Active rail damage monitoring method and terminal based on nonlinear ultrasonic wave

Abstract

The present invention discloses an active rail damage monitoring method and terminal based on nonlinear ultrasonic wave, which belongs to the technical field of rail health monitoring, and includes a data training step: calculating the damage level threshold of the test signal under different states of the rail, and then establishing a mapping relationship between different damage levels of the rail and the test signal under different states; a monitoring and judgment step: judging the state corresponding to the track detection signal, comparing the detection signal with the damage level threshold under the corresponding state, and then determining the rail damage level. The present invention can achieve directional damage level judgment of detection signals under different states by establishing a mapping relationship between different damage levels of the track under different states and the test signal, eliminating the influence of different states on the characteristic parameters of the detection signal, thereby reducing the false alarm rate of damage level detection; further, judging the detection signal based on multiple damage thresholds can further improve the accuracy of damage level judgment.

Description

Technical Field

[0001] The present invention relates to the technical field of rail health monitoring, and in particular to an active rail damage monitoring method and a terminal based on nonlinear ultrasound. Background Art

[0002] With the continuous extension of railway operating mileage, the increase in train speed, traffic density, and line service time, higher requirements have been put forward for train operation safety, track equipment inspection efficiency, and maintenance costs. As an important component of track equipment, the service status of turnouts plays a vital role in the safe passage of trains. Since the point rail and the heart rail are the key components for turnouts to switch, they have complex and changeable boundary conditions, large cross-section changes, and conventional flaw detection operations cannot be quickly carried out. Inspection and other problems have become the weak link in ensuring the safe service of turnouts.

[0003] At present, the main methods for detecting damage to the turnout point rail are flaw detection trolleys and portable flaw detectors. The flaw detection trolleys can only work during the "window period" when the line is idle, with slow detection speed and still blind spots. Portable flaw detectors can focus on suspected damage and cover some blind spots of the flaw detection trolleys, but the detection efficiency is low and is greatly affected by human factors. These existing detection technologies can only rely on manual inspection and identification. This manual inspection cannot achieve uninterrupted real-time detection throughout the day, and there are serious safety hazards between two maintenance cycles. In order to achieve real-time monitoring of the degree of track damage, active ultrasonic guided wave detection technology is currently introduced into the field of turnout anti-breakage monitoring. However, due to the multi-mode and dispersion characteristics of ultrasonic guided waves, the accuracy of guided wave signal extraction is not high, and the accuracy of damage detection cannot be effectively guaranteed; and the point rail in different states such as repulsion and close contact will also cause changes in the characteristic value of the guided wave signal, further increasing the difficulty of turnout anti-breakage monitoring. Therefore, how to achieve real-time and accurate turnout anti-breakage monitoring is a technical problem that needs to be solved urgently. Summary of the invention

[0004] The purpose of the present invention is to overcome the problem that real-time and accurate turnout anti-breakage monitoring cannot be achieved in the prior art, and to provide an active rail damage monitoring method and terminal based on nonlinear ultrasonic waves.

[0005] The objective of the present invention is achieved through the following technical solution: an active rail damage monitoring method based on nonlinear ultrasound, the method comprising the following steps:

[0006] Data training step: calculating the damage level threshold of the test signal under different states of the rail, and then establishing a mapping relationship between different damage levels of the rail under different states and the test signal; the damage level threshold includes a damage threshold of microcracks and / or a damage threshold of macrocracks, and the damage threshold of microcracks is a nonlinear parameter threshold; the damage threshold of macrocracks includes a time domain signal similarity threshold, a time domain signal cross-correlation threshold, a spectrum amplitude difference threshold, a spectrum similarity threshold, an empirical mode decomposition threshold and a Hilbert-Huang transform threshold;

[0007] Monitoring and judgment steps: judge the state corresponding to the track detection signal, compare the detection signal with the damage level threshold under the corresponding state, and then determine the rail damage level.

[0008] In one example, the calculation of the damage level threshold of the test signal under different rail states specifically includes:

[0009] Obtaining a nonlinear parameter threshold of microcrack damage according to the nonlinear parameter value of the test signal of microcrack damage under different states; and / or

[0010] Obtaining a domain signal similarity threshold of macro crack damage according to the time domain similarity value between the test signal of macro crack damage in different states and the test signal sent by its transmitting end; and / or,

[0011] The cross-correlation threshold of the time domain signal of macro crack damage is obtained according to the cross-correlation damage index value of the test signal of macro crack damage under different states and its reference signal, and the reference signal represents the lossless signal under the corresponding state; and / or,

[0012] Obtaining a spectrum amplitude difference threshold of macro crack damage according to the spectrum amplitude change of the test signal of macro crack damage under different states and the test signal sent by its transmitting end; and / or,

[0013] Calculate the correlation function value of the test signal of macro crack damage under different states and its reference signal to obtain the spectrum similarity threshold of macro crack damage; and / or,

[0014] Decomposing the collected test signals of macro crack damage under different states into multiple linear steady-state intrinsic mode functions, and calculating the similarity of time-frequency distribution as the empirical mode decomposition threshold of macro crack damage; and / or,

[0015] The collected test signals of macro crack damage under different states are decomposed into multiple linear steady-state intrinsic mode functions, and each intrinsic mode function is transformed to obtain the three-dimensional spectrum distribution value of time, frequency and amplitude, and then the Hilbert-Huang transform threshold of macro crack damage is obtained.

[0016] As an option, the step of calculating the damage level threshold of the test signal under different states of the rail includes calculating any one or more thresholds of the time domain signal similarity threshold, the time domain signal cross-correlation threshold, the spectrum amplitude difference threshold, the spectrum similarity threshold, the empirical mode decomposition threshold and the Hilbert-Huang transform threshold, such as calculating the time domain signal similarity threshold, the time domain signal cross-correlation threshold, the spectrum amplitude difference threshold and the spectrum similarity threshold to determine the damage level.

[0017] In one example, comparing the detection signal with the damage level threshold in the corresponding state further includes:

[0018] The damage threshold of the macro crack is weighted to obtain the damage coefficient threshold, and the damage coefficient of the detection signal is compared with the damage coefficient threshold to determine whether the rail has a macro crack.

[0019] In one example, before the step of calculating the damage level thresholds under different states and / or before the step of determining the state corresponding to the track detection signal, a signal preprocessing step is also included:

[0020] The test signal and / or the detection signal are sequentially subjected to discrete Fourier transform processing, filtering processing, and discrete Fourier inverse transform processing.

[0021] In one example, the data training step further includes establishing a state judgment model:

[0022] Extract characteristic parameters of the test signal under a certain state;

[0023] A corresponding state judgment model is established based on the characteristic parameters of the test signal in the corresponding state.

[0024] In one example, the method further includes a state determination step:

[0025] The test signal or the detection signal is classified and processed to obtain the test signal or the detection signal in different states.

[0026] In one example, the classification process step further includes:

[0027] Perform principal component analysis on the test signal.

[0028] In one example, the monitoring and judging step further includes a rail break judging sub-step:

[0029] Determine whether the detection signal is a detection signal of the specified frequency. If not, the turnout is completely broken.

[0030] In one example, the method further includes a hierarchical alarm step:

[0031] Output different levels of alarm instructions according to the damage level of the rail.

[0032] It should be further explained that the technical features corresponding to the above examples can be combined or replaced with each other to form a new technical solution.

[0033] The present invention also includes a storage medium on which computer instructions are stored. When the computer instructions are executed, the steps of the active rail damage monitoring method based on nonlinear ultrasound formed by any one or more of the above examples are executed.

[0034] The present invention also includes a terminal, including a memory and a processor, wherein the memory stores computer instructions that can be run on the processor, and when the processor runs the computer instructions, it executes the steps of the active rail damage monitoring method based on nonlinear ultrasound formed by any example or multiple examples.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] (1) In one example, by establishing a mapping relationship between different damage levels of the track in different states and the test signal, it is possible to achieve directional damage level judgment of the detection signal in different states, eliminate the influence of different states on the characteristic parameters of the detection signal, and thus reduce the false alarm rate of damage level detection; further, judging the detection signal based on multiple damage thresholds can further improve the accuracy of damage level judgment.

[0037] (2) In one example, the preprocessing step can filter out external interference signals, thereby ensuring the credibility of the acquired detection signal, and further ensuring the accuracy and reliability of subsequent damage level detection.

[0038] (3) In one example, principal component analysis can greatly reduce the amount of redundant data calculations, save computing resources, and improve data processing efficiency, thereby ensuring the real-time nature of track damage detection.

[0039] (4) In one example, a graded alarm strategy is used to monitor the damage of turnouts throughout their life cycle, including microcracks / microdefects, macrocracks, and complete breakage of rails. Relevant departments can specify corresponding treatment measures based on the alarm level to improve the efficiency of turnout maintenance and avoid the occurrence of safety accidents.

[0040] (5) The LamB wave analogy method is used to program and solve the dispersion curve of the turnout rail, and the non-dispersive A0 mode is selected for excitation according to the dispersion curve to reduce the dispersion phenomenon.

[0041] (6) The signal collected when the rail is in an intact state is used as a reference signal. The signal contains information such as dispersion, multi-mode, and different working states such as tip-center rail repulsion or close contact. This eliminates the influence of non-damage factors on rail defect monitoring, thereby reducing the false alarm rate of the system.

[0042] (7) The ultrasonic wave (excitation signal) emitted by the ultrasonic sensor is a mechanical wave that is almost unaffected by electromagnetic and turnout parameters, and will not affect existing electrical equipment. The ultrasonic transmitter and receiver are both attached to the surface of the rail through a coupling agent, so there is no need to place the sensor (ultrasonic transmitter and ultrasonic receiver) inside the rail or make any changes such as drilling or grooving. This means that real-time monitoring of the entire cross-section of the turnout point rail can be achieved around the clock. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The specific implementation methods of the present invention are further described in detail below in conjunction with the accompanying drawings. The accompanying drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The same reference numerals are used in these drawings to represent the same or similar parts. The schematic embodiments of the present application and their descriptions are used to explain the present application and do not constitute improper limitations on the present application.

[0044] Figure 1 A method flow chart in an example of the present invention;

[0045] Figure 2 A flow chart of pre-processing and self-checking steps in an example of the present invention;

[0046] Figure 3 Establish a flow chart for the state judgment model in the preferred example of the present invention;

[0047] Figure 4 This is a schematic diagram of the installation of a sensor for a naturally damaged rail according to the present invention;

[0048] Figure 5 The present invention is a schematic diagram of the installation of a sensor for a non-destructive rail;

[0049] Figure 6 The present invention is a comparison diagram of signal collection for a non-damaged rail and a naturally damaged rail;

[0050] Figure 7 This is the damage index variation trend diagram of the R3 sensor at different damage levels;

[0051] Figure 8 This is a damage index variation trend chart of the R4 sensor at different damage degrees. DETAILED DESCRIPTION

[0052] The technical solution of the present invention is described clearly and completely below 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 are within the scope of protection of the present invention.

[0053] In the description of the present invention, it should be noted that the directions or positional relationships indicated by "center", "up", "down", "left", "right", "vertical", "horizontal", "inside", "outside", etc. are based on the directions or positional relationships described in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0054] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, "installation", "connection" and "connection" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0055] In addition, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0056] The method of the present invention is used for real-time monitoring of turnout rail damage and broken rails, especially for accurately judging the damage level of the track, so as to facilitate relevant staff to make corresponding prevention strategies. The following is an example of monitoring the damage level and breakage of the turnout tip / heart rail.

[0057] In one example, an active rail damage monitoring method based on nonlinear ultrasound, such as Figure 1 As shown, the method specifically comprises the following steps:

[0058] S1 data training step: calculate the damage level threshold of the test signal under different states of the rail, and then establish the mapping relationship between different damage levels of the rail and the test signal under different states; among which, different states correspond to the repulsion state or the close contact state of the tip / center rail; the test signal is the rail detection signal used for data training. At this time, the current state of the tip / center rail and the service state of the rail under the corresponding state are known, that is, the damage level of the rail, whether it is damaged, and whether it has been completely broken; in order to ensure the accuracy of damage level detection, the early data training needs to continuously correct the damage level threshold through a large amount of test data. The damage level threshold is used to represent the difference between the reference signal and the test signal (detection signal) when the track is not damaged. It is only related to the damage on the track structure and is not affected by changes in the surrounding environment. It is the only external factor that affects the detection result, that is, the damage level threshold is used to divide the current damage level of the rail, that is, the degree of damage. If the rail detection signal under the current state exceeds the current damage level threshold, it means that the rail has reached the damage level. More specifically, the damage level threshold includes the damage threshold of microcracks and the damage threshold of macrocracks, and the corresponding damage index includes the damage value of microcracks and the damage value of macrocracks, that is, the damage level of the present invention includes the damage level of microcracks and the damage level of macrocracks. The damage threshold of microcracks is a nonlinear parameter threshold; the damage threshold of macrocracks includes the time domain signal similarity threshold, the time domain signal cross-correlation threshold, the spectrum amplitude difference threshold, the spectrum similarity threshold, the empirical mode decomposition threshold and the Hilbert-Huang transform threshold. Judging the detection signal based on multiple damage thresholds can further improve the accuracy of damage level judgment.

[0059] S2 monitoring and judgment step: judge the state corresponding to the track detection signal, compare the detection signal with the damage level threshold under the corresponding state, and then determine the damage level of the rail. Among them, the detection signal is used to detect the service status of the rail. In this example, the test signal and the detection signal are both ultrasonic signals. The only difference is whether the current state and service status of the rail have been obtained when the signal is emitted, that is, it is only an ultrasonic signal obtained at different stages in the data processing process. The test signal and the detection signal are collectively referred to as monitoring signals. More specifically, when comparing the detection signal with the damage level threshold under the corresponding state, it is necessary to calculate the damage index of the detection signal, including the damage value of microcracks and the damage value of macrocracks, and then further compare the size of the damage index and the threshold to determine the damage level. In this example, a mapping relationship between different damage levels of the track in different states and the test signal is established, that is, the detection data of the tip / center rail repulsion state and the close fitting state are classified, and the damage level of the track in the repulsion state is judged by the damage level threshold of the track in the repulsion state, and the damage level of the track in the close fitting state is judged by the damage level threshold of the track in the close fitting state, thereby realizing directional damage level judgment of the detection signal in a specific state, eliminating the influence of different states on the characteristic parameters of the detection signal (amplitude and phase parameters, number of cycles, etc.), thereby reducing the false alarm rate of damage level detection.

[0060] Furthermore, an ultrasonic transmitter and an ultrasonic receiver are provided at corresponding horizontal positions on the rails. The excitation signal (detection signal) generated by the ultrasonic transmitter propagates along the rails and is received and collected at a specific rail position (the installation position of the ultrasonic receiver) and converted into an electrical signal. If the rails in the detection section (the rail area between the ultrasonic transmitter and its corresponding ultrasonic receiver) are not broken, the ultrasonic receiver will collect and identify the correct excitation signal; and once the rails in the detection section are broken, the excitation signal propagates to the cracks, and reflection, scattering, transmission and nonlinear phenomena will occur, thereby causing changes in the excitation signal, and the degree of damage to the rails is identified according to the changes in the excitation signal. In this application, since the wavelength of the excited ultrasonic signal is at the same level as the rail bottom angle thickness, rail waist thickness and rail head thickness of the rail, the LamB wave analogy method is used to program and solve the dispersion curve of the turnout rail, and the non-dispersive A0 mode is selected according to the dispersion curve for excitation, thereby reducing the dispersion phenomenon. This application uses a hanning window to modulate a sinusoidal wave signal of a specified frequency, and the number of cycles n is 10. The signal is stored in a signal source automatic transmitting device, which can amplify the signal by a certain multiple and then transmit the signal at a set time interval.

[0061] In this example, the ultrasonic wave (excitation signal) emitted by the ultrasonic sensor is a mechanical wave, which is almost unaffected by electromagnetic and turnout parameters and will not affect existing electrical equipment. The ultrasonic transmitter and receiver are both attached to the surface of the rail through a coupling agent, so there is no need to place the sensor (ultrasonic transmitter and ultrasonic receiver) inside the rail or make any changes such as drilling holes or grooving, thus enabling real-time monitoring of the entire cross-section of the turnout point rail around the clock.

[0062] In one example, calculating the damage level threshold of the test signal under different rail conditions specifically includes:

[0063] The nonlinear parameter threshold of microcrack damage is obtained according to the nonlinear parameter value of the test signal of microcrack damage under different states, and the maximum nonlinear coefficient in the lossless state is preferably used as the nonlinear parameter threshold; specifically, when encountering microcracks / microdamage during the propagation of ultrasonic signals in the waveguide, nonlinear phenomena such as high-order harmonics and subharmonics will be generated, and microcrack damage monitoring can be achieved by utilizing nonlinear phenomena; more specifically, the present application utilizes nonlinear characteristics such as second-order and third-order nonlinear coefficients to determine whether there is a nonlinear phenomenon in the detection signal.

[0064] The domain signal similarity threshold of macro-crack damage is obtained according to the time domain similarity value between the test signal of macro-crack damage in different states and the test signal sent by its transmitter. Specifically, since the rail is in a damage-free service state, no matter how the amplitude of the excitation signal changes, as long as the center frequency and boundary conditions remain unchanged, the shape of the received signal is similar. Therefore, it is possible to judge whether the rail has macro-crack damage by the similarity between the transmitter and receiver signals.

[0065] The cross-correlation damage index value of the time domain signal of macro crack damage under different states and its reference signal is used to obtain the cross-correlation threshold of the time domain signal of macro crack damage. The reference signal represents the lossless signal under the corresponding state. Specifically, the cross-correlation damage index of the time domain signal represents the cross-correlation of the signal in the time domain. According to a large number of test results, the cross-correlation performance of the time domain signal under different crack sizes is better for quantitatively analyzing the size of the damage. The cross-correlation damage index DI of the time domain signal is selected to characterize the size of the damage, that is, to judge whether the rail has macro crack damage. The specific calculation formula is:

[0066]

[0067] Among them, DIc(H,D)∈[0,1], H(t) represents the healthy (lossless) reference signal; D(t) represents the monitored damage signal; t1 represents the start time of the signal; t2 represents the end time of the signal.

[0068] The spectrum amplitude difference threshold of macro crack damage is obtained according to the spectrum amplitude change of the test signal of macro crack damage under different states and the test signal sent by its transmitter; specifically, if the monitoring area of ​​the rail is damaged, the received signal is mainly composed of the direct signal and the signal scattered by the damage. When the damage is not on the propagation path, the propagation distance of the damage scattered signal is farther than the direct signal, which is manifested as time shift in the time domain of the signal, and the signal after the scattered signal with time shift and the direct signal are superimposed is mainly manifested as the change of spectrum amplitude in the frequency domain, and then it can be judged whether the rail has macro crack damage by the change of spectrum amplitude signal. Specifically, the damage of the rail causes the "lateral movement" of the frequency distribution of the waveguide energy in the rail, which may be related to the modal conversion of the waveguide; if the deterministic relationship between the ultrasonic guided wave power spectrum and typical damage can be determined, it means that the damage in the rail can be identified by the change of the ultrasonic guided wave power spectrum. The calculation formula of the spectrum amplitude difference damage index DIf is as follows:

[0069]

[0070] Among them, 0 <DIf<1, It means taking the complex coefficient modulus value after Fourier transforming the signal. The spectrum amplitude difference damage index can determine the damage index in the area affected by the excitation-sensing path (detection section), and is also related to the amplitude of the scattered signal, that is, the signal amplitude contains the size information of some damage. In order to ensure the accuracy of the damage level judgment, in this example, the spectrum amplitude is not used alone to determine the area affected by the excitation-sensing path, but is further combined with the relative time shift coefficient to determine it.

[0071] The correlation function value of the test signal of macro crack damage collected under different states and its reference signal is calculated to obtain the spectrum similarity threshold of macro crack damage; specifically, if damage occurs in the monitoring area of ​​the rail, the signal propagates to the crack, and reflection, scattering, transmission and nonlinear phenomena will occur, causing the signal to change, which is different from the signal in the most original lossless state. This change becomes more obvious with the passage of time. This example uses the correlation function to quantify this difference.

[0072] The test signals of macro crack damage collected under different states are decomposed into multiple linear steady-state intrinsic mode functions (EMD) to obtain the empirical mode decomposition threshold of macro crack damage;

[0073] Each intrinsic mode function is transformed to obtain the instantaneous frequency and amplitude that change with time, and then the three-dimensional spectrum distribution value of time, frequency and amplitude is obtained, which is the Hilbert-Huang transform threshold used for macro crack damage judgment.

[0074] In one example, comparing the detection signal with the damage level threshold in the corresponding state further includes:

[0075] The damage threshold of the macro crack is weighted to obtain the damage coefficient threshold, and the damage coefficient of the detection signal is compared with the damage coefficient threshold to determine whether the rail has a macro crack. When a macro crack appears on the rail, the ultrasonic signal will produce reflection, multipath and other phenomena after passing through the damage, causing the waveform of the collected signal to change. The degree of change is directly related to the degree of damage. This degree of change is quantified as the damage coefficient, and the damage coefficient is used to achieve effective monitoring of macro cracks in turnouts (rails).

[0076] In one example, the process of calculating the damage level threshold of the test signal under different rail conditions and the process of calculating the damage index of the test signal also include:

[0077] The characteristic parameters of the reference signal in the non-destructive state are extracted to establish a prediction model; the reference signal in the non-destructive state is collected, which contains information such as dispersion, multi-mode and different working states such as tip-center rail repulsion or close contact, so as to eliminate the influence of non-damage factors on rail defect monitoring.

[0078] The prediction model is used to extract characteristic parameters of the test signal and the detection signal under different states, thereby realizing rapid calculation of the damage level threshold of the test signal and the damage index of the detection signal under different states.

[0079] As a preferred embodiment, the feature parameter extraction process can be processed through a sliding window to control the data processing volume while ensuring the parameter extraction accuracy.

[0080] In one example, before the step of calculating the damage level threshold under different states and before the step of determining the state corresponding to the track detection signal, a signal preprocessing step is also included:

[0081] The test signal and / or detection signal are processed by discrete Fourier transform, filtering, and inverse discrete Fourier transform in sequence. Specifically, in the broken rail monitoring process, the detection signal received by the piezoelectric sensor may be subject to various external interferences, including known or unknown, continuous or intermittent electromagnetic interference, especially in the high-speed railway turnout environment. In order to improve the signal recognizability to a certain extent, a fast and efficient signal processing method suitable for guided wave flaw detection is required to meet the basic signal processing requirements of on-site detection. The excitation signal sent by the transmitter of the present application is essentially a narrowband signal, which is then filtered out by filtering the external interference signal.

[0082] In one example, if Figure 2 As shown, the pre-processing step also includes a self-checking step, which specifically includes:

[0083] S04: Extract monitoring signal features to determine whether the sensor's transmission and reception are normal, that is, determine whether there is a signal component of a specified frequency in the detection signal. If not, issue a fault prompt message.

[0084] In one example, the data training step further includes establishing a state judgment model:

[0085] Extract characteristic parameters of the test signal under a certain state; wherein the characteristic parameters include amplitude and phase parameters, number of cycles, etc. of the test signal, that is, the current state, i.e., repulsion or close contact, can be determined through the characteristic parameters.

[0086] A corresponding state judgment model is established based on the characteristic parameters of the test signal in the corresponding state, that is, a repulsion state judgment model and a close contact state judgment model are realized.

[0087] In one example, the method further includes a state determination step:

[0088] The test signal or detection signal is classified and processed to obtain the test signal or detection signal in different states. Specifically, in a lossless state, ultrasonic signals in different states are extracted; then the principal components of these ultrasonic signals are extracted; the first principal component is used as the cluster center to extract the first type of signal; then the remaining signals are calculated repeatedly in the previous step to obtain the second type of signal; and so on, and all the first principal components are used as reference signals, so as to achieve the purpose of classification and removal of redundant data.

[0089] In one example, the classification step further includes:

[0090] The test signal is processed by principal component analysis. Specifically, in order to make the model parameters more robust, sufficient data is required to be collected during the data training phase; when the number of training data reaches the set requirements, PCA is used to extract the main components of the data, and the main components of the data of the turnout point rail in the two different states of repulsion and close contact are obtained, which reduces the amount of calculation of redundant data, saves computing resources, improves data processing efficiency, and thus ensures the real-time performance of track damage detection.

[0091] Combining the above examples to obtain a preferred example of establishing a state judgment model is as follows: Figure 3 As shown, the following steps are included:

[0092] S05: Perform principal component analysis on the monitoring signal;

[0093] S06: Classify and process the monitoring to obtain test signals in different states;

[0094] S07: Extract characteristic parameters of the test signal under the corresponding state to establish a corresponding state judgment model.

[0095] In one example, the monitoring and judging step further includes a rail break judging sub-step:

[0096] It is determined whether the detection signal is a detection signal of a specified frequency. If not, the turnout is completely broken. The specified frequency signal is a specific frequency signal emitted by an ultrasonic transmitter, and the specified frequency signal can be defined according to different user requirements.

[0097] As a preferred example, the monitoring and judgment steps of the track detection of the present application specifically include:

[0098] S21: judging whether the track has slight damage according to the detection signal; specifically, extracting the nonlinear parameters of the detection signal, and judging whether the frequency of the nonlinear phenomenon exceeds the nonlinear parameter threshold, indicating that the rail has slight damage.

[0099] S22: Determine whether the track has macro damage based on the detection signal; specifically, extract the damage index of the detection signal, namely, the time domain signal similarity value, the time domain signal cross-correlation value, the spectrum amplitude difference value, the spectrum similarity value, the EMD damage index value and the Hilbert-Huang damage index value. When the damage index exceeds the threshold, it indicates that the rail head or rail bottom is damaged.

[0100] S23: Determine whether the track is completely broken based on whether a detection signal of a specified frequency is received.

[0101] In one example, the method further includes a hierarchical alarm step:

[0102] S4: Output different levels of alarm instructions according to the damage level of the rail. Specifically, when it is judged that the current tip / center rail is in micro-crack damage, it will not affect the safety of passing vehicles in a short period of time. It is recommended to inspect the track at the nearest skylight point and output a third-level alarm instruction; when it is judged that the current tip / center rail is in macro-crack damage, it is recommended that the staff conduct key inspections and output a second-level alarm instruction; when it is judged that the current tip / center rail is completely broken, it must be inspected on the track in the shortest possible time, and a first-level alarm instruction is output. The alarm gestures of the first-level alarm instruction, the second-level alarm instruction, and the third-level alarm instruction are gradually weakened. This example can promptly inform the staff of the current damage level of the track through graded alarms, and prompt relevant departments to take corresponding measures to avoid the occurrence of safety accidents.

[0103] Combining the above examples, a preferred monitoring method of the present invention is obtained, comprising the following steps:

[0104] S1': classify and process the rail detection signals collected in real time to obtain the current state of the rail;

[0105] S2': Calculate the nonlinear parameter of the detection signal and determine whether the nonlinear parameter exceeds the threshold (the nonlinear parameter threshold in the current state). If so, proceed to step S3'; if not, proceed to step S4';

[0106] S3': Calculate the damage coefficient of the macro crack by using the sliding window to calculate the development trend of the nonlinearity, and judge whether the damage coefficient exceeds the threshold (the damage coefficient threshold in the current state). If not, continue monitoring; if exceeded, output a third-level alarm instruction;

[0107] S4': Calculate the damage coefficient of the macro crack and determine whether the damage coefficient exceeds the threshold (the damage coefficient threshold in the current state). If not, continue monitoring; if exceeded, output a secondary alarm instruction and enter step S5';

[0108] S5': Determine whether the real-time detection signal of the rail is a detection signal of a specified frequency. If not, the turnout is completely broken and a first-level alarm instruction is output; if yes, continue monitoring.

[0109] The principle of active broken rail monitoring in the present application is to use the collected signal in the non-destructive state as the reference signal (reference signal). The signal contains information such as frequency dispersion, multi-modality, and characteristic structure of sensors and rails, so as to eliminate the influence of non-damage factors on rail defect monitoring; when macro cracks appear on the rail, the ultrasonic wave will produce reflection, multi-path and other phenomena after passing through the damage, thereby affecting the sensor collection signal, and the degree of influence is directly related to the degree of damage, so the change of this signal is used to realize rail broken monitoring.

[0110] In order to verify the effectiveness of the rail crack identification of the present invention, a rail with natural damage is selected for verification. The sensor is installed at the following position: Figure 4 As shown; in order to compare the test results, a non-damaged rail with the same length and material was selected, and the sensor was installed at the following position: Figure 5 As shown, artificial injury is made at position X1 at the same time. The test steps are as follows:

[0111] S11': Install the transmitting sensor at F1 and the receiving sensors R1~R4, where R1 and R2 are installed on the naturally damaged rails; R3 and R4 are installed on the undamaged rails. The transmitting point F1 is connected to the same transmitting channel, that is, the undamaged and damaged rails transmit the same signal; the transmitting signal comes from the signal source automatic transmitting device;

[0112] S12': collecting N ultrasonic data received by the receiving sensors and comparing the damaged rails with the undamaged rails;

[0113] S13': make a flaw at the rail bottom angle on the undamaged rail, with the flaw size of 30*5mm. After the flaw is made, continue to transmit the signal source at the F1 position, and the R1~R4 sensors receive it, and calculate the damage index of the detection signal;

[0114] S14': Continue to make a flaw on the rail bottom angle on the undamaged rail, with the flaw size being 40*9mm. After the flaw is made, continue to transmit the signal source at the F1 position, and the R1~R4 sensors receive it, and calculate the damage index of the detection signal;

[0115] S15': Continue to make a flaw on the rail bottom angle on the undamaged rail, with a flaw size of 40*10. After the flaw is made, continue to transmit the signal source at the F1 position, and the R1~R4 sensors receive it, and calculate the damage index of the detection signal;

[0116] Furthermore, if Figure 6 The data collection comparison of damaged rails and undamaged rails shows the receiving signals collected by the receiving sensors of the naturally damaged rails and undamaged rails at the bottom angle of the rails. It can be seen from the figure that the receiving sensors of the two rails have collected obvious end face reflection wave signals, but in the receiving signals on the damaged rail, obvious damage reflection wave signals are collected between 0.003 and 0.004s. According to the arrival time of the reflection wave, the location of the damage can be calculated to be 2.43m, which is 0.07m different from the actual damage location of 2.5m, and the error is only 2.8%. This result shows that the monitoring method described in the present invention can accurately identify natural damage.

[0117] Furthermore, if Figure 7-8 As shown, Figure 7 This is the damage index variation trend diagram of the R3 sensor under different damage sizes (different damage degrees); Figure 8 This is a damage index variation trend diagram of the R4 sensor under different damage sizes. It can be seen that under different damage degrees, the various indicators of the R3 and R4 sensors show significant differences. The active broken rail monitoring method in this application can accurately identify the macroscopic damage of the rail.

[0118] This embodiment provides a storage medium, which has the same inventive concept as the active rail damage monitoring method based on nonlinear ultrasound formed by any one of the above examples or a combination of multiple examples, and stores computer instructions on the storage medium, which execute the steps of the active rail damage monitoring method based on nonlinear ultrasound formed by any one of the above examples or a combination of multiple examples when running.

[0119] Based on this understanding, the technical solution of this embodiment, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0120] This embodiment also provides a terminal, which has the same inventive concept as the active rail damage monitoring method based on nonlinear ultrasound formed by any one or more of the above examples, and includes a memory and a processor, wherein the memory stores computer instructions that can be run on the processor, and the processor executes the steps of the active rail damage monitoring method based on nonlinear ultrasound formed by any one or more of the above examples when running the computer instructions. The processor can be a single-core or multi-core central processing unit or a specific integrated circuit, or one or more integrated circuits configured to implement the present invention.

[0121] Each functional unit in the embodiment provided by the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0122] The present invention also includes an active turnout anti-breakage monitoring system based on nonlinear ultrasound. The monitoring method formed by any one of the above examples or a combination of multiple examples is applied based on this system. The system includes a signal transmitting end, a signal receiving end, a signal acquisition end and a signal processing end. The signal receiving end, the signal acquisition end and the signal processing end are connected in sequence. The signal transmitting end and the signal receiving end are both arranged on the rail and are on the same horizontal line. When the rail is not damaged, the signal receiving end can effectively receive the excitation signal sent by the transmitting end. As an option, the ultrasonic transmitter and the ultrasonic receiver are any one of a waveguide transducer, a piezoelectric sensor, and a piezoelectric transducer.

[0123] In another example, the present invention also includes an active turnout anti-breakage monitoring system based on nonlinear ultrasonic waves. The monitoring method formed by any one of the above examples or a combination of multiple examples is applied based on the system, and the system includes a training unit and a judgment unit; the training unit is used to calculate the damage level threshold of the test signal under different states of the rail, and then establish a mapping relationship between different damage levels of the rail and the test signal under different states; the judgment unit is used to judge the state corresponding to the track detection signal, compare the detection signal with the damage level threshold under the corresponding state, and then determine the damage level of the rail.

[0124] In one example, the system further includes a first preprocessing unit for sequentially performing discrete Fourier transform processing, filtering processing, and discrete Fourier inverse transform processing on the test signal and / or the detection signal.

[0125] In one example, the system further includes a model building unit for extracting characteristic parameters of a test signal under a certain state, and building a corresponding state judgment model based on the characteristic parameters of the test signal under the corresponding state.

[0126] In one example, the system further includes a state judgment unit, which is used to classify the test signal or the detection signal to obtain the test signal or the detection signal in different states.

[0127] In one example, the system further includes a second preprocessing unit configured to perform principal component analysis on the test signal.

[0128] In one example, the judgment unit further includes a rail break judgment module, which is used to judge whether the detection signal is a detection signal of a specified frequency. If not, the turnout is completely broken.

[0129] In one example, the system further includes an alarm unit for outputting alarm instructions of different levels according to the level of rail damage.

[0130] The above specific implementation methods are detailed descriptions of the present invention. It cannot be determined that the specific implementation methods of the present invention are limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, several simple deductions and substitutions can be made without departing from the concept of the present invention, which should be regarded as belonging to the protection scope of the present invention.

Claims

1. An active rail damage monitoring method based on nonlinear ultrasound, characterized in that: It includes the following steps: Data training step: calculating the damage level threshold of the test signal under different states of the rail, and then establishing a mapping relationship between different damage levels and the test signal under different states of the rail, and different states correspond to the repulsion state or the close contact state of the point rail and the heart rail; the damage level threshold includes the damage threshold of microcracks and the damage threshold of macrocracks, and the damage threshold of microcracks is a nonlinear parameter threshold; the damage threshold of macrocracks includes the time domain signal similarity threshold, the time domain signal cross-correlation threshold, the spectrum amplitude difference threshold, the spectrum similarity threshold, the empirical mode decomposition threshold and the Hilbert-Huang transform threshold; The data training step also includes establishing a state judgment model: Extract characteristic parameters of the test signal under a certain state; Based on the characteristic parameters of the test signal in the corresponding state, a corresponding state judgment model is established, which is a repulsion state judgment model and a close contact state judgment model; Monitoring and judgment steps: judge the state corresponding to the track detection signal, compare the detection signal with the damage level threshold under the corresponding state, and then determine the damage level of the rail; classify the detection data of the point rail and the heart rail in the repulsive state and the tightly fitted state, judge the damage level of the track in the repulsive state by the damage level threshold under the repulsive state, and judge the damage level of the track in the tightly fitted state by the damage level threshold under the tightly fitted state, so as to realize the directional damage level judgment of the detection signal under a specific state; The calculation of the damage level threshold of the test signal under different rail states specifically includes: The nonlinear parameter threshold of microcrack damage is obtained according to the nonlinear parameter value of the test signal of microcrack damage under different states; The domain signal similarity threshold of the macro crack damage is obtained according to the time domain similarity value between the test signal of the macro crack damage in different states and the test signal sent by the transmitting end; The cross-correlation threshold of the time domain signal of macro crack damage is obtained according to the cross-correlation damage index value of the test signal of macro crack damage under different states and its reference signal, and the reference signal represents the lossless signal under the corresponding state; The spectrum amplitude difference threshold of the macro crack damage is obtained according to the spectrum amplitude change of the test signal of the macro crack damage under different states and the test signal sent by the transmitting end; The correlation function value between the test signal of macro crack damage under different states and its reference signal is calculated to obtain the spectrum similarity threshold of macro crack damage; The test signals of macro crack damage collected under different states are decomposed into multiple linear steady-state intrinsic mode functions, and the similarity of time-frequency distribution is calculated as the empirical mode decomposition threshold of macro crack damage; The collected test signals of macro crack damage under different states are decomposed into multiple linear steady-state intrinsic mode functions, and each intrinsic mode function is transformed to obtain the three-dimensional spectrum distribution value of time, frequency and amplitude, and then the Hilbert-Huang transform threshold of macro crack damage is obtained; The comparing the detection signal with the damage level threshold under the corresponding state also includes: The damage threshold of the macro crack is weighted to obtain the damage coefficient threshold, and the damage coefficient of the detection signal is compared with the damage coefficient threshold to determine whether the rail has a macro crack. The method further comprises a state determination step: Classify and process the test signal or detection signal to obtain the test signal or detection signal in different states; An ultrasonic transmitter and an ultrasonic receiver are provided on the rail. The excitation signal generated by the ultrasonic transmitter propagates along the rail and is received and collected at a specific rail position and converted into an electrical signal. If the rail in the rail area between the ultrasonic transmitter and its corresponding ultrasonic receiver is not broken, the ultrasonic receiver will collect and identify the correct excitation signal; once the rail breaks somewhere in the rail area between the ultrasonic transmitter and its corresponding ultrasonic receiver, the excitation signal propagates to the crack, and reflection, scattering, transmission and nonlinear phenomena will occur, thereby causing a change in the excitation signal. The degree of damage to the rail is identified based on the change in the excitation signal; the wavelength of the excited ultrasonic signal is at the same level as the rail bottom angle thickness, rail waist thickness and rail head thickness of the rail. The LamB wave analogy method is used to program and solve the dispersion curve of the turnout rail, and the non-dispersive A0 mode is selected for excitation based on the dispersion curve.

2. The active rail damage monitoring method based on nonlinear ultrasound according to claim 1 is characterized in that: Before the step of calculating the damage level thresholds under different states and / or the step of determining the state corresponding to the track detection signal, a signal preprocessing step is also included: The test signal and / or the detection signal are sequentially subjected to discrete Fourier transform processing, filtering processing, and discrete Fourier inverse transform processing.

3. The active rail damage monitoring method based on nonlinear ultrasound according to claim 1 is characterized in that: The classification processing step also includes: Perform principal component analysis on the test signal.

4. The active rail damage monitoring method based on nonlinear ultrasound according to claim 1 is characterized in that: The monitoring and judging step also includes a rail break judging sub-step: Determine whether the detection signal is a detection signal of the specified frequency. If not, the turnout is completely broken.

5. The active rail damage monitoring method based on nonlinear ultrasound according to claim 1 is characterized in that: The method also includes a hierarchical alarm step: Output different levels of alarm instructions according to the damage level of the rail.

6. A terminal comprising a memory and a processor, wherein the memory stores computer instructions that can be executed on the processor, characterized in that: When the processor runs the computer instructions, the processor executes the steps of the active rail damage monitoring method based on nonlinear ultrasound as described in any one of claims 1-5.

Images