possible structural diseases, the optical fiber grating array sensing technology in the prior art can provide full-time and full-domain, accurate data collection, and has characteristics such as anti-electromagnetic interference, large networking capacity, and high sensitivity, which can well overcome the pain points of electrical sensors being affected by electromagnetic interference, signals being unable to be transmitted over long distances, and limited detection range. However, at the same time, the monitoring data volume of the optical fiber grating array sensing network is huge, and there are many challenges in processing it. Especially when extracting disease signal features based on train vibration, due to the low signal-to-noise ratio, it is difficult to effectively identify disease signals, which brings additional difficulties to the identification and processing of structural diseases.

[0004] Therefore, there is an urgent need to propose a method and device for identifying track structure diseases based on an optical fiber grating array to solve the technical problem in the prior art that when using the optical fiber grating array sensing technology to identify track diseases, it is difficult to effectively identify disease signals due to the low signal-to-noise ratio. Summary of the Invention

[0005] In view of this, it is necessary to provide a method and device for identifying track structure diseases based on an optical fiber grating array to solve the technical problem in the prior art that it is difficult to effectively identify disease signals by the optical fiber grating array sensing technology.

[0006] To solve the above problems, in the first aspect, the present invention provides a method for identifying track structure diseases based on an optical fiber grating array, including: When a train passes through the sensing measurement points of the optical fiber grating array on the track, obtain the initial vibration signal of each sensing measurement point, and extract the effective information from the initial vibration signal to obtain an effective vibration signal; Perform sequence calculations on the effective vibration signals based on the standard basic database to obtain multiple OSA sequences, and obtain the error estimate of each sensing measurement point according to the multiple OSA sequences; Divide continuous multiple sensing measurement points with disease data in the error estimate into regions to obtain sensing measurement areas; Perform curve fitting and disease category judgment based on all error estimates in the sensing measurement area to obtain a disease recognition result.

[0007] In a possible implementation, the extracting effective information from the initial vibration signal to obtain an effective vibration signal includes: Determine all initial signal frames in the initial vibration signal according to a preset fixed length; Calculate each initial signal frame respectively to obtain the corresponding short-time energy and short-time zero-crossing rate; Determine the first initial signal frame in which the short-time energy or the short-time zero-crossing rate in all the initial signal frames is greater than a preset threshold as the start frame; Determine the first initial signal frame in which the short-time energy and the short-time zero-crossing rate after the start frame in all the initial signal frames are not greater than the preset threshold as the end frame; Obtain an effective vibration signal according to the start frame and the end frame.

[0008] In a possible implementation, the obtaining an effective vibration signal according to the start frame and the end frame includes: When the number of frames is greater than a preset number-of-frames threshold, determine all effective signal frames between the start frame and the end frame as the effective vibration signal; the number of frames is the number of effective signal frames between the start frame and the end frame; the effective signal frame is a signal frame in which the short-time energy or the short-time zero-crossing rate is greater than the preset threshold; When the number of frames is not greater than the preset number-of-frames threshold, determine the first initial signal frame in which the short-time energy and the short-time zero-crossing rate after the end frame in all the initial signal frames are not greater than the preset threshold as the new end frame, and obtain an effective vibration signal according to the start frame and the new end frame.

[0009] In a possible implementation, the performing sequence calculations on the effective vibration signal based on the standard basic database to obtain multiple OSA sequences includes: Compare the effective vibration signal with all standard signals in the standard basic database to determine a reference signal; Synchronize the effective vibration signal and the reference signal in the time domain to obtain a synchronized vibration signal and a synchronized reference signal; Calculate the synchronous vibration signal and the synchronous reference signal respectively to obtain a first OSA sequence and a second OSA sequence.

[0010] In a possible implementation manner, the error estimation includes mean square error and mean absolute error; obtaining the error estimation of each sensing measurement point according to the multiple OSA sequences includes: Calculate the mean square error and the mean absolute error respectively by calculating the first OSA sequence and the second OSA sequence. After obtaining the error estimation of each sensing measurement point according to the multiple OSA sequences, it further includes: Judge whether the mean square error or the mean absolute error is greater than a preset error threshold; If so, determine that there is disease data in the effective vibration signal; If not, determine that there is no disease data in the effective vibration signal.

[0011] In a possible implementation manner, the step of synchronizing the effective vibration signal and the reference signal in the time domain to obtain a synchronous vibration signal and a synchronous reference signal includes: Calculate the maximum value of the cross-correlation of the effective vibration signal and the reference signal to obtain a cross-correlation peak; Determine the delay time between the effective vibration signal and the reference signal according to the position of the cross-correlation peak; Translate one of the effective vibration signal and the reference signal according to the delay time to obtain a synchronous vibration signal and a synchronous reference signal.

[0012] In a possible implementation manner, the disease recognition result includes a disease recognition position and a disease category; obtaining the disease recognition result by performing curve fitting and disease category judgment according to all the error estimations in the sensing measurement area includes: Perform curve fitting according to all the error estimations in the sensing measurement area to obtain a fitting curve; Calculate the centroid of the fitting curve to obtain the disease recognition position; Compare the disease data at the disease recognition position with the error estimations of the disease-free signal and the reference signal in the standard basic database to obtain the disease category.

[0013] In a possible implementation manner, the method further includes: Set multiple preset error intervals and corresponding disease levels; Calculate the error estimation at the disease recognition position with the error estimations of the disease-free signal and the reference signal in the standard basic database to obtain a target error; Determine the target disease level corresponding to the target error according to the multiple preset error intervals; Give an early warning to the track according to the target disease level.

[0014] In a possible implementation manner, the method further includes: Perform clustering analysis on the vibration signals in the sensing area to obtain the similarity of each signal frame; Store the signals with the similarity greater than the preset similarity threshold into the standard basic database as standard signals.

[0015] In a second aspect, the present invention further provides an apparatus for identifying track structure diseases based on an optical fiber grating array, including: A data detection module, configured to obtain the initial vibration signal of each sensing point when a train passes through the sensing points of the optical fiber grating array on the track, and perform effective information extraction on the initial vibration signal to obtain an effective vibration signal; An error estimation module, configured to perform sequence calculation on the effective vibration signal based on the standard basic database to obtain multiple OSA sequences, and obtain the error estimation of each sensing point according to the multiple OSA sequences; A region division module, configured to divide a continuous plurality of sensing points with the error estimation being disease data into sensing areas; A result identification module, configured to perform curve fitting and disease category judgment according to all the error estimations in the sensing area to obtain a disease identification result.

[0016] The beneficial effects of the present invention are as follows: When a train passes through the sensing points of the optical fiber grating array on the track, the initial vibration signal of each sensing point is detected, and effective information extraction is performed on the initial vibration signal to obtain an effective vibration signal, so that the data of the optical fiber grating array can be screened, reducing the amount of data to be processed; further, sequence calculation is performed on the effective vibration signal based on the standard basic database to obtain multiple OSA sequences, so as to obtain the error estimation of each sensing point, and further, the frequency characteristics of the signal can be retained on the premise of reducing the influence of noise, so that when dividing the regions of multiple sensing points according to the disease data of the error estimation, the accuracy of region division can be improved, and further, curve fitting and disease category judgment can be performed according to all the error estimations in the sensing area to obtain a disease identification result, realizing the identification of track structure diseases and improving the accuracy of disease identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a schematic flowchart of an embodiment of the method for identifying track structure diseases based on an optical fiber grating array provided by the present invention; Figure 2For the present invention Figure 1 Schematic flowchart of an embodiment of step S101 in the present invention; Figure 3 For the present invention Figure 1 Schematic flowchart of an embodiment of step S102 in the present invention; Figure 4 Schematic structural diagram of an embodiment of the sensing area division of the fiber grating array sensing network provided by the present invention; Figure 5 For the present invention Figure 1 Schematic flowchart of an embodiment of step S104 in the present invention; Figure 6 Schematic structural diagram of an embodiment of the rail structure disease identification device based on the fiber grating array provided by the present invention. Specific embodiments

[0018] The following combines the accompanying drawings to specifically describe the preferred embodiments of the present invention. Among them, the accompanying drawings form a part of this application and are used together with the embodiments of the present invention to explain the principles of the present invention, and are not used to limit the scope of the present invention.

[0019] As Figure 1 shown, a specific embodiment of the present invention discloses a method for identifying rail structure diseases based on a fiber grating array, including: S101. When a train passes through the sensing measurement points of the fiber grating array on the track, obtain the initial vibration signal of each sensing measurement point, and extract the effective information from the initial vibration signal to obtain the effective vibration signal.

[0020] The method for identifying rail structure diseases based on the fiber grating array provided by the embodiments of this application can be applied to a rail structure disease identification system. Among them, the rail structure disease identification can be based on a software system running on a terminal device. The terminal device can be a server, a tablet computer, a vehicle-mounted device, an augmented reality (AR) / virtual reality (VR) device, a notebook computer, an ultra-mobile personal computer (UMPC), a netbook, a personal digital assistant (PDA), a mobile phone, or other terminal devices. The embodiments of this application do not impose any restrictions on the specific type of the terminal device.

[0021] In a specific embodiment of the present invention, an optical fiber grating array vibration sensing optical cable can be arranged on the train track. The optical fiber grating array vibration sensing optical cable can be laid between the two tracks along the track direction, and the optical fiber grating array vibration sensing optical cable is fixed to ensure that the optical fiber grating array vibration sensing optical cable is closely adhered to the track structure. A modulated optical signal is sent by a pulsed light source, and the modulated optical signal is transmitted to the optical fiber grating array vibration sensing optical cable through an optical circulator. The optical signal reflected by the optical fiber grating array vibration sensing optical cable is transmitted to a demodulator, and the demodulator demodulates the reflected optical signal to obtain the initial vibration signal of each sensing measurement point. The initial vibration signal excitation source is the train load. When the train passes through the sensing measurement point, a vibration signal is generated by the wheel-rail coupling effect of the train and the track, that is, the initial vibration signal. The amount of monitoring data of the optical fiber grating array sensing network is too large. In order to perform better detection, the effective information of the initial vibration signal can be extracted, and then the effective vibration signal with redundant information removed can be obtained.

[0022] In some embodiments of the present invention, as Figure 2 shown, step S101 includes: S201. Determine all initial signal frames in the initial vibration signal according to a preset fixed length; S202. Calculate the corresponding short-time energy and short-time zero-crossing rate for each initial signal frame respectively; S203. Determine the first initial signal frame in all initial signal frames whose short-time energy or short-time zero-crossing rate is greater than a preset threshold as the start frame; S204. Determine the first initial signal frame in all initial signal frames after the start frame whose short-time energy and short-time zero-crossing rate are not greater than the preset threshold as the end frame; S205. Obtain the effective vibration signal according to the start frame and the end frame.

[0023] In a specific embodiment of the present invention, before extracting the effective signal frames, in order to prevent the number of frames in the previous round of processing from being stored in the system, the number of counted effective signal frames can be cleared. Before performing this embodiment, the preset fixed length can also be set according to the actual situation, so that the initial vibration signal can be detected according to the preset fixed length, and the initial signal frames that meet the preset fixed length in the initial vibration signal can be determined, so as to obtain all initial signal frames. Then, the short-time energy and short-time zero-crossing rate of each initial signal frame can be calculated respectively. Among them, the calculation of the short-time energy is shown in formula (1): (1) In the formula, represents the short-time energy, represents the signal length of the current signal frame, represents the signal at thei The square of the signal value within a time interval.

[0024] The calculation of the short-time zero-crossing rate is shown in Equation (2): (2) In the formula, represents the short-time zero-crossing rate, represents the sign function, , n represents the signal length, represents the position of the sensor i of.

[0025] A preset threshold can be set. The preset threshold can be set according to the actual situation, and the embodiments of the present invention do not limit it here. After calculating the short-time energy and the short-time zero-crossing rate, the short-time energy and the short-time zero-crossing rate can be judged according to the preset threshold to determine whether the short-time energy or the short-time zero-crossing rate is greater than the preset threshold. When one of them is greater than the preset threshold, the initial signal frame can be determined as a valid signal frame. The first initial signal frame in all initial signal frames whose short-time energy or short-time zero-crossing rate is greater than the preset threshold is determined as the start frame, and then the initial signal frames after the start frame can be judged one by one. The first initial signal frame whose short-time energy and short-time zero-crossing rate after the start frame are not greater than the preset threshold is determined as the end frame. The detected end frame is a normal signal frame, so that the valid vibration signal can be obtained according to the start frame and the end frame.

[0026] , where, after obtaining the valid signal frame, the number of valid signal frames can be determined. For example, if the current signal frame is the start frame, the number of valid signal frames should be 1, and after the next round of detecting a valid signal frame, the number of valid signal frames should be 2, and so on; when the short-time energy and the short-time zero-crossing rate are both not greater than the preset threshold during the subsequent detection process, it means that the currently detected signal is a normal signal frame, and the subsequent signal frames do not need to be detected anymore. Then, according to the valid signal frames detected in the previous round and the determined number of frames, the valid vibration signal is obtained.

[0027] In some embodiments of the present invention, step S205 includes: When the number of frames is greater than the preset number-of-frames threshold, all valid signal frames between the start frame and the end frame are determined as the valid vibration signal; the number of frames is the number of valid signal frames between the start frame and the end frame; the valid signal frame is a signal frame whose short-time energy or short-time zero-crossing rate is greater than the preset threshold; When the number of frames is not greater than the preset number-of-frames threshold, the first initial signal frame whose short-time energy and short-time zero-crossing rate after the end frame in all initial signal frames are not greater than the preset threshold is determined as the new end frame, and the valid vibration signal is obtained according to the start frame and the new end frame.

[0028] In a specific embodiment of the present invention, after determining the start frame and the end frame, it is possible to determine whether the number of valid signal frames between the start frame and the end frame is greater than a preset frame number threshold. Herein, the preset frame number threshold can be set according to the actual situation, and the embodiments of the present invention do not limit it here. The frame number is the number of all valid signal frames between the start frame and the end frame; a valid signal frame is a signal frame whose short-time energy or short-time zero-crossing rate is greater than a preset threshold. Therefore, the frame number is the number of signal frames whose short-time energy or short-time zero-crossing rate between the start frame and the end frame is greater than the preset threshold, including the start frame. When the frame number is greater than the preset frame number threshold, it indicates that the number of all valid signal frames between the start frame and the end frame has met the condition, and all valid signal frames can be determined as valid vibration signals for subsequent processes. When the frame number is not greater than the preset frame number threshold, it indicates that the number of all valid signal frames between the start frame and the end frame is insufficient, and the end frame needs to be re-determined to increase the number of valid signal frames. Then, the first initial signal frame whose short-time energy and short-time zero-crossing rate after the end frame in all initial signal frames are not greater than the preset threshold is determined as the new end frame. Then, it is possible to determine again whether the number of all valid signal frames between the start frame and the new end frame is greater than the preset frame number threshold. If so, all valid signal frames between the start frame and the new end frame are determined as valid vibration signals for subsequent processes. If not, the new end frame is determined again until the condition is met.

[0029] Further, after obtaining the valid vibration signal, the valid vibration signal can be filtered to obtain a signal with the vibration frequency characteristics of the track structure, so that the valid vibration signal is a signal with the vibration frequency characteristics of the track structure, and the valid vibration signals in subsequent processes are all signals including the vibration frequency characteristics of the track structure.

[0030] When there is no external excitation to make the track structure vibrate in the embodiments of the present invention, the signals transmitted back by the vibrating optical cable are environmental background noises, and these signals cannot reflect the structural disease conditions. When the train passes through the measurement area, the vibration of the track structure can reflect the structural disease information. Therefore, it is necessary to accurately extract the vibration signal of the train passing through this measurement area at the current time from the vibration signals collected for a long time through the extraction of valid vibration signals. Based on the real-time vibration data signal stream collected, the signal stream is processed by frame division to remove the redundant signal stream and obtain the valid vibration signal.

[0031] S102. Perform sequence calculation on the valid vibration signal based on the standard basic database to obtain multiple OSA sequences, and obtain the error estimation of each sensing measurement point according to the multiple OSA sequences.

[0032] In a specific embodiment of the present invention, a standard basic database may be set up. The standard basic database may store historical standard data or include standard data obtained through acquisition and other means. Thus, sequence calculations can be performed on the effective vibration signals based on the data in the standard basic database, and then multiple OSA sequences can be obtained. Subsequently, error estimates for each sensing measurement point can be obtained based on the multiple OSA sequences.

[0033] In order to reduce the influence of noise, in some embodiments of the present invention, as Figure 3 shown, step 102 includes: S301. Compare the effective vibration signal with all standard signals in the standard basic database to determine the reference signal; S302. Synchronize the effective vibration signal and the reference signal in the time domain to obtain a synchronized vibration signal and a synchronized reference signal; S303. Calculate the synchronized vibration signal and the synchronized reference signal respectively to obtain a first OSA sequence and a second OSA sequence.

[0034] In a specific embodiment of the present invention, the effective vibration signal can be compared with all standard signals in the standard basic database, and the one with the highest similarity can be determined as the reference signal. Then, the effective vibration signal and the reference signal can be synchronized in the time domain to obtain a synchronized vibration signal and a synchronized reference signal. Subsequently, the synchronized vibration signal and the synchronized reference signal can be calculated respectively to obtain the first OSA sequence of the synchronized vibration signal and the second OSA sequence of the synchronized reference signal. The calculation of the OSA sequence is shown in formula (3): (3) In the formula, represents the calculated OSA sequence, is the delay parameter. When is the synchronized vibration signal, the OSA sequence is denoted as , which is the first OSA sequence of the synchronized vibration signal. When is the synchronized reference signal, the OSA sequence is denoted as , which is the second OSA sequence of the synchronized reference signal.

[0035] Furthermore, after obtaining the first OSA sequence of the synchronized vibration signal and the second OSA sequence of the synchronized reference signal, the first OSA sequence of the synchronized vibration signal and the second OSA sequence of the synchronized reference signal can be calculated to obtain the error estimate for each sensing measurement point.

[0036] In order to reduce errors in subsequent calculation processes, in some embodiments of the present invention, step S302 includes: Calculate the maximum value of the cross-correlation between the effective vibration signal and the reference signal to obtain the cross-correlation peak value; Determine the delay time between the effective vibration signal and the reference signal according to the position of the cross-correlation peak value; Translate one of the effective vibration signal and the reference signal according to the delay time to obtain the synchronous vibration signal and the synchronous reference signal.

[0037] In a specific embodiment of the present invention, the maximum value of the cross-correlation between the effective vibration signal and the reference signal can be calculated to obtain the cross-correlation peak value, wherein the calculation of the maximum value of the cross-correlation is shown in formula (4): (4) In the formula, represents the effective vibration signal, represents the reference signal.

[0038] The position of the cross-correlation peak value (i.e., the maximum value) corresponds to the delay time between the two signals. Translating one of the signals can achieve the time-domain alignment of the effective vibration signal and the reference signal, and obtain the synchronous vibration signal and the synchronous reference signal after synchronization. For example, adjust the reference signal according to the effective vibration signal, or adjust the effective vibration signal according to the reference signal. When adjusting the reference signal according to the effective vibration signal, the effective vibration signal is the synchronous vibration signal, and the adjusted reference signal is the synchronous reference signal. The same applies to other cases, so that the synchronous vibration signal and the synchronous reference signal can be obtained.

[0039] S103. Divide the continuous multiple sensing measurement points with error estimation as disease data into regions to obtain sensing measurement regions.

[0040] In a specific embodiment of the present invention, when a disease occurs at a certain position, its vibration signal will propagate along the track direction, so there will be multiple consecutive sensors that can monitor this vibration signal. As Figure 4As shown in the figure, it is a schematic diagram of an embodiment of the sensing area division of the fiber Bragg grating array sensing network provided by the embodiment of the present invention, including: track bed 1, fiber Bragg grating array sensing optical cable 2, fiber Bragg grating sensor 3, disease location 4, and sensing area 5. Among them, the fiber Bragg grating sensor 3 is integrated on the fiber Bragg grating array sensing optical cable 2; the disease location 4 can be any position along the track direction; the sensing area 5 is the monitoring range of continuous n sensors including the disease location 4. In order to improve the signal monitoring efficiency and accuracy, continuous n sensors can be divided into 1 sensing area. When a disease appears in the sensing area, several adjacent continuous sensors will detect the disease signal, and the disease of the track structure can be identified and located through multiple sensors. Specifically, according to the error estimation of OSA, multiple sensing measurement points with error estimation of disease data can be divided into one sensing area, such as Figure 4 the sensing area 5 in

[0041] S104. Perform curve fitting and disease category judgment based on all error estimations in the sensing area to obtain the disease recognition result.

[0042] In a specific embodiment of the present invention, in order to identify the location of the disease, curve fitting and disease category judgment can be performed based on all error estimations in the sensing area to obtain the disease recognition result.

[0043] In some embodiments of the present invention, as Figure 5 shown, the disease recognition result includes the disease recognition location and the disease category; step 104 includes: S501. Perform curve fitting based on all error estimations in the sensing area to obtain a fitting curve; S502. Calculate the centroid of the fitting curve to obtain the disease recognition location; S503. Compare the disease data at the disease recognition location with the disease database based on the disease database to obtain the disease category.

[0044] In a specific embodiment of the present invention, curve fitting can be performed on all error estimations in the sensing area. For example, curve fitting can be performed on the mean square error among all error estimations in the sensing area, or curve fitting can be performed on the mean absolute error among all error estimations, and a fitting curve can be obtained. Then, the disease location can be located by calculating the centroid in the horizontal direction of the fitting curve to obtain the disease recognition location. The calculation of the centroid is shown in formula (5): (5) In the formula, represents the position of the sensor i , represents the sensor on the fitting curve iThe OSA sequence value of the corresponding effective vibration signal is the position of the center of gravity in the horizontal direction, that is, the disease identification position.

[0045] A disease database can also be set up. The disease database includes historical disease vibration signals stored historically. The historical disease vibration signals can be signals obtained in actual applications or signals simulated according to modules. Specifically, the embodiments of the present invention are not limited thereto. Thus, the disease data at the disease identification position can be compared with the historical disease vibration signals in the disease database, and the category of the historical disease vibration signal with the largest similarity in the disease database is determined as the disease category at the disease identification position.

[0046] In summary, when the train passes through the sensing measurement points of the fiber Bragg grating array on the track in this embodiment, the initial vibration signal of each sensing measurement point is detected, and the effective information of the initial vibration signal is extracted to obtain the effective vibration signal, so that the data of the fiber Bragg grating array can be screened, reducing the amount of data to be processed; further, sequence calculations are performed on the effective vibration signals based on the standard basic database to obtain multiple OSA sequences, thereby obtaining the error estimate of each sensing measurement point. Furthermore, the frequency characteristics of the signal can be retained on the premise of reducing the influence of noise. Thus, when dividing the areas of multiple sensing measurement points according to the disease data of the error estimate, the accuracy of the area division is improved. Furthermore, curve fitting and disease category judgment can be performed based on all the error estimates in the sensing measurement area to obtain the disease identification result, realizing the disease identification of the track structure and improving the accuracy of the disease identification.

[0047] In some embodiments of the present invention, step 102 further includes: Calculating the first OSA sequence and the second OSA sequence respectively to obtain the mean square error and the mean absolute error.

[0048] In the specific embodiments of the present invention, the first OSA sequence and the second OSA sequence can be calculated through a regression loss function to obtain the mean square error and the mean absolute error respectively. Among them, the mean square error calculation is shown in formula (6): (6) In the formula, represents the calculated mean square error, is the OSA sequence of the synchronous reference signal, is the OSA sequence of the synchronous vibration signal.

[0049] The mean absolute error calculation is shown in formula (7): (7) In the formula, represents the calculated mean absolute error, The OSA sequence for the synchronous reference signal The OSA sequence for the synchronous vibration signal

[0050] In some embodiments of the present invention, after step 102, it further includes: Determine whether the mean square error or the mean absolute error is greater than a preset error threshold; If so, it is determined that there is disease data in the effective vibration signal; If not, it is determined that there is no disease data in the effective vibration signal.

[0051] In a specific embodiment of the present invention, a preset error threshold can be set. For example, set the threshold , It represents the mean square error or the mean absolute error between the disease-free signal and the reference signal in the standard basic database. The preset error threshold can be 1.3· , and the mean square error or the mean absolute error can be judged respectively. If one of them is greater than the preset error threshold, it can be determined that there is disease data in the effective vibration signal. If not, it can be determined that there is no disease data in the effective vibration signal.

[0052] In some embodiments of the present invention, the method further includes: Set multiple preset error intervals and corresponding disease levels; Calculate the error estimate at the disease recognition position and the error estimate between the disease-free signal and the reference signal in the standard basic database to obtain the target error; Determine the target disease level corresponding to the target error according to multiple preset error intervals; Warn the track according to the target disease level.

[0053] In a specific embodiment of the present invention, in order to give an alarm, multiple preset error intervals and corresponding disease levels can be set according to the actual situation. For example, the alarm levels of the hierarchical alarm are set as disease-free, minor disease, moderate disease, and severe disease: Disease-free: , the structural condition is normal and there is no disease situation, Minor disease: , the defects or damages of the structure are not significant and will not immediately affect the overall stability and safety of the structure; Moderate disease: , the structure has a certain degree of damage or defect, and monitoring and planned maintenance are required to prevent the disease from deteriorating further; Severe disease: , the structure is severely damaged, which may have affected the normal use of the track structure, and immediate measures need to be taken for repair or reinforcement.

[0054] Among them, is the error estimate between the effective vibration signal and the reference signal, is the error estimate between the disease-free signal and the reference signal in the standard basic database.

[0055] Therefore, the error estimate at the disease recognition position can be calculated with the error estimate between the disease-free signal and the reference signal in the standard basic database to obtain the target error, that is, = target error. Then, the preset error interval is determined according to the magnitude of the target error, so as to determine the target disease level corresponding to the target error. Furthermore, early warning of the track can be carried out according to the target disease level.

[0056] In some embodiments of the present invention, the method further includes: Performing clustering analysis on the vibration signals in the sensing area to obtain the similarity of each signal frame; Storing the signals with similarity greater than the preset similarity threshold into the standard basic database as standard signals.

[0057] In a specific embodiment of the present invention, the standard basic database may include the structural characteristics of each sensing area itself and the location where the sensing area is located. Then, by performing clustering analysis on the extracted effective vibration signals, the signals with higher similarity are stored in the standard basic database as standard signals, thereby updating the standard basic database. The information of the standard database can also be updated regularly to ensure the accuracy of data processing. The specific clustering process may be as follows: Set the number of clusters K, and randomly select K data points as the initial centroids; for each data point in the dataset, calculate its Euclidean distance from all centroids, and assign it to the cluster represented by the nearest centroid; recalculate the centroid of each cluster by taking the mean of all points in the cluster; check whether the centroid has changed. If the centroid has changed, return to step 2; otherwise, output the final cluster division result; among them, the calculation of the Euclidean distance is shown in formula (8): (8) In the formula, is the data point, is the th clustering center, is the data dimension, and are respectively and the values on the th dimension.

[0058] Furthermore, the disease database determines the data information of diseases by statistically analyzing the OSA calculation results of multiple train passages and the changing trends of the calculation results, which can reflect the change process of the structure, and the disease database can update the disease data according to the actual situation.

[0059] In the embodiment of the present invention, the vibration signal of the track structure when the train passes is obtained through the fiber Bragg grating array sensing optical cable, which has the advantages of anti-electromagnetic interference, simple structure, convenient installation, high sensitivity, etc. A single optical cable can realize the functions of signal measurement and transmission without the need to additionally establish signal transmission cables, and there is no need for power supply in the monitoring line, which is safer and more reliable, and can conveniently realize the full-time, full-domain, large-scale, and long-distance monitoring of track structure diseases; the vibration signal is processed and analyzed by the method of unilateral autocorrelation sequence, and the noise in the signal can be effectively analyzed and removed, while the frequency characteristics of the track structure vibration can be retained; by determining the reference signal and constructing the disease database, it can quickly calculate whether there are diseases in the track structure and identify the disease types for the reference of relevant personnel; by setting the threshold, the sensitivity of the disease recognition method can be adjusted, and an early warning can be issued at the initial stage of the structural disease, and the maintenance personnel can perform fixed-point maintenance in a timely manner according to the warning information to avoid major losses caused by the deterioration of the track structure diseases, and protect the safety of people's lives and property to the greatest extent; online data analysis and hierarchical early warning enable the dispatching center to always master the health status of the track structure and respond to diseases of different degrees in a timely manner.

[0060] In order to better implement the track structure disease recognition method based on the fiber Bragg grating array in the embodiment of the present invention, correspondingly, the embodiment of the present invention also provides a track structure disease recognition device based on the fiber Bragg grating array, as Figure 6 shown, the track structure disease recognition device 600 based on the fiber Bragg grating array includes: A data detection module 601, configured to obtain the initial vibration signal of each sensing point when the train passes through the sensing points of the fiber Bragg grating array on the track, and extract the effective information from the initial vibration signal to obtain the effective vibration signal; An error estimation module 602, configured to perform sequence calculation on the effective vibration signal based on the standard basic database to obtain multiple OSA sequences, and obtain the error estimation of each sensing point according to the multiple OSA sequences; A region division module 603, configured to divide a continuous plurality of sensing points with error estimation of disease data into regions to obtain a sensing area; A result recognition module 604, configured to perform curve fitting and disease category judgment according to all the error estimations in the sensing area to obtain a disease recognition result.

[0061] The rail structure disease recognition device 600 based on the fiber Bragg grating array provided in the above embodiments can implement the technical solutions described in the above embodiments of the rail structure disease recognition method based on the fiber Bragg grating array. For the specific implementation principles of the above modules or units, reference can be made to the corresponding content in the above embodiments of the rail structure disease recognition method based on the fiber Bragg grating array, which will not be elaborated here.

[0062] The above has introduced in detail the rail structure disease recognition method and device based on the fiber Bragg grating array provided by the present invention. Specific examples are used in this article to elaborate on the principles and implementation manners of the present invention. The description of the above embodiments is only used to help understand the method and its core idea of the present invention; at the same time, for those skilled in the art, according to the idea of the present invention, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

1. A method for identifying rail structure defects based on fiber grating array, characterized in that: include: When the train passes through the sensing point of the fiber grating array on the track, the initial vibration signal of each sensing point is obtained, and effective information is extracted from the initial vibration signal to obtain an effective vibration signal; Performing sequence calculation on the effective vibration signal based on a standard basic database to obtain a plurality of OSA sequences, and obtaining an error estimate of each sensing point according to the plurality of OSA sequences; Dividing the plurality of continuous sensing points where the error is estimated to contain disease data into regions to obtain sensing regions; Curve fitting and disease category determination are performed based on all error estimates in the sensing area to obtain a disease identification result.

2. The method for identifying rail structure defects based on fiber grating array according to claim 1, characterized in that: The extracting effective information from the initial vibration signal to obtain an effective vibration signal includes: Determining all initial signal frames in the initial vibration signal according to a preset fixed length; Calculate each initial signal frame separately to obtain the corresponding short-time energy and short-time zero-crossing rate; Determine the first initial signal frame in which the short-time energy or the short-time zero-crossing rate is greater than a preset threshold among all the initial signal frames as the start frame; Determine the first initial signal frame after the start frame in all the initial signal frames, in which both the short-time energy and the short-time zero-crossing rate are not greater than the preset threshold, as the end frame; A valid vibration signal is obtained according to the start frame and the end frame.

3. The method for identifying rail structure defects based on fiber grating array according to claim 2, characterized in that: The obtaining of a valid vibration signal according to the start frame and the end frame comprises: When the frame number is greater than a preset frame number threshold, all valid signal frames between the start frame and the end frame are determined as valid vibration signals; the frame number is the number of valid signal frames between the start frame and the end frame; the valid signal frame is a signal frame whose short-time energy or the short-time zero-crossing rate is greater than the preset threshold; When the frame number is not greater than the preset frame number threshold, the first initial signal frame in which the short-time energy and the short-time zero-crossing rate after the end frame in all the initial signal frames are not greater than the preset threshold is determined as a new end frame, and a valid vibration signal is obtained based on the start frame and the new end frame.

4. The method for identifying rail structure defects based on fiber grating array according to claim 1, characterized in that: The effective vibration signal is sequenced based on the standard basic database to obtain multiple OSA sequences, including: Comparing the effective vibration signal with all standard signals in the standard basic database to determine a reference signal; Performing signal synchronization on the effective vibration signal and the reference signal in the time domain to obtain a synchronized vibration signal and a synchronized reference signal; The synchronous vibration signal and the synchronous reference signal are calculated respectively to obtain a first OSA sequence and a second OSA sequence.

5. The method for identifying rail structure defects based on fiber grating array according to claim 4, characterized in that: The error estimation includes a mean square error and a mean absolute error; and obtaining the error estimation of each sensing point according to the multiple OSA sequences includes: Calculating the first OSA sequence and the second OSA sequence to obtain a mean square error and a mean absolute error respectively; After obtaining the error estimate of each sensing point according to the multiple OSA sequences, the method further includes: Determine whether the mean square error or the mean absolute error is greater than a preset error threshold; If yes, it is determined that there is defect data in the valid vibration signal; If not, it is determined that no defect data exists in the valid vibration signal.

6. The method for identifying rail structure defects based on fiber grating array according to claim 4, characterized in that: The step of performing signal synchronization on the effective vibration signal and the reference signal in the time domain to obtain a synchronized vibration signal and a synchronized reference signal comprises: Calculating a maximum value of a cross-correlation between the effective vibration signal and the reference signal to obtain a cross-correlation peak value; Determine the delay time between the effective vibration signal and the reference signal according to the position of the cross-correlation peak; One of the effective vibration signal and the reference signal is shifted according to the delay time to obtain a synchronous vibration signal and a synchronous reference signal.

7. The method for identifying rail structure defects based on fiber grating array according to claim 1, characterized in that: The disease identification result includes the disease identification location and disease category; The curve fitting and disease category judgment are performed based on all error estimates in the sensing area to obtain a disease identification result, including: Performing curve fitting based on all error estimates in the sensing region to obtain a fitting curve; Calculating the centroid of the fitting curve to obtain the disease identification position; Based on the disease database, the disease data at the disease identification position is compared in categories to obtain the disease category.

8. The method for identifying rail structure defects based on fiber grating array according to claim 7, characterized in that: The method further comprises: Set multiple preset error intervals and corresponding disease levels; Calculating the error estimate of the disease identification position and the error estimate of the disease-free signal and the reference signal in the standard basic database to obtain a target error; Determining a target disease level corresponding to the target error according to the plurality of preset error intervals; An early warning is given to the track according to the target disease level.

9. The method for identifying rail structure defects based on fiber grating array according to claim 1, characterized in that: The method further comprises: Performing cluster analysis on the vibration signals in the sensing area to obtain the similarity of each signal frame; The signal whose similarity is greater than a preset similarity threshold is stored in the standard basic database as a standard signal.

10. A rail structure defect identification device based on fiber grating array, characterized in that: include: A data detection module is used to obtain an initial vibration signal of each sensing point when the train passes through the sensing point of the fiber grating array on the track, and extract effective information from the initial vibration signal to obtain an effective vibration signal; An error estimation module, configured to perform sequence calculation on the effective vibration signal based on a standard basic database to obtain a plurality of OSA sequences, and obtain an error estimate of each sensing point according to the plurality of OSA sequences; A region division module is used to divide the plurality of continuous sensing points where the error is estimated to have disease data into regions to obtain sensing regions; The result recognition module is used to perform curve fitting and disease category judgment according to all error estimates in the sensing area to obtain a disease recognition result.

Images