Device and method for detecting damage in the area of ​​anchor clamp of load-bearing cable based on vibration analysis

Abstract

The present invention discloses a device and method for detecting damage in the anchor wire clamp area of ​​a load-bearing cable based on vibration analysis. The detection device includes a mobile lifting unit, a claw arm unit and a signal acquisition unit. The mobile lifting unit includes a guide rail trolley that can move along the rail, a lifting platform that is slidably connected to the top of the guide rail trolley and can move in a direction perpendicular to the rail, and a pillar vertically installed on the lifting platform. The claw arm unit includes a mounting platform arranged at the top of the pillar, a claw assembly arranged on the mounting platform for clamping the cross beams at both ends of the anchor wire clamp, and two sets of rotating mechanical arms. The signal acquisition unit includes a magnetic vibration sensor and an exciter, an acquisition card and a computer. The anchor wire clamp area is detected using the aforementioned device for detecting damage in the anchor wire clamp area of ​​a load-bearing cable based on vibration analysis. The use of the detection device and method can realize non-destructive, real-time and efficient detection of damage in the anchor wire clamp area of ​​railway load-bearing cables.

Description

Technical Field

[0001] The present invention belongs to the field of railway detection equipment, relates to the diagnosis of defects in railway contact network catenary cables, and specifically to a device and method for detecting damage in the anchor clamp area of ​​the catenary cable based on vibration analysis. Background Art

[0002] The load-bearing cable is a single-core multi-layer stranded wire. It is an important conductor that performs the dual functions of power transmission and locomotive suspending slideway lines. It is necessary to ensure that the load-bearing cable has little deformation and uniform sag after installation, and can withstand the vibration transmitted by the locomotive and the external force loads of the natural environment such as wind, ice, rain and snow. Therefore, during use, some stranded wires will suffer fatigue fracture due to the alternating load of tension changes.

[0003] With the vigorous promotion of high-speed rail construction and the continuous increase in the speed of high-speed rail operation, the requirements for the stability of the tension of the catenary cables are getting higher and higher. If the stranded wire of the catenary cable breaks due to fatigue, the overall risk of wire breakage will be greatly increased. The two broken ends or one of the two ends will relax to the lower part of the contact line or even fall to the ground. On the one hand, it will directly cause the contact network to short-circuit and discharge to the ground, burning the contact line or rails; on the other hand, part of the suspension string at the broken part of the catenary cable will lose strength and relax to the lower part of the contact line with the broken part of the catenary cable. If it is not discovered in time by the locomotive crew, it will not only cause the contact network to short-circuit and discharge the locomotive vehicle, expanding the scope of the accident and endangering personal safety, but also cause a larger bow-snapping accident, greatly reducing the safety of railway operation. According to reports, in recent years, there have been many railway operation failures caused by structural damage to the catenary cable in my country alone, and a considerable part of the accidents were caused by damage to the stranded wire area covered by the anchor wire clamp of the catenary cable. When the anchor wire clamp area of ​​the catenary cable is damaged, it needs to be reinforced, connected or replaced according to the degree of cross-sectional damage. Among them, effective detection and determination of the damage in the wire clamp area is a prerequisite. Therefore, it is of great significance to ensure the safe and stable operation of high-speed railways to study and design a railway contact network cable anchor wire clamp area wire break monitoring technology based on abnormal vibration signal diagnosis, formulate a cable strand maintenance plan, regularly conduct all-round inspections of the contact network cable, and take effective protective measures according to the inspection results.

[0004] At present, the operation status of the catenary cables is generally detected by manual inspection or tension detection equipment at home and abroad. The method of relying solely on manual detection requires manual climbing to the height of the contact network, opening the cable seat, destroying the cable structure to check whether there is a broken wire in the area with the anchor wire clamp, and the reinstallation operation after inspection and replacement is cumbersome, and there are problems such as low efficiency and high randomness; using the existing domestically introduced tension monitoring equipment for detection, although the detection speed has been improved to a certain extent, the detection error is large, the price is expensive, and due to operation restrictions, it can generally only be carried out during the skylight time of railway operation, lacking continuity in time. The utility model patent with patent application number 201320854397.1 discloses a contact network tension measurement system, including a measuring sensor device for measuring the contact network tension, the measuring sensor device is installed at the contact wire and the anchor position of the contact wire of the contact network, the measuring sensor device includes an insulating part and a connecting rod connected to the contact wire and the catenary cable in sequence in the direction toward the anchor position, a wire clamp frame is provided in the connecting rod, and a measuring sensor is provided at the position where the wire clamp frame is connected to the connecting rod. Since the measuring sensor device is installed between the anchor position and the contact wire rope, it is necessary to disconnect the contact wire to connect the measuring sensor device, which destroys the original contact network structure and is inconvenient to install. Moreover, if the measuring sensor device is of poor quality, it is easy to cause a major safety accident. The invention patent with patent application number 201810414256.5 discloses a system and method for vibration and current-carrying tests on contact wires, including: a vibration device and an electric power supply device determined according to the on-site working conditions; the contact wire is erected in the erection device; a vibration device for simulating vibration of the contact wire for vibration testing; and an electric power supply device for loading current onto the contact wire for current-carrying tests. The installation method is cumbersome and fixed, and the installation and inspection costs of long-line load-bearing cables are extremely high, making it difficult to achieve regular inspections of the contact wires.

[0005] In summary, the existing traditional contact network catenary cable operating status detection technology and solutions have a series of technical pain points. Their detection efficiency and accuracy cannot meet the growing detection needs year by year. It is urgent to research and develop a method and system for detecting broken wires in the anchor wire clamp area of ​​the catenary cable without destroying the original structure of the catenary cable, so as to improve the efficiency and accuracy of the catenary cable status detection and shorten the "window time" of railway operation. Summary of the invention

[0006] In view of the problems in the current technology for detecting the degree of wear in the anchor clamp area of ​​railway load-bearing cables, such as slow manual detection speed, low degree of automation, and inability to accurately detect damage in the anchor clamp area of ​​railway load-bearing cables in a complex multi-climate environment, the purpose of the present invention is to provide a device and method for detecting damage in the anchor clamp area of ​​railway load-bearing cables based on vibration analysis, which can realize rapid and accurate detection of the degree of wear in the anchor clamp area of ​​railway load-bearing cables caused by unavoidable reasons such as external forces during construction, water leakage in the tunnel above, and short-circuit current generated by overlapping foreign objects.

[0007] The invention idea of ​​the present invention is: according to the railway industry standard TB / T3111-2005 of the People's Republic of China, the material of the catenary stranded wire used for 50HZ, 25kV electrified railway contact network is magnesium copper alloy, composed of 19 cores, and the average vibration life is 2×10 6 times, fatigue life 5×10 5 times. Through the structural analysis of the anchor clamp area of ​​the railway catenary, it was found that the structural shapes of the anchor clamp area of ​​the catenary are different, and the pressure exerted on the catenary in different areas of the anchor clamp varies greatly, making it difficult to effectively apply visual, infrared, and electromagnetic damage detection methods. Through vibration detection analysis, it was found that by using the active acoustic emission method, an electromagnetic exciter is installed on one side of the catenary anchor clamp to generate a controllable excitation signal to drive the catenary to vibrate, and a magnetic vibration sensor is installed on the other side to receive the signal, and the excitation signal energy is adjusted according to the transmission loss. This can effectively obtain damage information and detect the damage status of the catenary in the anchor clamp area.

[0008] Based on the above invention ideas, the present invention provides a device for detecting damage in the area of ​​a load-bearing cable anchor clamp based on vibration analysis, comprising a mobile lifting unit, a claw arm unit and a signal acquisition unit;

[0009] The mobile lifting unit comprises a guide rail trolley movable along the rail, a lifting platform slidably connected to the top of the guide rail trolley and movable in a direction perpendicular to the rail, and a support vertically mounted on the lifting platform;

[0010] The clamp arm unit includes a mounting platform arranged at the top of the pillar, a clamp assembly arranged on the mounting platform for clamping the cross beams at both ends of the anchor wire clamp, and two sets of rotating mechanical arms symmetrically arranged on both sides of the mounting platform and having the same structure. The end surface of the rotating mechanical arm with the mounting platform as the center is provided with a clamping slot I, a support clamp and a clamping slot II in sequence from the outside to the inside;

[0011] The signal acquisition unit includes a magnetic vibration sensor and a vibrator respectively installed in the card slot I and the card slot II, an acquisition card installed on the installation platform, and a computer for preprocessing the received vibration signal and identifying damage in the area of ​​the anchor clamp of the load-bearing cable. The computer is installed on the mobile lifting unit, and the two ends of the acquisition card are respectively connected to the magnetic vibration sensor and the computer;

[0012] When the rotary manipulator arm rotates to the working position, the support clamp clamps the load-bearing cable, and the exciter and the magnetic vibration sensor are respectively located under the load-bearing cable and the anchor wire clamp.

[0013] The above-mentioned damage detection device for the anchor wire clamp area of ​​the catenary cable based on vibration analysis is based on the law of conservation of energy. During the transmission of the vibration signal, the larger the resonance area generated by the system vibration, the faster the vibration energy dissipates, and the higher the signal discreteness and mixing degree. The study found that the simpler the rigid system structure, the fewer the resonance peaks of the system, so the energy dissipation is slower, the signal discreteness is low, the signal-to-noise ratio is high, and it is easier to analyze and identify the signal. Based on the above analysis, the present invention adopts a claw arm unit for rigid connection of the upper end of the catenary cable anchor wire clamp area, and adopts a lifting guide rail trolley to move along the railway track at the lower end to perform target alignment on the catenary cable anchor wire clamp area of ​​the railway contact network. The upper rigid connection is achieved by clamping the catenary cable with a support clamp and positioning the claw arm unit of the crossbeams at both ends of the catenary cable anchor wire clamp with a claw assembly. In this process, the V-shaped mechanical claw provides an upward support force to counteract the downward clamping force provided by the catenary cable anchor wire clamp, and the tension of the catenary cable can be regarded as a rigid connection. The claw arm unit adopts a rigid structure and is fixed to the pillar through a rigid connection. The connection is designed with a lifting platform base for installing a vibration signal acquisition card. The claw arm unit, the load-bearing cable anchor wire clamp and the load-bearing cable constitute a vibration system. The rigid structure can absorb the vibration signal of the vibration system without generating a large resonance. Based on the certain flexibility of the load-bearing cable itself, the rotating mechanical arm can apply an upward thrust to the load-bearing cable. When the load-bearing cable in the anchor wire clamp area is applied with an upward thrust, the load-bearing cable forms a better rigid structure under the dual constraints of the wire clamp and the mechanical arm. Furthermore, the rigid structure preferably adopts a high-hardness material, such as steel, silicon steel sheet or ferrite material.

[0014] The above-mentioned load-bearing cable anchor clamp area damage detection device based on vibration analysis, the present invention adopts a magnetic vibration sensor, which can prevent the vibration of the exciter installed on the rotating mechanical arm from interfering with the magnetic vibration sensor during operation. The function of the rotating mechanical arm is to drive the exciter and the magnetic vibration sensor to the working position through rotational motion. Therefore, on the basis of realizing rotational motion, the rotating mechanical arm can adopt the mechanical arm structure conventionally used in this field. In the present invention, the rotating mechanical arm adopts a motor drive mechanism, which relies on the motor to rotate the mechanical arm at a certain angle. In order to facilitate the installation of the exciter and the magnetic vibration sensor, the end of the rotating mechanical arm is preferably a flat structure, and a support seat is provided on the end surface of the rotating mechanical arm, and the support clamp and the card slot II are provided on the support seat.

[0015] When the mechanical arm rotates to the working position, the vibrator and the magnetic vibration sensor are respectively located below the load-bearing cable and below the cable clamp. When working, the magnetic vibration sensor is energized and separated from the mechanical arm by magnetic force to be adsorbed on the lower end of the load-bearing cable anchor cable clamp. The lower end of the load-bearing cable and the lower end of the load-bearing cable anchor cable clamp are tightly fitted under the constraint, so the vibration signal generated by the vibrator on the load-bearing cable can be used for indirect measurement. The claw arm unit adopts a two-end symmetrical design, and the load-bearing cable anchor cable clamp area to be detected is fixed to the symmetry center of the rotating mechanical arm at both ends. For the vibrator and the magnetic vibration sensor on the same side mechanical arm, in order to ensure that the received vibration signal passes through the damaged area of ​​the load-bearing cable and prevent the vibration of the vibrator on the same side from causing signal interference to the magnetic vibration sensor, the present invention designs two sets of rotating mechanical arms, that is, a detection method in which the vibrator of one end mechanical arm generates a vibration excitation signal and the magnetic vibration sensor of the other end mechanical arm receives the vibration signal. Finally, the stress wave generated by the vibration passes through the damaged area, and the damage information of the load-bearing cable contained in the signal acquisition card is transmitted to the computer. Furthermore, in order to fix the magnetic vibration sensor in the mechanical arm slot, a magnet for attracting the magnetic vibration sensor is provided at the slot opening or the bottom of the slot II.

[0016] In the above-mentioned vibration analysis-based regional damage detection device for the anchor wire clamp of the load-bearing cable, the function of the claw assembly is to position and clamp the crossbeams at both ends of the anchor wire clamp so that the lower end of the load-bearing cable and the lower end of the load-bearing cable anchor wire clamp fit closely under constraint. The claw assembly can adopt a positioning clamping structure commonly used in the art. In the present invention, the claw assembly preferably includes two hooks that are inverted and installed on the mounting platform and have the same opening direction, and the hook portion of the hook is in a U-shaped structure. During operation, the guide rail trolley and the lifting platform are used to align the hook and the area to be tested. By controlling the horizontal movement of the guide rail trolley and the lifting platform and adjusting the height of the lifting platform, it is ensured that the hooks at both ends of the anchor wire clamp are tightly attached to the cantilever beams at both ends of the anchor wire clamp, thereby adjusting the height of the lifting platform to clamp the cantilever beam. After the detection is completed, the height of the lifting platform is adjusted to allow the hook to relax the cantilever beam, and then the horizontal movement of the guide rail trolley and the lifting platform is controlled to make the hooks unobstructed below the hooks, and then the lifting platform is controlled to descend back to the original position.

[0017] In the above-mentioned vibration analysis-based damage detection device for the anchoring wire clamp area of ​​the catenary, the function of the support clamp is to apply an upward thrust to the catenary and clamp the catenary. When the catenary in the anchoring wire clamp area is applied with an upward thrust, the catenary anchor can form a better rigid structure under the dual constraints of the wire clamp and the mechanical arm. On the basis of being able to apply an upward thrust to the catenary and clamp the catenary, the structure of the support clamp can be designed according to the conventional method. In the present invention, the support clamp is preferably in a V-shaped structure.

[0018] The above-mentioned vibration analysis-based damage detection device for the load-bearing cable anchor clamp area, the detection device is accurately aligned with the target in the load-bearing cable anchor clamp area by moving the lifting unit (guide rail trolley and lifting platform). After the detection device is transported as a whole to the vicinity of the load-bearing cable anchor clamp area to be tested by the guide rail trolley, on the one hand, the release and locking of the cantilever beam by the claw assemblies at both ends of the anchor clamp are controlled by adjusting the height of the lifting platform, and on the other hand, the lifting platform is moved in a direction perpendicular to the rail until the symmetry center of the rotating mechanical arms at both ends is aligned with the load-bearing cable anchor clamp area. The guide rail trolley can move along the rails, mainly to carry other components and move the entire device. There is no special restriction on the specific structure of the guide rail trolley, and it can be designed according to the conventional structure. The guide rail trolley described in the present invention includes a car body and rail wheels installed on the car body. Furthermore, the sliding connection between the lifting platform and the guide rail trolley can adopt the sliding connection method commonly used in the art. In the present invention, the upper surface of the vehicle body is preferably installed with a guide rail assembly for controlling the guided movement of the lifting platform, and the bottom of the lifting platform is connected to the guide rail assembly. The guide rail assembly includes two screw guide rails installed on the upper surface of the vehicle body and perpendicular to the railway track, a feed screw rod I installed between the two screw guide rails, and a handwheel I connected to one end of the feed screw rod. The bottom of the lifting platform is connected to the screw guide rail and the feed screw rod I, and the lifting platform performs a feeding motion under the rotation of the handwheel I, pushing the lifting platform to move along the screw guide rail.

[0019] Furthermore, the main function of the lifting platform is to adjust the height and horizontal position of the claw arm unit. The adjustment of the horizontal position is achieved by slidingly connecting with the top of the guide rail trolley. The lifting platform can adopt conventional lifting mechanisms in the field, such as hydraulic lifting platforms, screw lifting platforms, etc. In the present invention, the lifting platform adopts a stepping screw mechanism. Specifically, the lifting platform is preferably a screw lifting platform. The screw lifting platform uses a screw lifter screw to drive a scissor mechanism for lifting and lowering. It has the characteristics of a hydraulic lifting platform and is free of oil pollution. It is particularly suitable for workplaces with high requirements for the working environment. The screw lifting platform includes a lifting platform base and a lifting platform table arranged in parallel, a scissor bracket arranged between the lifting platform base and the lifting platform table for maintaining the height of the lifting platform table, and a screw lift connected to the scissor bracket. One end of the feed screw II of the screw lift is connected to a handwheel II, and the lifting platform base is slidably connected to the top of the guide rail trolley. The feed screw II performs a feeding motion under the rotation of the handwheel II, pushing the scissor bracket of the lifting platform to rotate around the bracket shaft, thereby realizing the lifting motion of the lifting platform.

[0020] In the above-mentioned vibration analysis-based load-bearing cable anchor clamp area damage detection device, the computer is connected to the vibration signal acquisition card, and is used to pre-process the received vibration signal, and extract features based on the IMF component of the vibration signal extracted by time domain analysis, frequency domain analysis, wavelet packet decomposition and EMD, and identify the damage of the railway load-bearing cable anchor clamp area based on the obtained characteristic coefficient combined with neural network technology analysis. The computer can be installed on a guide rail trolley or a lifting platform, preferably installed on the lifting platform, located on one side of the pillar.

[0021] The present invention also provides a method for detecting damage to the anchor wire clamp area of ​​a catenary cable based on vibration analysis. The above-mentioned device for detecting damage to the anchor wire clamp area of ​​a catenary cable based on vibration analysis is used. After the claw arm unit at the upper end of the device clamps the catenary cable at the position to be tested and positions and clamps the crossbeams at both ends of the anchor wire clamp of the catenary cable, the rotary mechanical arm rotates to the working position. At this time, the vibrator and the magnetic vibration sensor are respectively located below the catenary cable and below the wire clamp. The vibrator on one set of rotary mechanical arms works to rigidly collide with the catenary cable to generate an excitation signal, and the magnetic vibration sensor on the other set of rotary mechanical arms is adsorbed below the anchor wire clamp. The claw arm unit, the catenary cable anchor wire clamp and the catenary cable constitute a vibration system. The vibration signals generated by the vibration system and the vibrator are collected by the magnetic vibration sensor and transmitted to the computer through the vibration signal acquisition card. During the signal acquisition process, the vibrators and magnetic vibration sensors on the two sets of rotary mechanical arms work alternately. The computer preprocesses the received vibration signal, and extracts features based on the IMF component of the vibration signal extracted by time domain analysis, frequency domain analysis, wavelet packet decomposition and EMD, and realizes the identification of the damage state of the railway load-bearing cable anchor clamp area based on the obtained characteristic coefficient combined with neural network technology, and determines the damage category and damage degree. The damage categories involved in the present invention include cracks, wear, broken strands, etc., and the damage degree includes crack depth, wear area, number of broken strands, etc.

[0022] Specifically, the method for detecting damage to the anchor clamp area of ​​a load-bearing cable based on vibration analysis comprises the following steps:

[0023] S1. The guide rail trolley moves to the area below the anchor wire clamp of the load-bearing cable. The lifting platform is adjusted so that the claw assembly clamps the beams at both ends of the anchor wire clamp. The two sets of rotary mechanical arms rotate to the support clamps to clamp the load-bearing cable. At the same time, the exciter and the magnetic vibration sensor enter the working position. The exciter contacts the load-bearing cable. After the magnetic vibration sensor is powered on, it is adsorbed under the anchor wire clamp by magnetic force.

[0024] S2, a vibration exciter on one set of rotating mechanical arms generates a vibration excitation signal, a magnetic vibration sensor on another set of rotating mechanical arms receives the vibration signal, and an acquisition card receives the vibration signal and transmits it to a computer;

[0025] S3. The computer performs preprocessing, time domain analysis, frequency domain analysis, wavelet packet decomposition and EMD extraction of the received vibration signal IMF component, and builds a damage collaborative identification model to identify the damage status of the load-bearing cable anchor clamp area.

[0026] In order to improve the detection accuracy of the anchor cable clamp area damage of the above-mentioned vibration analysis-based method, the present invention further adopts a feedback optimization mechanism: First, by setting the feature evaluation factor method, the vector feature selection criteria under different environments are determined; Second, the optimal recognition feature reconstruction is performed by using feature vector clustering and real-time detection information feedback after signal extraction. Specifically, the detection method also includes the following steps:

[0027] S4. Reconstruct features by at least one of feature selection, feature vector clustering and real-time detection information feedback based on setting feature evaluation factors, and then input the reconstructed features into the established damage collaborative identification model to re-identify the damage status of the load-bearing cable anchor clamp area.

[0028] In the above-mentioned method for detecting damage in the area of ​​the anchor clamp of the catenary based on vibration analysis, the step S1 comprises the following sub-steps:

[0029] S11, push the guide rail trolley to move to the vicinity of the anchor wire clamp area of ​​the load-bearing cable to be tested, raise the lifting platform, and align the claw arm unit 6 and the area to be tested through the guide rail assembly;

[0030] S12, lower the lifting platform, ensure that the claw assembly is tightly attached to the cantilever beams at both ends of the anchor wire clamp, rotate the rotary mechanical arm to the support clamp to clamp the load-bearing cable, and the exciter and magnetic vibration sensor enter the working position;

[0031] S13, after the magnetic vibration sensor is powered on, it is adsorbed under the anchor wire clamp by magnetic force;

[0032] In the above-mentioned method for detecting damage in the area of ​​the anchor clamp of the catenary based on vibration analysis, the step S2 comprises the following sub-steps:

[0033] S21, controlling the vibration frequency and pressure of the vibrator, as well as the excitation time and the signal receiving time of the magnetic vibration sensor through the MCU control board;

[0034] S22, controlling the vibrator on one set of rotary mechanical arms to generate a vibration excitation signal through the MCU control board, and the magnetic vibration sensor on the other end of the rotary mechanical arm to receive the vibration signal, and the acquisition card receives and transmits the signal to the computer;

[0035] S23, in the above step S22, the vibrators and the working ends of the magnetic vibration sensors on the two sets of rotary mechanical arms are exchanged, and the acquisition card receives and transmits the signal to the computer;

[0036] S24. Repeat the above steps S21, S22, and S23 to collect a set number of vibration signals for damage detection of the load-bearing cable anchor clamp area.

[0037] In the above-mentioned method for detecting damage in the area of ​​the anchor clamp of the catenary based on vibration analysis, the step S3 comprises the following sub-steps:

[0038] S31, pre-emphasize the collected vibration signal and obtain frequency spectrum information corresponding to the vibration signal through discrete fast Fourier transform FFT;

[0039] S32, time domain analysis, selecting the time domain peak value, crest factor, energy value, and standard deviation of the vibration signal as time domain features x1, x2, x3, and x4;

[0040] S33, frequency domain analysis, select spectrum peak and crest factor as frequency domain features: x5, x6;

[0041] S34, wavelet packet decomposition features, wavelet packet db05 is decomposed into three layers of wavelet packets, and the energy of the third layer of wavelet packet reconstruction coefficients is selected as the wavelet packet decomposition features x7~x 14 ;

[0042] S35, EMD feature extraction, use EMD to decompose the active acoustic emission detection signal sample, determine the number of IMF components, and extract the two IMF energies with the largest proportion of the total energy as the EMD feature x 15 , x 16 ;

[0043] S36. Build a multi-layer perceptron network as a damage collaborative identification model, identify the extracted features through the damage collaborative identification model, and preliminarily determine the damage status of the load-bearing cable anchor clamp area.

[0044] The damage state includes damage category and damage degree. The damage category includes but is not limited to common damage categories in the field such as cracks, wear, and broken strands. The damage degree includes but is not limited to common damage degrees in the field such as crack depth, wear area, and number of broken strands.

[0045] In the above-mentioned method for detecting damage in the area of ​​the anchor clamp of the load-bearing cable based on vibration analysis, in the step S4, the feature reconstruction method includes:

[0046] (1) By setting the feature evaluation factor method, the termination condition of the feature value is determined, and the feature signal is determined according to the termination condition, thereby obtaining the reconstructed feature;

[0047] (2) K-means cluster analysis is used to reconstruct the damage state vector and obtain the reconstructed features;

[0048] (3) Use the superposition method of online detection signals to correct the damage state vector and obtain the reconstructed features.

[0049] Through the railway load-bearing cable anchor clamp area damage detection method based on vibration analysis provided by the present invention, the collected vibration signal is preprocessed, analyzed in the time domain, analyzed in the frequency domain, and subjected to wavelet packet decomposition and EMD extraction of the vibration signal IMF component for feature extraction, and a damage collaborative identification model is built to realize the railway load-bearing cable anchor clamp area damage detection. From positioning and clamping the load-bearing cable anchor clamp area to be tested to the completion of defect identification, the manual detection time is greatly shortened and the detection efficiency is improved. The present invention further introduces a feedback optimization mechanism, which adjusts the environmental adaptability of the detection device by setting feature evaluation factors to achieve specified specific vector feature selection criteria under different environments; through the method of feature vector clustering and real-time detection information feedback, feedback adjustment parameters are introduced to achieve optimal reconstruction of feature vectors.

[0050] Compared with the prior art, the device and method for detecting damage in the area of ​​the anchoring wire clamp of the load-bearing cable based on vibration analysis provided by the present invention have the following beneficial effects:

[0051] 1. The damage detection device for the anchor wire clamp area of ​​the load-bearing cable based on vibration analysis provided by the present invention firstly uses the rotary mechanical arm in the claw arm unit to clamp the load-bearing cable from bottom to top, and the claw assembly positions and clamps the cross beams at both ends of the load-bearing cable anchor wire clamp to form a rigid connection fixing mode; then the exciter on one group of rotary mechanical arms generates a vibration excitation signal, and the magnetic vibration sensor on the other group of rotary mechanical arms receives the vibration signal, and the acquisition card receives and transmits the signal to the computer; the computer then processes the collected vibration signal, performs pre-processing, time domain analysis, frequency domain analysis, wavelet packet decomposition and EMD extraction of the vibration signal IMF component for feature extraction, and builds a damage collaborative identification model to identify the extracted features, from positioning and clamping the load-bearing cable anchor wire clamp area to be tested to defect identification, which greatly shortens the manual inspection time, improves the inspection efficiency, and realizes non-destructive, real-time and efficient detection of damage in the railway load-bearing cable anchor wire clamp area.

[0052] 2. The device for detecting damage to the anchoring wire clamp area of ​​the catenary cable based on vibration analysis provided by the present invention is provided with two groups of mechanical arms, wherein the exciter on one group of the mechanical arms generates a vibration excitation signal, the magnetic vibration sensor on the other group of the mechanical arms receives the vibration signal, and the acquisition card receives and transmits the signal to the computer; the claw arm unit, the catenary cable anchoring wire clamp and the catenary cable constitute a rigid vibration system, and the rigid system has few resonance peaks, low signal discreteness and high signal-to-noise ratio, so that the wear information of the catenary cable contained in the damaged area when the stress wave generated by the vibration is almost not disturbed, thereby realizing the damage detection of the anchoring wire clamp area of ​​the railway catenary cable, filling the gap in the damage detection technology of the anchoring wire clamp area of ​​the railway catenary cable, and having great application prospects in the field of railway catenary cable detection technology.

[0053] 3. The method for detecting damage to the anchor cable clamp area based on vibration analysis provided by the present invention comprises the following steps: a computer pre-processes the received vibration signal, and extracts features based on the IMF component of the vibration signal extracted by time domain analysis, frequency domain analysis, wavelet packet decomposition and EMD, and identifies the damage to the anchor cable clamp area of ​​the railway according to the obtained characteristic coefficients combined with the neural network technology analysis. Therefore, since the present invention is a vibration signal analysis and identification method based on time domain analysis, frequency domain analysis, wavelet packet decomposition characteristics and EMD characteristics, it is different from the limitations of a single feature identification method. By combining the intuitive response capabilities of time domain characteristics and frequency domain characteristics and the suitability of wavelet packet decomposition and empirical mode decomposition (EMD) for analyzing non-stationary signals, it can make up for the missing damage information between various features, and ensure that the damage detection process of the anchor cable clamp area is more comprehensive and precise.

[0054] 4. The damage detection method for the anchor clamp area of ​​the catenary cable based on vibration analysis provided by the present invention introduces a feedback optimization mechanism. By setting the feature evaluation factor, the specified specific vector feature selection criteria in different environments are realized, which greatly improves the environmental adaptability of the detection device. By means of feature vector clustering and real-time detection information feedback, feedback adjustment parameters are introduced to realize the optimal reconstruction of the feature vector. The detection of different damage types of the wear of the anchor clamp area of ​​the railway catenary cable in different environments can be realized (different environments include icing, high temperature, extreme cold, etc., and different damage types include catenary cable wear, catenary cable breakage, catenary cable cracks, etc.), which is extremely suitable for efficient detection of damage to the anchor clamp area of ​​the railway catenary cable in various working environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 Schematic diagram of the damage detection device for the anchor clamp area of ​​the load-bearing cable based on vibration analysis.

[0056] Figure 2 for Figure 1 Schematic diagram of middle guide rail trolley

[0057] Figure 3 for Figure 1 Schematic diagram of the middle lifting platform.

[0058] Figure 4 for Figure 1 Schematic diagram of the middle jaw arm unit.

[0059] Figure 5 The diagrams are before and after the claw assembly positions the area to be measured, where (a) is a diagram showing the position calibration before the hook is lowered, and (b) is a diagram showing the hooks at both ends after they are lowered.

[0060] Figure 6 This is a schematic diagram of the claw arm unit entering the working position.

[0061] Figure 7 This is a schematic diagram of the magnetic vibration sensor being powered and adsorbed under the wiring clamp.

[0062] Figure 8 This is a schematic diagram of the load-bearing cable anchor clamp area damage detection device based on vibration analysis when detecting the area to be tested.

[0063] Fig. 9 The preprocessed vibration signal diagram collected by the magnetic vibration sensor, where (a) is the time domain diagram of the vibration signal and (b) is the frequency domain diagram of the vibration signal.

[0064] Fig.10 The feature diagram extracted by wavelet packet decomposition of vibration signal, where (a) is the dendrogram of db5 wavelet packet decomposition, (b) is the spectrum diagram of vibration signal of each node after wavelet packet decomposition, and (c) is the energy proportion diagram of each frequency band of vibration signal after wavelet packet decomposition.

[0065] Fig.11 This is the time domain diagram of the intrinsic mode function component (IMF component) after the empirical mode decomposition (EMD) of the vibration signal.

[0066] Fig.12 A multi-layer perceptron network structure diagram is constructed for the present invention.

[0067] Explanation of the reference numerals: 1. ground contact surface; 11. railway track bed; 12. rails; 2. guide rail trolley; 21. vehicle body; 22. rail wheels; 231. screw guide rail I; 232. screw guide rail II; 241. hand wheel I; 242. feed screw I; 3. lifting platform; 311. lifting platform base; 312. lifting platform table; 321. scissors bracket; 322. bracket coupling; 33. hand wheel II; 34. feed screw II; 4. computer; 5. pillar; 6. claw arm unit; 61. mounting platform; 62. rotary robot arm; 63. magnetic vibration sensor; 64. exciter; 65. hook; 661. bolt; 662. nut; 7. anchor wire clamp; 8. load-bearing cable; 9. crossbeam. DETAILED DESCRIPTION

[0068] The technical solutions of the embodiments of the present invention are clearly and completely described in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work belong to the present invention.

[0069] Example 1

[0070] The present embodiment provides a device for detecting damage to the area of ​​the load-bearing cable anchor clamp 7 based on vibration analysis, such as Figure 1As shown, it includes a mobile lifting unit, a claw arm unit 6 and a signal acquisition unit.

[0071] like Figure 1 and Figure 2 As shown, the mobile lifting unit includes a guide rail trolley 2, a lifting platform 3 and a support 5 of an integrated structure. The guide rail trolley 2 includes a car body 21 and a rail wheel 22 installed on the car body 21. The ground contact surface is paved with a railway track bed 11 and a rail 12, and the car body 21 runs on the rail through the rail wheel 22. A guide rail assembly for controlling the guided movement of the lifting platform 3 is installed on the upper surface of the car body 21. The guide rail assembly includes two screw guides (screw guide Ⅰ231 and screw guide Ⅱ232) and a feed screw Ⅰ242. The two screw guides are installed on the upper surface of the car body 21 and are perpendicular to the railway track. The feed screw Ⅰ242 is installed between the two screw guides, and one end of the feed screw is connected to a handwheel Ⅰ241. The lifting platform 3 is driven to feed forward and backward by rotating the handwheel Ⅰ241, which is mainly used to align the two ends of the area of ​​the anchor wire clamp 7 of the load-bearing cable. The maintenance worker can control the guide trolley 2 to move on the rails and shake the hand wheel Ⅰ241 to control the lifting platform 3 to move in a direction perpendicular to the rails, thereby controlling the lifting platform 3 to the projection position of the railway load-bearing cable anchor clamp 7 on the ground.

[0072] like Figure 1 and Figure 3 As shown, the lifting platform 3 is a stepping screw mechanism, including a lifting platform base 311 and a lifting platform surface 312 arranged in parallel, a scissor bracket 321 arranged between the lifting platform surface 312 and the lifting platform base 311 for maintaining the height of the lifting platform surface 312, and a screw elevator connected to the scissor bracket 321. The pillar 5 is vertically arranged on the lifting platform 3. There is a bracket coupling 322 at the intersection of the scissor bracket 321, and the screw elevator includes a handwheel Ⅱ33 and a feed screw Ⅱ34. It is connected through the screw guide rails arranged on the lifting platform base 311 and the vehicle body 21. The feed screw Ⅱ34 performs a feeding motion under the rotation of the handwheel Ⅱ33, pushing the scissor bracket 321 to rotate around the bracket coupling 322, realizing the lifting motion of the lifting platform 3, thereby accurately controlling the distance between the lifting platform 3 and the equipment thereon and the ground, that is, the position in the vertical direction. As shown Figure 5 As shown in (a), during the test, the hand wheel II 33 is turned clockwise, and the feed screw II 34 performs a feeding motion under the rotation of the hand wheel II 33, pushing the scissor brackets 321 on both sides to rotate around the bracket connecting shaft 322, and the lifting platform 3 rises upward, and then the height of the lifting platform 3 is adjusted in conjunction with the support 5 and the claw arm unit 6, so that the claw assembly (the hook 65 in this embodiment) is locked on the cantilever beam 9; Figure 5As shown in (b), after the test is completed, the hand wheel II 33 is first turned clockwise to loosen the hook 65 on the cantilever beam 9, and then the horizontal movement of the guide trolley 2 and the lifting platform 3 is controlled to make the bottom of the hook 65 clear, and then the hand wheel II 33 is turned counterclockwise to lower the lifting platform 3 back to its original position.

[0073] like Figure 1 and Figure 4 As shown, the claw arm unit 6 adopts a two-end symmetrical design, and the area of ​​the load-bearing cable anchor wire clamp 7 to be detected is fixed at the symmetry center of the two-end mechanical arms. The claw arm unit 6 includes a mounting platform 61, a claw assembly and two sets of rotating mechanical arms 62. The mounting platform 61 is set at the top of the pillar 5. The claw assembly is used to clamp the cross beams 9 at both ends of the anchor wire clamp 7, including two hooks 65 invertedly installed on the mounting platform 61 and with the same opening direction, and the hook part of the hook 65 is a U-shaped structure. The two sets of rotating mechanical arms 62 are symmetrically arranged on both sides of the mounting platform 61 and have the same structure. The rotating mechanical arm 62 is driven to rotate by a motor. The end of the rotating mechanical arm 62 is a flat plate structure, and a support seat is arranged on its end surface. The end surface of the rotating mechanical arm 62 with the mounting platform 61 as the center is sequentially provided with a slot I, a support clamp and a slot II from the outside to the inside. The support clamp and the slot II are located on the support seat, and a magnet for attracting the magnetic vibration sensor 63 is arranged at the bottom of the slot II. The support clamp is a V-shaped structure. In this embodiment, the claw arm unit 6 is a rigid structure, fixed to the pillar 5 by a rigid connection, and the claw arm unit 6, the load-bearing cable anchor clamp 7 and the load-bearing cable 8 constitute a vibration system. This rigid structure made of high-hardness materials such as steel, silicon steel sheet or ferrite material can absorb the vibration signal of the vibration system without generating a large resonance. Figure 6 As shown, during the test, the rotating mechanical arm 62 is used to apply an upward thrust to the catenary 8. Since the catenary 8 itself also has a certain flexibility, when the catenary 8 in the area of ​​the anchor wire clamp 7 is applied with an upward thrust, the catenary 8 can form such a good rigid structure under the dual constraints of the wire clamp and the mechanical arm. When working, the mechanical arm rotates at a certain angle, such as Figure 7 As shown, when it is turned to the working position, the support clamp clamps the load-bearing cable 8, and the magnetic vibration sensor 63 is energized and separated from the mechanical arm by magnetic force to be adsorbed on the lower end of the load-bearing cable anchor clamp 7. The lower end of the load-bearing cable 8 and the lower end of the anchor clamp 7 are tightly fitted under constraint.

[0074] The signal acquisition unit includes a magnetic vibration sensor 63, an exciter 64, an acquisition card and a computer 4. The exciter 64 is installed in the card slot I by means of a bolt 661 and a nut 662. The magnetic vibration sensor 63 is adsorbed in the card slot II by means of a magnet. The acquisition card is built into the mounting platform 61. The computer 4 is installed on the lifting platform 3 and is located on one side of the pillar 5. One end of the acquisition card is connected to the magnetic vibration sensor 63, and the other end is connected to the computer 4 along the pillar 5. The computer 4 is connected to the acquisition card, and the computer 4 is used to process and extract features of the received vibration signal, and to build a damage collaborative identification model to identify the damage status of the load-bearing cable anchor clamp 7 area.

[0075] In this embodiment, the mechanical arm in the claw arm unit 6 is used to clamp the catenary cable 8 from bottom to top, and the claw assembly is positioned to clamp the cross beams 9 at both ends of the catenary cable anchor wire clamp 7, so as to form a rigid connection fixing mode, and then the exciter 64 on one group of rotating mechanical arms 62 generates a vibration excitation signal, and the magnetic vibration sensor 63 on the other group of rotating mechanical arms 62 receives the vibration signal, and finally the vibration generates a stress wave passing through the damaged area, and the signal and the damage information of the catenary cable 8 contained therein are transmitted to the computer 4 by the signal acquisition card. This rigid system has few resonance peaks, low signal discreteness, and high signal-to-noise ratio, so that the damage information of the catenary cable 8 contained in the vibration-generated stress wave passing through the damaged area will hardly be disturbed, ensuring the validity of the vibration signal data used for the damage detection of the railway catenary cable anchor wire clamp 7 area. Therefore, it is possible to realize the damage detection of the railway catenary cable anchor wire clamp 7 area, fill the gap in the damage detection technology of the railway catenary cable anchor wire clamp 7 area, and has great application prospects in the field of railway catenary cable 8 detection technology. The computer 4 processes the signal, performs preprocessing, time domain analysis, frequency domain analysis, wavelet packet decomposition and EMD extraction of the collected vibration signal IMF component for feature extraction, and builds a damage collaborative identification model to identify the extracted features, thereby achieving non-destructive, real-time and efficient detection of damage in the railway load-bearing cable anchor clamp 7 area.

[0076] Example 2

[0077] This embodiment provides a method for detecting damage in the anchor clamp area of ​​a catenary cable based on vibration analysis. Before damage detection, a damage collaborative identification model is established through orthogonality experiments, and then defect detection is performed on the anchor clamp area of ​​the railway catenary cable.

[0078] (I) Orthogonality experiment

[0079] Select several standard workpieces in the railway anchor cable clamp area with various types of defects (including normal workpieces without defects), and then perform orthogonal tests on the selected railway anchor cable clamp area according to the following steps to calculate the damage assessment factor β of each eigenvalue m , and reconstruct the feature vector at the same time:

[0080] S1. The guide trolley moves to the bottom of the anchor wire clamp area of ​​the load-bearing cable. The lifting platform is adjusted so that the claw assembly clamps the beams at both ends of the anchor wire clamp. The two sets of rotary mechanical arms rotate to the support clamps to clamp the load-bearing cable. At the same time, the exciter and the magnetic vibration sensor enter the working position. The exciter contacts the load-bearing cable. After the magnetic vibration sensor is powered on, it is adsorbed under the anchor wire clamp by magnetic force.

[0081] The above step S1 specifically includes the following sub-steps:

[0082] S11, push the guide rail trolley 2 to move to the vicinity of the load-bearing cable anchor clamp 7 area to be measured, lift the lifting platform 3, and align the claw arm unit 6 and the area to be measured through the screw guide rail I 241 and the screw guide rail II 242;

[0083] S12, lower the lifting platform 3, ensure that the hook 65 is in close contact with the cantilever beam 9 at both ends of the anchor wire clamp 7, rotate the rotary mechanical arm 62 to the upper end support clamp to clamp the load-bearing cable 8, and the exciter 64 and the magnetic vibration sensor 63 enter the working position;

[0084] S13, after being powered on, the magnetic vibration sensor 63 is adsorbed under the anchor clamp 7 by magnetic force.

[0085] S2, the exciter on one set of rotating mechanical arms generates a vibration excitation signal, the magnetic vibration sensor on another set of rotating mechanical arms receives the vibration signal, and the acquisition card receives the vibration signal and transmits it to the computer, such as Figure 8 shown.

[0086] The above step S2 includes the following sub-steps:

[0087] S21, controlling the vibration frequency and pressure of the vibrator, as well as the excitation time and the signal receiving time of the magnetic vibration sensor through the MCU control board;

[0088] S22, controlling the vibrator on one set of rotary mechanical arms to generate a vibration excitation signal through the MCU control board, and the magnetic vibration sensor on the other end of the rotary mechanical arm to receive the vibration signal, and the acquisition card receives and transmits the signal to the computer;

[0089] S23, in the above step S22, the vibrators and the working ends of the magnetic vibration sensors on the two sets of rotary mechanical arms are exchanged, and the acquisition card receives and transmits the signal to the computer;

[0090] S24. Repeat the above steps S21, S22 and S23 multiple times according to actual detection requirements to collect sufficient vibration signals for damage detection of the load-bearing cable anchor clamp area.

[0091] S3. The computer performs preprocessing, time domain analysis, frequency domain analysis, wavelet packet decomposition and feature extraction of the IMF component of the vibration signal extracted by EMD, and builds a damage collaborative identification model to identify the damage state of the anchor cable clamp area, such as Fig. 9 , Fig.10 , Fig.11 The damage state includes the damage category and the damage degree, wherein the damage category includes but is not limited to common damage categories in the art such as crack, wear, and broken strands, and the damage degree includes but is not limited to common damage degrees in the art such as crack depth, wear area, and number of broken strands.

[0092] Preprocess the signal and extract the eigenvalue component eigenvector: Use the feature evaluation method based on the damage state distance to evaluate the extracted signal features to achieve a more accurate classification of the damage state. First, determine the number of damage states M of the cable strand node structure, then perform vibration signal detection, and collect n groups of signal samples in each type of damage state; use the following feature extraction method for vibration signals.

[0093] The above step S3 specifically includes the following sub-steps:

[0094] S31, first preprocessing the vibration signal, that is, intercepting the same time length to ensure that all intercepted signals contain the same discrete data points, and obtaining the spectrum information corresponding to the vibration signal through discrete fast Fourier transform FFT;

[0095] S32, time domain extraction, select the time domain peak value and the peak coefficient of the direct wave of the vibration signal as the first and second type of feature vectors respectively In addition, the energy value and standard deviation of the signal are selected as time domain features and

[0096] S33, frequency domain extraction, perform corresponding spectrum analysis on the signal at the receiving end of the vibration sensor, and select the peak value of the spectrum curve and the peak coefficient of the spectrum curve as the feature vector The crest factor of the spectrum curve is calculated using the following formula:

[0097]

[0098] Where A represents the signal discrete point spectrum vector, δ represents half of the original discrete signal data length;

[0099] S34, feature extraction based on wavelet packets, such as Fig.10 As shown in (a) and (b), the vibration signal is decomposed into three layers of wavelets using the db5 wavelet packet to obtain the wavelet packet coefficients. Where L is the wavelet coefficient; reconstruct the wavelet packet coefficients of each layer to obtain the reconstruction coefficient (For wavelet packet decomposition and reconstruction, see "Sun Chuanqi, Teng Zhaosheng, Zou Liming. Application of wavelet packet decomposition and reconstruction algorithm in reactive energy measurement [J]. Electric Power Science and Engineering, 2005 (4): 45-48. DOI: 10.3969 / j.issn.1672-0792.2005.04.013."), the energy of the third-layer wavelet packet reconstruction coefficient is used as the seventh to fourteenth category feature quantity Wavelet energy Fig.10 (c) as shown.

[0100]

[0101] S35, EMD feature extraction, such as Fig.11 As shown in FIG. 1 , the vibration detection signal sample is decomposed by EMD (Empirical Mode Decomposition) (EMD decomposition see “Zhang Xiaoji, Zhang Xin, Sun Jincai. Target feature extraction and selection based on empirical mode decomposition [J]. Journal of Northwestern Polytechnical University, 2006, 24 (4): 453-456. DOI: 10.3969 / j.issn.1000-2758.2006.04.012.”), the number of IMF components is determined, and the corresponding intrinsic mode function τ(y) is obtained, where y represents time; the two IMF energies with the largest energy proportion are obtained as the fifteenth type of feature quantity. and by For example:

[0102]

[0103] For the i-th signal sample under a certain damage category k, the feature vector can be extracted and constructed using the above feature extraction method:

[0104]

[0105] represents the i-th signal sample under a certain impairment category k.

[0106] S36. Build a multi-layer perceptron network as a damage collaborative identification model, identify the extracted features through the damage collaborative identification model, and preliminarily identify the damage status of the load-bearing cable anchor clamp area, such as Fig.12 As shown:

[0107] A damage collaborative identification model was established, which is a multi-layer perceptron network with one hidden layer, one input layer, and one output layer (such as Fig.12). Each unit in the input unit layer represents the input of the initial mode, that is, each unit receives the required characteristic vector of the structural damage signal of the tested online catenary strand node; the hidden layer (intermediate layer) represents the intermediate quantity, each unit in the first layer is connected to each unit in the second layer through a calculation with weights and biases (or each unit in the second layer is calculated), and each unit in the second layer is calculated with weights and biases to obtain each unit in the third layer (output layer). This is a forward propagation process. In the training stage of constructing a multi-layer perceptron network, the weights and biases of each layer are iterated and updated with the help of loss functions and back propagation methods. After several steps of iteration, the constructed damage collaborative identification model converges and reaches stability. Each node in the output layer corresponds to the identified structural damage state of a certain catenary strand node. This damage collaborative identification model is used to make a preliminary judgment on the wear state of the catenary anchor wire clamp area.

[0108] S4. Reconstruct features based on at least one of feature selection, feature vector clustering and real-time detection information feedback using feature evaluation factors, and then input the reconstructed features into the damage collaborative identification model (i.e., the multi-layer perceptron network) to re-identify the damage status of the load-bearing cable anchor clamp area.

[0109] In the present invention, a feedback optimization mechanism is adopted to reconstruct the optimal recognition features by setting feature evaluation factors for feature selection, feature vector clustering, real-time detection information feedback and other methods, so as to re-identify the damage state of the load-bearing cable anchor clamp area.

[0110] The feature reconstruction in step S4 includes:

[0111] (1) By setting the feature evaluation factor method, the termination condition of the feature value is determined, and the feature signal is determined according to the termination condition, thereby obtaining the reconstructed feature.

[0112] The evaluation process first needs to determine the number of damage states M of the anchor line node structure of the load-bearing cable, and then perform vibration signal detection. In each type of damage state, n groups of signal samples are collected. 16 values ​​are extracted from each group of signal samples, and these 16 values ​​are evaluated. Represents a feature of a feature set under a certain damage state, that is, the mth feature value extracted from the i-th signal sample under the k-th damage state. The specific steps are:

[0113] Calculate the mean of the sample characteristics of each vibration signal under the same damage state:

[0114]

[0115] The average distance of sample features under different damage states is calculated from the feature average:

[0116]

[0117] Calculate the average distance between the same feature values ​​in different signal samples under the same damage state:

[0118]

[0119] Calculate the average distance between various damage states for the same feature value in different samples:

[0120]

[0121] Using the two distances calculated in the above steps, calculate the damage assessment factor of the damage characteristic value:

[0122] β m =d m / d′ m (9)

[0123] Damage Assessment Factor β m The larger the value of , the easier it is for the mth feature to distinguish different damage states.

[0124] Based on the above steps, we first calculate the recognition evaluation factor β of each extracted feature. m , and then determine the termination condition for the selection of eigenvalues, that is, set the accuracy threshold of structural damage identification of the anchor line node of the load-bearing cable to a corresponding threshold, and the features with damage assessment factors lower than this threshold will be eliminated; then set β m Sort from small to large and remove them one by one, reorganizing them into a new input vector The features selected when the termination conditions are reached are the sensitive features for damage identification under this working condition. The set of these reconstructed features is the new input vector after feature evaluation, which is input into the damage collaborative identification model to identify the damage status of the anchor clamp area of ​​the load-bearing cable.

[0125] (2) Cluster analysis of signal feature vectors: K-means cluster analysis is used to reconstruct the damage state vectors under different environmental factors to obtain the reconstructed features.

[0126] The optimal damage state vector can be further reconstructed using the classic K-means clustering analysis, that is, some obvious eigenvectors are used as cluster centers, and other eigenvectors are integrated into each class. A representative eigenvector is selected from each class to reconstruct the optimal damage state vector.

[0127] Cluster initialization: For n vibration detection signal samples, extract the evaluated signal feature vector The i-th signal sample under a certain impairment category k.

[0128] Determine the cluster center (s): that is, set it to the number of damage status categories that need to be identified.

[0129] Select the initial feature center: {z1(h),z2(h),…,z2(h)}, and then perform the corresponding signal feature division, where h represents the number of iterations;

[0130] If the division process satisfies: where ‖·‖ is the Euclidean norm, z 21 (h), z 22 (h) represents different loss status categories, z 21 (h) represents the cluster center of the loss status category to be determined, z 22 (h) indicates the non-z in s loss status categories 21 (h) other loss state category cluster centers; then calculate the new cluster centers under different environmental states:

[0131]

[0132] Among them, Q j is the number of sample vectors in a certain damage category after clustering, j = 1, 2, ..., s. After calculating the new cluster center, check whether it converges. If it satisfies the following formula, clustering ends:

[0133] z j (h+1)=z j (h) (11)

[0134] The obtained cluster center z j (h) as the optimal damage state vector v′ k , and re-input into the damage collaborative identification model for re-identification.

[0135] (3) Information superposition of online detection signals: Use the superposition method of online detection signals to correct the damage state vector and obtain the reconstructed features.

[0136] The damage state vector is continuously corrected by superposition of online detection signals, and the signal feature vector samples of the online detection are fed back to the damage state vector to which they belong for reconstruction. The process is as follows:

[0137] Calculate the initial damage state vector v k , the average value is calculated here:

[0138]

[0139] In the formula, represents the characteristic vector value of the i-th sample under damage category k, n k Represents the total number of signal samples under damage category k.

[0140] The measured characteristic vector q(t) of the collected signal at a certain moment is used as the feedback quantity and corrected as follows:

[0141] v″ k =v k ×[1-R(t)]+q(t)×R(t) (13)

[0142] Construct the real-time feedback parameter R(t), which represents the characteristics of the vibration signal collected at that moment.

[0143] The difference between the vector and the initial damage state vector is calculated using the following formula:

[0144]

[0145] The reconstructed damage state vector v″ k Re-input into the damage collaborative identification model for re-identification.

[0146] The vibration signal information collected online is continuously superimposed on the initial damage state vector. Obviously, the reconstructed damage state vector v″ k It contains the real-time collected signal information to be identified, so there is a strong consistency between the two.

[0147] This feature reconstruction method is particularly suitable for the case of loss recognition errors. This reconstruction process only feeds back and reconstructs the misidentified signal samples, thus avoiding information oversaturation to a certain extent, effectively speeding up the training speed of the damage collaborative recognition model, and having little impact on the samples that were originally correctly identified.

[0148] (II) Defect detection of railway load-bearing cable anchor clamp area

[0149] The diagnostic device provided in Example 1 is used to identify defects in the railway load-bearing cable anchor clamp area, which specifically includes the following steps:

[0150] S1. The guide trolley moves to the bottom of the anchor wire clamp area of ​​the load-bearing cable. The lifting platform is adjusted so that the claw assembly clamps the beams at both ends of the anchor wire clamp. The two sets of rotary mechanical arms rotate to the support clamps to clamp the load-bearing cable. At the same time, the exciter and the magnetic vibration sensor enter the working position. The exciter contacts the load-bearing cable. After the magnetic vibration sensor is powered on, it is adsorbed under the anchor wire clamp by magnetic force.

[0151] The above step S1 specifically includes the following sub-steps:

[0152] S11, push the guide rail trolley 2 to move to the vicinity of the load-bearing cable anchor clamp 7 area to be measured, lift the lifting platform 3, and align the claw arm unit 6 and the area to be measured through the screw guide rail I 241 and the screw guide rail II 242;

[0153] S12, lower the lifting platform 3, ensure that the hook 65 is in close contact with the cantilever beam 9 at both ends of the anchor wire clamp 7, rotate the rotary mechanical arm 62 to the upper end support clamp to clamp the load-bearing cable 8, and the exciter 64 and the magnetic vibration sensor 63 enter the working position;

[0154] S13, after being powered on, the magnetic vibration sensor 63 is adsorbed under the anchor clamp 7 by magnetic force.

[0155] S2, the exciter on one set of rotating mechanical arms generates a vibration excitation signal, the magnetic vibration sensor on another set of rotating mechanical arms receives the vibration signal, and the acquisition card receives the vibration signal and transmits it to the computer, such as Figure 8 shown.

[0156] The above step S2 includes the following sub-steps:

[0157] S21, controlling the vibration frequency and pressure of the vibrator, as well as the excitation time and the signal receiving time of the magnetic vibration sensor through the MCU control board;

[0158] S22, controlling the vibrator on one set of rotary mechanical arms to generate a vibration excitation signal through the MCU control board, and the magnetic vibration sensor on the other end of the rotary mechanical arm to receive the vibration signal, and the acquisition card receives and transmits the signal to the computer;

[0159] S23, in the above step S22, the vibrators and the working ends of the magnetic vibration sensors on the two sets of rotary mechanical arms are exchanged, and the acquisition card receives and transmits the signal to the computer;

[0160] S24. Repeat the above steps S21, S22 and S23 multiple times according to actual detection requirements to collect sufficient vibration signals for damage detection of the load-bearing cable anchor clamp area.

[0161] S3, preprocessing the received vibration signal, time domain analysis, frequency domain analysis, wavelet packet decomposition and feature extraction of the IMF component of the vibration signal extracted by EMD, and building a damage collaborative identification model to preliminarily identify the damage status of the anchor clamp area of ​​the load-bearing cable;

[0162] The above step S3 specifically includes the following sub-steps:

[0163] S31, pre-emphasize the collected vibration signal and obtain frequency spectrum information corresponding to the vibration signal through discrete fast Fourier transform FFT;

[0164] S32, time domain analysis: select the time domain peak value, crest factor, energy value, and standard deviation of the vibration signal as time domain features x1, x2, x3, and x4;

[0165] S33, frequency domain analysis: select spectrum peak and crest factor as frequency domain features: x5, x6;

[0166] S34, wavelet packet decomposition features: wavelet packet db5 is decomposed into three layers of wavelet packets, and the energy of the third layer of wavelet packet reconstruction coefficients is selected as the wavelet packet decomposition features x7~x 14 ;

[0167] S35, EMD feature extraction: Use EMD to decompose the active acoustic emission detection signal sample, determine the number of IMF components, and extract the two IMF energies with the largest proportion of the total energy as the EMD feature x 15 , x 16 ;

[0168] S36. Build a multi-layer perceptron network as a damage collaborative identification model, identify the extracted features through the damage collaborative identification model, and preliminarily determine the damage status of the load-bearing cable anchor clamp area;

[0169] Steps S31-S36 are the same as steps S31-S36 in the orthogonal test and will not be repeated here.

[0170] S4. Reconstruct the identification features by at least one of feature selection based on setting feature evaluation factors, feature vector clustering, and real-time detection information feedback. Then input the reconstructed identification features into the established damage collaborative identification model to re-identify the damage status of the load-bearing cable anchor clamp area.

[0171] Step S4 is basically the same as step S4 in the orthogonal experiment, except that:

[0172] (1) Since the damage state of the catenary anchor clamp region is preliminarily determined in step S36, based on the preliminary damage state determination result of the catenary anchor clamp region, the collected vibration signal can be classified according to the damage state determination result; then, feature reconstruction is performed based on the feature evaluation factor method;

[0173] (2) When using cluster analysis for feature reconstruction, a large amount of data samples is required. After collecting data for a period of time, it is necessary to combine the preliminary damage state judgment results of the load-bearing cable anchor clamp area and follow the steps given above;

[0174] (3) The method of reconstructing features by superimposing the information of online detection signals can be performed in real time; it can also be used to optimize the already constructed damage collaborative identification model. For example, when the initial damage state judgment result of the anchor clamp area of ​​the load-bearing cable is obviously inconsistent with the actual detection environment factors, or in special extreme environments (ice cover, high temperature, extreme cold, etc.), it is easy to cause recognition errors. It is necessary to manually detect the current environment, then construct label samples, and reconstruct features by superimposing the information of the previous online detection signal, and then optimize and train the already constructed damage collaborative identification model.

[0175] In summary, this embodiment performs preprocessing, time domain analysis, frequency domain analysis, wavelet packet decomposition and EMD extraction of the collected vibration signal to extract the features of the IMF component of the vibration signal, and builds a multi-layer perceptron network to realize the damage detection of the railway load-bearing cable anchor clamp area. The vibration signal analysis and identification method based on time domain analysis, frequency domain analysis, wavelet packet decomposition features and EMD features is different from the limitations of the single feature recognition method. By combining the intuitive response capabilities of time domain features and frequency domain features and the characteristics of wavelet packet decomposition and empirical mode decomposition (EMD) suitable for analyzing non-stationary signals, it can make up for the missing damage information between various features, and ensure that the damage detection process of the load-bearing cable anchor clamp area is more comprehensive and precise. From positioning and clamping the load-bearing cable anchor clamp area to be tested to the completion of defect identification, the manual detection time is greatly shortened and the detection efficiency is improved. Furthermore, the feedback optimization mechanism introduced in this embodiment adjusts the environmental adaptability of the detection device by setting feature evaluation factors to implement specific vector feature selection criteria in different environments; through feature vector clustering and real-time detection information feedback methods, feedback adjustment parameters are introduced to achieve optimal reconstruction of feature vectors, which can realize the detection of different damage types of railway cable anchor clamp area wear in different environments (including icing, cable wear, cable fracture, cable cracks, etc.), and is extremely suitable for efficient detection of railway cable anchor clamp area damage in various working environments.

[0176] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.

Claims

1. A device for detecting damage to the anchor clamp area of ​​a load-bearing cable based on vibration analysis, characterized in that: It comprises a mobile lifting unit, a claw arm unit (6) and a signal collection unit; The mobile lifting unit comprises a guide rail trolley (2) movable along a rail, a lifting platform (3) slidably connected to the top of the guide rail trolley (2) and movable in a direction perpendicular to the rail, and a support (5) vertically mounted on the lifting platform (3); The claw arm unit (6) comprises a mounting platform (61) arranged at the top of the pillar (5), a claw assembly arranged on the mounting platform (61) for clamping the cross beams at both ends of the anchor wire clamp, and two sets of rotating mechanical arms (62) symmetrically arranged on both sides of the mounting platform (61) and having completely the same structure, wherein a clamping groove I, a support clamp and a clamping groove II are sequentially arranged on the end surface of the rotating mechanical arm (62) with the mounting platform (61) as the center from the outside to the inside; the claw assembly comprises two hooks (65) invertedly mounted on the mounting platform (61) and having the same opening direction, wherein the hook portion of the hook (65) is in a U-shaped structure; the end of the rotating mechanical arm (62) is in a flat plate structure, and a support seat is arranged on the end surface of the rotating mechanical arm (62), and the support clamp and the clamping groove II are arranged on the support seat; the support clamp is in a V-shaped structure; the notch or the bottom of the notch of the clamping groove II is provided with a magnet for attracting the magnetic vibration sensor (63); The signal acquisition unit comprises a magnetic vibration sensor (63) and a vibration exciter (64) respectively installed in the card slot I and the card slot II, an acquisition card installed on the installation platform (61), and a computer (4) for pre-processing the received vibration signal and identifying damage in the area of ​​the load-bearing cable anchor clamp, the computer (4) being installed on the mobile lifting unit, and the two ends of the acquisition card being respectively connected to the magnetic vibration sensor (63) and the computer (4); When the rotary mechanical arm (62) rotates to the working position, the support clamp clamps the load-bearing cable, and the vibration exciter (64) and the magnetic vibration sensor (63) are respectively located below the load-bearing cable and the anchor wire clamp.

2. The device for detecting damage to the anchoring wire clamp area of ​​a load-bearing cable based on vibration analysis according to claim 1 is characterized in that: The guide rail trolley (2) comprises a vehicle body (21) and rail wheels (22) mounted on the vehicle body (21); a guide rail assembly for controlling the guided movement of a lifting platform (3) is mounted on the upper surface of the vehicle body (21); and the bottom of the lifting platform (3) is connected to the guide rail assembly.

3. The device for detecting damage to the anchor clamp area of ​​a load-bearing cable based on vibration analysis according to claim 2 is characterized in that: The guide rail assembly comprises two screw guide rails installed on the upper surface of the vehicle body (21) and in a vertical direction to the railway track, a feed screw rod I (242) installed between the two screw guide rails, and a hand wheel I (241) connected to one end of the feed screw rod.

4. The device for detecting damage to the anchoring wire clamp area of ​​a load-bearing cable based on vibration analysis according to any one of claims 1 to 3, characterized in that: The lifting platform (3) is a screw lifting platform (3), comprising a lifting platform base (311) and a lifting platform surface (312) arranged in parallel, a scissor bracket (321) arranged between the lifting platform base (311) and the lifting platform surface (312) for maintaining the height of the lifting platform surface (312), and a screw lifting machine connected to the scissor bracket (321), one end of a feed screw II (34) of the screw lifting machine is connected to a hand wheel II (33), and the lifting platform base (311) is slidably connected to the top of the guide rail trolley (2).

5. A method for detecting damage in the anchor clamp area of ​​a load-bearing cable based on vibration analysis, characterized in that: The use of the load-bearing cable anchor clamp area damage detection device based on vibration analysis as described in any one of claims 1 to 4 comprises the following steps: S1, the guide rail trolley (2) moves to the area below the anchor wire clamp of the load-bearing cable, the lifting platform (3) is adjusted so that the claw assembly clamps the crossbeams at both ends of the anchor wire clamp, and the two sets of rotating mechanical arms (62) rotate to the support clamp to clamp the load-bearing cable, and at the same time, the vibrator (64) and the magnetic vibration sensor (63) enter the working position, the vibrator (64) contacts the load-bearing cable, and the magnetic vibration sensor (63) is adsorbed under the anchor wire clamp by magnetic force after being powered on; S2, a vibration exciter (64) on a set of rotating mechanical arms (62) generates a vibration excitation signal, a magnetic vibration sensor (63) on another set of rotating mechanical arms (62) receives the vibration signal, and a collection card receives the vibration signal and transmits it to a computer (4); S3, the computer (4) performs preprocessing, time domain analysis, frequency domain analysis, wavelet packet decomposition and feature extraction of the IMF component of the vibration signal extracted by EMD on the received vibration signal, and builds a damage collaborative identification model to identify the damage status of the load-bearing cable anchor clamp area.

6. The method for detecting damage to the anchor clamp area of ​​a load-bearing cable based on vibration analysis according to claim 5 is characterized in that: The detection method also includes the following steps: S4. Reconstruct features by at least one of feature selection, feature vector clustering and real-time detection information feedback based on setting feature evaluation factors, and then input the reconstructed features into the established damage collaborative identification model to re-identify the damage status of the load-bearing cable anchor clamp area.

7. The method for detecting damage to the anchor clamp area of ​​a load-bearing cable based on vibration analysis according to claim 5 is characterized in that: The step S2 comprises the following sub-steps: S21, controlling the vibration frequency and pressure of the vibration exciter (64), the excitation time of the vibration exciter (64) and the signal receiving time of the magnetic vibration sensor (63) through the MCU control board; S22, controlling the vibration exciter (64) on one set of rotating mechanical arms (62) through the MCU control board to generate a vibration excitation signal, and the magnetic vibration sensor (63) on the other end of the rotating mechanical arm (62) receives the vibration signal, and the acquisition card receives and transmits the signal to the computer (4); S23, in the above step S22, the vibrators (64) and the working ends of the magnetic vibration sensors (63) on the two sets of rotary mechanical arms (62) are interchanged, and the acquisition card receives and transmits the signal to the computer (4); S24. Repeat the above steps S21, S22, and S23 to collect a set number of vibration signals for damage detection of the load-bearing cable anchor clamp area.

8. The method for detecting damage to the anchor clamp area of ​​a catenary cable based on vibration analysis according to any one of claims 5 to 7, characterized in that: The step S3 comprises the following sub-steps: S31, pre-emphasize the collected vibration signal and obtain frequency spectrum information corresponding to the vibration signal through discrete fast Fourier transform FFT; S32, time domain analysis, selecting the time domain peak value, crest factor, energy value, and standard deviation of the vibration signal as time domain features x1, x2, x3, and x4; S33, frequency domain analysis, select spectrum peak and crest factor as frequency domain features: x5, x6; S34, wavelet packet decomposition features, wavelet packet db05 is decomposed into three layers of wavelet packets, and the energy of the third layer of wavelet packet reconstruction coefficients is selected as wavelet packet decomposition features x7 to x14; S35, EMD feature extraction, using EMD to decompose the active acoustic emission detection signal sample, determine the number of IMF components, and obtain the two IMF energies with the largest proportion of the total energy as EMD features x15, x16; S36. Build a multi-layer perceptron network as a damage collaborative identification model, identify the extracted features through the damage collaborative identification model, and preliminarily determine the damage status of the load-bearing cable anchor clamp area.

Images