A method and system for processing rail ultrasonic guided wave signals in a variable temperature environment

Abstract

This invention discloses a method and system for processing ultrasonic guided wave signals from steel rails under varying temperature conditions. The method includes acquiring a reference signal from the rail under a non-destructive environment and a measurement signal from the rail under the current environment. After processing the reference and measurement signals, the minimum normalized energy root mean square error and the optimal time-domain stretching factor of the reference and measurement signals are obtained to determine whether there is damage to the rail. Determining rail damage through the normalized energy root mean square error has strong robustness to interference such as changes in ambient temperature, reducing the possibility of false detections and missed detections, and achieving reliable monitoring of the rail structural health.

Description

Technical Field

[0001] This invention relates to the field of rail transit technology, and in particular to a method and system for processing ultrasonic guided wave signals of rails under variable temperature environments. Background Technology

[0002] With the development of railways towards high-speed passenger transport and heavy-haul freight transport, the requirements for transportation safety are becoming increasingly stringent, making the structural health monitoring of rails a crucial foundation for safe railway operation. Ultrasonic guided wave structural health monitoring technology, a novel monitoring method, can achieve online real-time monitoring and crack detection across the entire end face of the rail, and has attracted widespread attention.

[0003] When a rail is damaged or broken, the guided wave signal will generate a corresponding echo at the defect location. To address this characteristic, the commonly used guided wave signal processing method is baseline subtraction, which involves subtracting the reference signal from the measured signal and determining the presence of damage based on the residual caused by the echo. This places high demands on the stability of the guided wave receiver signal. However, guided wave signals are easily affected by changes in environmental conditions. In practical applications, changes in external temperature can also cause changes in the rail guided wave signal, mainly in wave velocity and amplitude. Therefore, when using baseline subtraction for damage detection, it is impossible to distinguish between the effects of damage and temperature on the signal, leading to missed and false detections of rail damage.

[0004] To address the aforementioned problems in ultrasonic guided wave monitoring of rails, this invention provides a method and system for processing ultrasonic guided wave signals of rails under varying temperature conditions. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for processing ultrasonic guided wave signals of rails under variable temperature environments. Considering the impact of temperature changes on the wave velocity and amplitude of the rail guided waves, the invention achieves adaptive stretching of the measurement signal in the time domain to compensate for wave velocity changes, and judges rail damage by normalizing the energy mean square error. It has strong robustness to interference such as changes in ambient temperature, reduces the possibility of false detection and missed detection, and realizes reliable monitoring of the health of rail structures.

[0006] To achieve the above objectives, the present invention provides a method for processing ultrasonic guided wave signals of steel rails under variable temperature conditions, comprising:

[0007] Collect reference signals from the rail under non-destructive conditions and measurement signals from the rail under the current conditions;

[0008] The minimum normalized energy mean square error and the optimal time-domain stretch factor between the reference signal and the measurement signal are obtained to determine whether there is damage to the rail.

[0009] Furthermore, the acquisition of the reference signal in the rail under non-destructive conditions and the current signal in the rail under the current conditions includes,

[0010] A transducer is installed on the rail, and a Hanning window modulation signal is used to excite the transducer to generate an ultrasonic guided wave signal.

[0011] An ultrasonic sensor installed on the rail is used to receive the echo signal in the rail; the echo signal includes a reference signal and a measurement signal. The reference signal represents the echo signal obtained at temperature T0, and the measurement signal represents the echo signal obtained after a temperature change δT.

[0012] Furthermore, the transducer sends a sinusoidal signal modulated by the Hanning window function to excite the transducer to generate an ultrasonic guided wave signal;

[0013] The modulated sinusoidal signal is:

[0014] s(t)=A0w(t)sin(2πft)

[0015] Where s(t) represents the sinusoidal signal modulated by the Hanning window function sent by the transducer, A0 represents the amplitude of the excitation ultrasonic guided wave signal, f is the frequency of the excitation ultrasonic guided wave signal, w(t) represents the Hanning window function, and t represents the time variable.

[0016] Furthermore, the Hanning window function is:

[0017]

[0018] Where w(t) represents the Hanning window function, f is the frequency of the excitation signal, n' is the number of sinusoidal signal cycles in the Hanning window, and t represents the time variable.

[0019] Furthermore, the reference signal is:

[0020]

[0021] Where u0(t) represents the reference signal, λ i s represents the reflection coefficient of the echo from the i-th rail feature structure. i (t) represents the i-th characteristic echo signal received at time t, where t i Let be the arrival time of the i-th characteristic echo, s represent the sinusoidal signal modulated by the Hanning window sent by the transducer, and t represent the time variable.

[0022] Furthermore, the measurement signal is:

[0023]

[0024] Where u1(t) represents the measurement signal, λ i Let t' represent the reflection coefficient of the echo from the i-th rail feature structure, s represent the sinusoidal signal modulated by the Hanning window transmitted by the transducer, and t'i t represents the arrival time of the i-th characteristic echo after a temperature change. i Let represent the arrival time of the i-th characteristic echo, β be the time-domain stretching factor, and t represent the time variable.

[0025] Furthermore, obtaining the minimum normalized energy mean square error and the optimal time-domain stretching factor between the reference signal and the measurement signal includes,

[0026] The extreme range of the time-domain stretching factor β [β min ,β max and minimum normalized energy mean square error Perform initialization definition;

[0027] Within the extreme range of the time-domain stretching factor β [β min ,β max Within this scope, the time-domain stretching factor β is iterated through with a step size to perform time-domain stretching on the reference signal or the measured signal, and the normalized mean square error of their energy is calculated. Until the minimum normalized energy mean square error is obtained through iterative calculation. and the optimal time-domain stretching factor; where the optimal time-domain stretching factor is the minimum normalized energy mean square error. The corresponding time-domain stretching factor β.

[0028] Furthermore, the reference signal or measurement signal undergoes time-domain stretching and the normalized energy mean square error of both is calculated. include,

[0029] Time-domain stretching: performing a cubic spline interpolation stretching transformation on a reference signal or measurement signal;

[0030] Mean removal processing: The mean removal processing is performed on the reference signal and the measurement signal after time-domain stretching respectively;

[0031] Normalization processing: Energy normalization processing is performed on the mean-removed reference signal and the measurement signal respectively to obtain the normalized reference signal and the normalized measurement signal;

[0032] The normalized energy mean square error is obtained based on the normalized reference signal and the normalized measurement signal.

[0033] Furthermore, the time-domain stretching factor β = β is initialized and defined for the first calculation. min Minimum normalized energy mean square error

[0034] Determine whether the currently calculated time-domain stretching factor β is greater than the initially defined maximum time-domain stretching factor β. max :

[0035] If the currently calculated time-domain stretching factor β is not greater than the initially defined maximum time-domain stretching factor β. max Then determine whether the currently calculated time-domain stretching factor β is greater than 1;

[0036] If the currently calculated time-domain stretching factor β is greater than the initially defined maximum time-domain stretching factor β max The output is the calculated minimum normalized mean square error of energy.

[0037] Furthermore, it is determined whether the calculated time-domain stretching factor β is greater than 1 in order to perform time-domain stretching processing, including:

[0038] If the currently calculated time-domain stretching factor β is not greater than 1, it means that the measurement signal is stretched. Then, the currently calculated measurement signal is stretched and transformed, and the measured signal after stretching and transformed is compared with the currently calculated reference signal to obtain the normalized energy mean square error between the two.

[0039] If the calculated time-domain stretching factor β is greater than 1, it indicates that the measured signal is compressed. Then, a stretching transformation is performed on the currently calculated reference signal, and the normalized energy mean square error between the currently calculated measured signal and the stretched reference signal is calculated.

[0040] Furthermore, the process continues until the minimum normalized energy mean square error is obtained through iterative calculation. and the optimal time-domain stretching factor, including,

[0041] Determine the normalized energy mean square error obtained in each iteration of the calculation. Is it greater than the minimum normalized energy mean square error defined in the initialization?

[0042] If the normalized energy mean square error after stretching calculation The minimum normalized energy mean square error greater than the initial definition Then, iterate through the values ​​of the time-domain stretching factor β with a step size μ, i.e., β = β + μ; then continue to return to check whether the currently calculated time-domain stretching factor β is greater than the initialized maximum value β of the time-domain stretching factor. max ;

[0043] If the normalized energy mean square error after stretching calculation Not greater than the minimum normalized energy mean square error defined in the initialization. Then the normalized energy mean square error at this time The minimum normalized mean square error of energy, i.e. Then, it returns to check whether the currently calculated time-domain stretching factor β is greater than the initialized maximum time-domain stretching factor β. max .

[0044] Furthermore, the measurement signal after the stretching transformation is:

[0045]

[0046] Where u'1(t) represents the measured signal after stretching, u1(βt) represents the measured signal after stretching, β is the time-domain stretching factor, t represents the time variable, and λ i Let s represent the reflection coefficient of the echo from the i-th rail feature structure, s represent the sinusoidal signal modulated by the Hanning window transmitted by the transducer, and t represent the reflection coefficient of the echo from the i-th rail feature structure. i This represents the arrival time of the i-th characteristic echo.

[0047] Furthermore, the reference signal after the stretching transformation is:

[0048] u'0(t)=u0[(1 / β)·t]

[0049] Where u'0(t) represents the stretched reference signal, u0 represents the reference signal, β is the time-domain stretching factor, and t represents the time variable.

[0050] Furthermore, the mean-reduced reference signal and the mean-reduced measurement signal are obtained using the following formula:

[0051]

[0052] Among them, u' 0m (n) represents the baseline signal after mean removal processing, u' 1m u'0(n) represents the measurement signal after mean removal processing, u'1(n) represents the reference signal after time-domain stretching processing, N is the length of the acquired digital signal, and n is the number of sampling sequence points.

[0053] Furthermore, the normalized reference signal and the normalized measurement signal are obtained through the following formula:

[0054]

[0055] Among them, u' 0mn (n) represents the normalized reference signal, u' 1mn (n) represents the normalized measurement signal, u' 0m (n) represents the baseline signal after mean removal processing, u' 1m (n) represents the measurement signal after mean removal processing, where N is the length of the acquired digital signal and n is the number of sampling sequence points.

[0056] Furthermore, the normalized energy mean square error can be obtained using the following formula:

[0057]

[0058] in, U' represents the normalized mean squared energy error. 0mn (n) represents the normalized reference signal, u' 1mn (n) represents the normalized measurement signal, where N is the length of the acquired digital signal and n is the number of sampling sequence points.

[0059] This invention also provides a rail ultrasonic guided wave signal processing system for variable temperature environments, including an acquisition unit and a calculation and acquisition unit.

[0060] The acquisition unit is used to acquire the reference signal in the rail under non-destructive conditions and the measurement signal in the rail under the current conditions.

[0061] The calculation and acquisition unit is used to acquire the minimum normalized energy mean square error and the optimal time-domain stretching factor between the reference signal and the measurement signal in order to determine whether there is damage to the rail.

[0062] Furthermore, the acquisition unit is used to acquire reference signals from the rail under non-destructive conditions and measurement signals from the rail under the current conditions, including:

[0063] The acquisition unit is used to excite the transducer to generate an ultrasonic guided wave signal by using a Hanning window modulation signal based on the transducer installed on the rail.

[0064] The acquisition unit is used to receive echo signals from the rail using ultrasonic sensors installed on the rail; wherein the echo signal includes a reference signal and a measurement signal, the reference signal representing the echo signal obtained at temperature T0, and the measurement signal representing the echo signal obtained after a temperature change δT.

[0065] Furthermore, the calculation and acquisition unit is used to acquire the minimum normalized energy mean square error and the optimal time-domain stretching factor between the reference signal and the measurement signal, including:

[0066] The calculation unit is used to determine the extreme range of the time-domain stretching factor β [β]. min ,β max and minimum normalized energy mean square error Perform initialization definition;

[0067] The calculation unit is used to calculate the extreme value range of the time-domain stretching factor β [β]. min ,β max Within this range, the time-domain stretching factor β is iterated through with a step size to stretch the reference signal or the measured signal, and the normalized mean square error of their energy is calculated. Until the minimum normalized energy mean square error is obtained and the optimal time-domain stretching factor; where the optimal time-domain stretching factor is the minimum normalized energy mean square error. The corresponding time-domain stretching factor β.

[0068] The technical effects and advantages of this invention are as follows: 1. Existing baseline signal stretching methods stretch in the time domain, while this invention uses cubic spline interpolation to stretch the reference signal or measurement signal in the time domain to compensate for the influence of temperature on the guide wave velocity of the rail. This eliminates the need for time-frequency transformation, thus reducing the complexity of the algorithm.

[0069] 2. Existing optimal baseline selection methods require building a baseline database of the structure at different temperatures. However, the temperature of rails can vary by hundreds of degrees Celsius during service. The optimal baseline method cannot traverse all temperature ranges in the early stages and lacks monitoring capabilities when building the baseline database. In contrast, this invention covers the range of rail temperature changes by traversing a wider range of tensile factors. Furthermore, it only requires collecting a small amount of signals as a reference in the early stages and has monitoring capabilities after installation.

[0070] 3. Existing baseline signal stretching methods use the minimum residual between the stretched reference signal and the measured signal as the criterion for selecting the optimal stretching factor, without considering the impact of temperature changes on signal amplitude. This invention, however, uses the normalized energy root mean square error as the criterion for selecting the optimal stretching factor, avoiding the influence of amplitude changes on the calculation results. Furthermore, the normalized energy root mean square error can also serve as a standard for judging rail damage, allowing for the assessment of rail health status without additional calculations after stretching.

[0071] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or will be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures pointed out in the description, claims and drawings. Attached Figure Description

[0072] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0073] Figure 1 This is a flowchart illustrating the steps of a method for processing ultrasonic guided wave signals of steel rails under varying temperature conditions, as described in this invention.

[0074] Figure 2 This is a flowchart of the calculation process in the ultrasonic guided wave signal processing method for rails under variable temperature conditions according to the present invention.

[0075] Figure 3This is a schematic diagram of the structure of an ultrasonic guided wave signal processing system for rails under variable temperature conditions according to the present invention.

[0076] Figure 4 This is a comparison diagram of the signal waveforms before time-domain stretching compensation during the experimental verification of the method of the present invention.

[0077] Figure 5 This is a schematic diagram of the normalized energy root mean square error and temperature before time-domain stretching compensation during the experimental verification of the method of the present invention.

[0078] Figure 6 This is a comparison diagram of the signal waveforms after time-domain stretching compensation during the experimental verification of the method of the present invention.

[0079] Figure 7 This is a schematic diagram of the normalized energy root mean square error and temperature after time-domain stretching compensation during the experimental verification of the method of the present invention. Detailed Implementation

[0080] The following will clearly and completely describe the technical solutions in the embodiments of the present invention 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 making creative efforts are within the scope of protection of the present invention.

[0081] To address the shortcomings of existing technologies, this invention discloses a method for processing ultrasonic guided wave signals from steel rails under varying temperature conditions, such as... Figure 1 As shown, it includes the following steps:

[0082] Collect reference signals from the rail under non-destructive conditions and measurement signals from the rail under the current conditions;

[0083] The minimum normalized energy mean square error and the optimal stretch factor between the reference signal and the measurement signal are obtained to determine whether there is damage to the rail.

[0084] Specifically,

[0085] During rail structural health monitoring, a transducer is installed at the rail web. The transducer sends a sinusoidal signal s(t) modulated by a Hanning window function to excite the transducer to generate an ultrasonic guided wave signal. An ultrasonic sensor installed near the transducer then receives the echo signal from the rail. The echo signal includes a reference signal and a measurement signal. The reference signal represents the echo signal obtained at temperature T0, and the measurement signal represents the echo signal obtained after a temperature change δT.

[0086] The modulated sinusoidal signal is:

[0087] s(t)=A0w(t)sin(2πft)

[0088] Where s(t) represents the sinusoidal signal modulated by the Hanning window sent by the transducer, A0 represents the amplitude of the excitation ultrasonic guided wave signal, f is the frequency of the excitation ultrasonic guided wave signal, w(t) represents the Hanning window function, and t represents the time variable.

[0089] The Hanning window function is as follows:

[0090]

[0091] Where e(t) represents the Hanning window function, f is the frequency of the excitation signal, n' is the number of sinusoidal signal cycles in the Hanning window, and t represents the time variable.

[0092] The reference signal, u0(t), represents the echo signal at temperature T0, and can be represented as the superposition of multiple characteristic echoes with different amplitudes and arrival times; the reference signal is:

[0093]

[0094] Where u0(t) represents the reference signal, λ i s represents the reflection coefficient of the echo from the i-th rail feature structure. i (t) represents the i-th characteristic echo signal received at time t, where t i Let be the arrival time of the i-th characteristic echo, s represent the sinusoidal signal modulated by the Hanning window sent by the transducer, and t represent the time variable.

[0095] The time shift of the rail characteristic echo during temperature changes is mainly caused by changes in wave velocity. Therefore, when the temperature changes by δT, the echo signal received by the sensor, i.e., the measurement signal, is expressed as:

[0096]

[0097] Where u1(t) represents the measurement signal, λ i Let t' represent the reflection coefficient of the echo from the i-th rail feature structure, s represent the sinusoidal signal modulated by the Hanning window transmitted by the transducer, and t' i t represents the arrival time of the i-th characteristic echo after a temperature change. i Let represent the arrival time of the i-th characteristic echo, β be the time-domain stretching factor, and t represent the time variable.

[0098] Then, the minimum normalized energy mean square error and the corresponding optimal stretching factor are calculated based on the reference signal and the measurement signal. Before the calculation, the time-domain stretching factor β and the minimum normalized energy mean square error are first evaluated. Perform initialization definition; define the extreme range of the time-domain stretching factor β as [β...]. min ,βmax The time-domain stretching factor β = β is defined in the first calculation. min And define the minimum normalized energy variance of the first calculation.

[0099] During calculation, the extreme range of the time-domain stretching factor β is [β...]. min ,β max Within this range, the time-domain stretching factor β is iterated through with a step size μ to stretch the reference signal or the measured signal and calculate the normalized mean square error of their energy. Until the minimum normalized energy mean square error is obtained and the optimal time-domain stretching factor; where the optimal time-domain stretching factor is the minimum normalized energy mean square error. The corresponding value of the time-domain stretching factor β.

[0100] like Figure 2 As shown, the specific calculation process is as follows:

[0101] First, regarding β and Perform initialization; then determine whether the currently calculated time-domain stretching factor β is greater than the maximum value β defined in the initialization. max :

[0102] If the currently calculated time-domain stretching factor β is not greater than the initially defined maximum time-domain stretching factor β. max Then determine whether the currently calculated time-domain stretching factor β is greater than 1;

[0103] If the currently calculated time-domain stretching factor β is greater than the initially defined maximum time-domain stretching factor β max The output is the calculated minimum normalized mean square error of energy.

[0104] The step of determining whether the currently calculated time-domain stretching factor β is greater than 1 includes:

[0105] If the calculated time-domain stretching factor β is not greater than 1, it means that the measurement signal is stretched. Then, cubic spline interpolation is performed on the measurement signal to achieve time-domain stretching, and the normalized energy mean square error between the stretched measurement signal and the reference signal is calculated.

[0106] If the calculated time-domain stretching factor β is greater than 1, it indicates that the measured signal is compressed. Then, cubic spline interpolation is performed on the reference signal to achieve time-domain stretching, and the normalized energy mean square error between the measured signal and the stretched reference signal is calculated.

[0107] Finally, the normalized energy mean square error calculated in each iteration is evaluated. Is it greater than the minimum normalized energy mean square error defined in the initialization?

[0108] If the normalized energy mean square error after stretching calculation The minimum normalized energy mean square error greater than the initial definition Then, iterate through the values ​​of the time-domain stretching factor β with a step size μ, i.e., β = β + μ; then continue to return to check whether the currently calculated time-domain stretching factor β is greater than the initialized maximum value β of the time-domain stretching factor. max This step;

[0109] If the normalized energy mean square error after stretching calculation Not greater than the minimum normalized energy mean square error defined in the initialization. Then the normalized energy mean square error at this time The minimum normalized mean square error of energy, i.e. Then, it returns to check whether the currently calculated time-domain stretching factor β is greater than the initialized maximum time-domain stretching factor β. max This step.

[0110] Among them, the normalized energy mean square error obtained after stretching calculation The calculation process includes: time-domain stretching, mean removal, and normalization.

[0111] Time-domain stretching: In order to reduce or eliminate signal differences caused by wave velocity variations, cubic spline interpolation is performed on the reference signal or measurement signal to achieve time-domain stretching, thereby ensuring that the data lengths of the reference signal and the measurement signal are consistent after time-domain stretching.

[0112] Mean removal processing: The reference signal and the measurement signal after the time-domain stretching step are respectively subjected to mean removal processing to eliminate the interference of DC component and obtain the mean-removed reference signal and the mean-removed measurement signal.

[0113] Normalization processing: Energy normalization processing is performed on the reference signal and the measurement signal after the mean removal processing step, which can eliminate the interference caused by the overall linear change of the signal amplitude with temperature, and obtain the normalized reference signal and the normalized measurement signal.

[0114] Based on the normalized reference signal and the normalized measurement signal, the normalized energy root mean square error is obtained. This normalized energy root mean square error reflects the difference between the energy-normalized reference signal and the measurement signal. It can be used as a criterion for determining the optimal tensile factor, and also as a characteristic parameter for determining the presence of damage in nondestructive testing. The β value corresponding to the minimum normalized energy root mean square error between the measurement signal and the reference signal is the optimal tensile factor.

[0115] Calculate the normalized energy mean square error The specific judgment process for time-domain stretching is as follows:

[0116] During the first loop calculation:

[0117] If the time-domain stretching factor β≤1, the first stretching transformation is performed on the measurement signal. The mean-removal and normalization processes are then performed on the measurement signal after the first stretching and the reference signal, respectively, until the normalized energy mean square error is obtained.

[0118] If the time-domain stretching factor β > 1, perform the first stretching transformation on the reference signal, and then perform de-meaning and normalization processing on the measured signal and the reference signal after the first stretching transformation respectively, until the normalized energy mean square error is obtained.

[0119] During the second loop calculation:

[0120] If β≤1 in the first loop and β≤1 in the second loop, perform a second stretching transformation on the measurement signal. Then, perform mean-reduction and normalization processing on the measurement signal and the reference signal after the second stretching transformation, respectively, until the normalized energy mean square error is obtained.

[0121] If β≤1 in the first loop judgment and β>1 in the second loop judgment, the reference signal is subjected to the first stretching transformation. The measured signal after the first stretching transformation and the reference signal after the first stretching transformation are subjected to de-meaning and normalization processing respectively until the normalized energy mean square error is obtained.

[0122] If β > 1 in the first loop judgment and β ≤ 1 in the second loop judgment, the measurement signal is subjected to the first stretching transformation. The measurement signal after the first stretching transformation and the reference signal after the first stretching transformation are subjected to de-meaning processing and normalization processing respectively until the normalized energy mean square error is obtained.

[0123] If β > 1 in the first loop judgment and β > 1 in the second loop judgment, a second stretching transformation is performed on the reference signal. The measured signal and the reference signal after the second stretching transformation are respectively subjected to de-meaning processing and normalization processing until the normalized energy mean square error is obtained. ...

[0125] The minimum normalized energy mean square error is obtained after the nth iteration.

[0126] Among them, the following is obtained from the derivation of the analog signal:

[0127] The measurement signal after stretching transformation is:

[0128]

[0129] Where u'1(t) represents the measured signal after stretching, u1(βt) represents the measured signal after stretching, β is the time-domain stretching factor, t represents the time variable, and λ i Let s represent the reflection coefficient of the echo from the i-th rail feature structure, s represent the sinusoidal signal modulated by the Hanning window transmitted by the transducer, and t represent the reflection coefficient of the echo from the i-th rail feature structure. i This represents the arrival time of the i-th characteristic echo.

[0130] During data processing, if β > 1, it indicates that the measured signal is compressed. When compared with the reference signal, this can be equivalent to stretching the reference signal by a stretching factor of 1 / β. Therefore, the reference signal after stretching transformation obtained from the derivation of the analog signal is:

[0131] u'0(t)=u0[(1 / β)·t]

[0132] Where u'0(t) represents the stretched reference signal, u0 represents the reference signal, β is the time-domain stretching factor, and t represents the time variable.

[0133] The actual acquired digital signals are processed to obtain:

[0134] The mean-reduced reference signal and the mean-reduced measurement signal are obtained using the following formula:

[0135]

[0136] Among them, u' 0m (n) represents the baseline signal after mean removal processing, u' 1m u'0(n) represents the measurement signal after mean removal processing, u'1(n) represents the reference signal after time-domain stretching processing, N is the length of the acquired digital signal, and n is the number of sampling sequence points.

[0137] The normalized reference signal and the normalized measurement signal are obtained using the following formula:

[0138]

[0139] Among them, u' 0mn (n) represents the normalized reference signal, u' 1mn (n) represents the normalized measurement signal, u' 0m (n) represents the baseline signal after mean removal processing, u' 1m (n) represents the measurement signal after mean removal processing, where N is the length of the acquired digital signal and n is the number of sampling sequence points.

[0140] The normalized energy mean square error is obtained by the following formula:

[0141]

[0142] in, U' represents the normalized mean squared energy deviation. 0mn (n) represents the normalized reference signal, u' 1mn (n) represents the normalized measurement signal, where N is the length of the acquired digital signal and n is the number of sampling sequence points.

[0143] This invention also discloses a rail ultrasonic guided wave signal processing system for variable temperature environments, such as... Figure 3 As shown, it includes a data acquisition unit and a computational acquisition unit.

[0144] The acquisition unit is used to acquire reference signals from the rail under non-destructive conditions and measurement signals from the rail under the current conditions.

[0145] The calculation and acquisition unit is used to obtain the minimum normalized energy mean square error and the optimal tensile factor between the reference signal and the measurement signal in order to determine whether there is damage to the rail.

[0146] Specifically, the acquisition unit is used to excite the transducer to generate an ultrasonic guided wave signal by using a Hanning window modulation signal based on the transducer installed on the rail.

[0147] The acquisition unit is used to receive echo signals from the rail using ultrasonic sensors installed on the rail; wherein the echo signal includes a reference signal and a measurement signal, the reference signal representing the echo signal obtained at temperature T0, and the measurement signal representing the echo signal obtained after a temperature change δT.

[0148] The calculation and acquisition unit is used to acquire the minimum normalized energy mean square error and the optimal time-domain stretching factor between the reference signal and the measurement signal, including:

[0149] The calculation unit is used to determine the extreme range of the time-domain stretching factor β [β]. min ,β max and minimum normalized energy mean square error Perform initialization definition;

[0150] The calculation unit is used to calculate the extreme value range of the time-domain stretching factor β [β]. min ,β max Within this range, the time-domain stretching factor β is iterated through with a step size to stretch the reference signal or the measured signal and calculate the normalized energy mean square error between the two until the minimum normalized energy mean square error is obtained. and the optimal time-domain stretching factor; where the optimal time-domain stretching factor is the minimum normalized energy mean square error. The corresponding time-domain stretching factor β.

[0151] Regarding the system in the above embodiments, the specific manner in which each unit module performs operations has been described in detail in the embodiments related to the method, and will not be elaborated here.

[0152] Experimental verification was conducted on the ultrasonic guided wave signal processing method for rails under variable temperature conditions proposed in this application:

[0153] After installing transducers and sensors on the rail, a 30kHz Hanning window modulation signal was used as excitation to collect guided wave signal data for four days, totaling 346 sets. Before the fourth day of testing (before collecting the 213th set of signals), a crack of approximately 2.8cm was created at the bottom of the rail. Simultaneously, temperature information of the rail was collected using temperature sensors; the total temperature fluctuation range over the four days was between 11.5 and 38.6℃.

[0154] The first group of signals is used as the reference signal, and the 212th group of signals is used as the measurement signal. Figure 4 This is a comparison of local waveforms of the reference signal and the measured signal; due to temperature changes, the wave velocity of the guided wave in the rail changes, and the amplitude of the guided wave signal also changes to some extent, such as... Figure 4 It can be seen that the amplitude fluctuation of the measured signal after temperature change is significantly smaller than that of the reference signal. The normalized mean square error of the energy between the measured signal and the reference signal is calculated, and the results are as follows: Figure 5 As shown, during the test on the second day, due to the increase in temperature, the normalized energy mean square error increased significantly. The numerical changes caused by temperature were difficult to distinguish from the numerical changes caused by the rail bottom crack, and the existence of the rail bottom crack could not be detected.

[0155] To reduce the impact of temperature on damage detection, time-domain stretching temperature compensation was applied to the data. The stretching factor β was iterated within the range of 0.99 to 1.01 with a step size of 0.001. The optimal stretching factor was obtained when the normalized energy mean square error was minimized, thus achieving wave velocity compensation. Figure 6 The waveforms of the 212th group of measured signals after compensation were compared with the reference signal. Simultaneously, the normalized energy root mean square error can be used as a parameter to determine whether the rail is damaged. Figure 7 To compensate for the normalized energy mean square error between each set of data and the reference signal, it can be seen that after processing by the method of this application, the energy mean square error increases after the temperature rises on the second day, but is still below 0.5. After manufacturing the rail bottom defect, the energy mean square error is higher than 0.75, which is clearly distinguishable from the numerical change caused by temperature, and can achieve accurate monitoring of damage.

[0156] This invention addresses the problems existing in current ultrasonic guided wave monitoring technology for rails. It proposes to process the digital signal in the time domain using cubic spline interpolation to stretch the signal, compensating for the influence of temperature on the guided wave velocity of the rail, thus reducing the complexity of the algorithm. Simultaneously, it introduces normalized energy mean square error as a criterion for selecting the optimal stretching factor and a standard for judging rail damage. While achieving temperature compensation, it eliminates the interference of signal amplitude changes caused by temperature or system voltage fluctuations on rail damage identification, ultimately achieving accurate and reliable monitoring of rail damage over a wide temperature range.

[0157] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for processing ultrasonic guided wave signals of steel rails under variable temperature conditions, characterized in that, include: Collect reference signals from the rail under non-destructive conditions and measurement signals from the rail under the current conditions; The minimum normalized energy mean square error and the optimal time-domain stretch factor between the reference signal and the measurement signal are obtained to determine whether there is damage to the rail. The acquisition of the minimum normalized energy mean square error and the optimal time-domain stretching factor between the reference signal and the measurement signal includes: Time-domain stretching factor The extreme range of [β] min ,β max and minimum normalized energy mean square error min Perform initialization definition; define the time-domain stretching factor. The extreme range of [β] min ,β max Within [the range], the time-domain stretching factor is traversed using a step size. The value is used to perform time-domain stretching on the reference signal or measurement signal and calculate the normalized energy mean square error of both. This continues until the minimum normalized energy mean square error is obtained through iterative calculation. min and the optimal time-domain stretching factor; where the optimal time-domain stretching factor is the minimum normalized energy mean square error. min Corresponding time-domain stretching factor ; in, The reference signal or measurement signal is subjected to time-domain stretching and the normalized energy mean square error of both is calculated. ,include, Time-domain stretching: performing a cubic spline interpolation stretching transformation on a reference signal or measurement signal; Mean removal processing: The mean removal processing is performed on the reference signal and the measurement signal after time-domain stretching respectively; Normalization processing: Energy normalization processing is performed on the mean-removed reference signal and the measurement signal respectively to obtain the normalized reference signal and the normalized measurement signal; The normalized energy mean square error is obtained based on the normalized reference signal and the normalized measurement signal. The measurement signal after the stretching transformation is: In the formula, This indicates the measurement signal after stretching. This indicates the measurement signal after stretching. The time-domain stretching factor. Represents a time variable. Indicates the first The reflection coefficient of the echo from the characteristic structure of a rail. This indicates that the transducer sends a sinusoidal signal modulated by a Hanning window. Indicates the first The arrival time of each characteristic echo; The reference signal after stretching transformation is: In the formula, This is represented as the reference signal after stretching. Represented as a reference signal, The time-domain stretching factor. Represents a time variable.

2. The method for processing ultrasonic guided wave signals of steel rails under variable temperature conditions according to claim 1, characterized in that, The acquisition of reference signals from the rail under non-destructive conditions and current signals from the rail under the current conditions includes, A transducer is installed on the rail, and a Hanning window modulation signal is used to excite the transducer to generate an ultrasonic guided wave signal. An ultrasonic sensor mounted on the rail receives echo signals from the rail; the echo signal includes a reference signal and a measurement signal, the reference signal representing the temperature... The echo signal obtained at that time, the measured signal represents the change in temperature. The echo signal obtained afterward.

3. The method for processing ultrasonic guided wave signals of steel rails under variable temperature conditions according to claim 2, characterized in that, The transducer sends a sinusoidal signal modulated by the Hanning window function to excite the transducer to generate an ultrasonic guided wave signal; The modulated sinusoidal signal is: in, This indicates that the transducer sends a sinusoidal signal modulated by the Hanning window function. This represents the amplitude of the excitation ultrasonic guided wave signal. It is the frequency of the excitation ultrasonic guided wave signal. This represents the Hanning window function. t Represents a time variable.

4. The method for processing ultrasonic guided wave signals of steel rails under variable temperature conditions according to claim 3, characterized in that, The Hanning window function is: in, This represents the Hanning window function. It is the frequency of the excitation signal. It is the number of sinusoidal signal cycles in the Hanning window. t Represents a time variable.

5. A method for processing ultrasonic guided wave signals of steel rails under variable temperature conditions according to claim 3 or 4, characterized in that, The reference signal is: in, Indicates the reference signal. Indicates the first The reflection coefficient of the echo from the characteristic structure of a rail. Indicates the number of received data at time t. A characteristic echo signal, For the first The arrival time of each characteristic echo, This indicates that the transducer sends a sinusoidal signal modulated by a Hanning window, and t represents the time variable.

6. A method for processing ultrasonic guided wave signals of steel rails under variable temperature conditions according to claim 3 or 4, characterized in that, The measurement signal is: in, Indicates the measurement signal. Indicates the first The reflection coefficient of the echo from the characteristic structure of a rail. This indicates that the transducer sends a sinusoidal signal modulated by a Hanning window. Indicates the number of times the temperature changes. The arrival time of each characteristic echo, Indicates the first The arrival time of each characteristic echo, t represents the time-domain stretching factor, where t represents the time variable.

7. The method for processing ultrasonic guided wave signals of steel rails under variable temperature conditions according to claim 1, characterized in that, Initialize and define the time-domain stretching factor value β=β during the first calculation. min Minimum normalized energy mean square error min =10; Determine the time-domain stretching factor currently being calculated. Is it greater than the maximum value of the time-domain stretching factor β defined during initialization? max : If the currently calculated time-domain stretching factor Not greater than the maximum value of the time-domain stretching factor β defined during initialization max Then determine the time-domain stretching factor currently being calculated. Is it greater than 1? If the currently calculated time-domain stretching factor Greater than the maximum value of the time-domain stretching factor β defined during initialization max The output is the calculated minimum normalized mean square error of energy. min .

8. A method for processing ultrasonic guided wave signals of steel rails under variable temperature conditions according to claim 1 or 7, characterized in that, Determine the time-domain stretching factor in the calculation Whether it is greater than 1, in order to perform time-domain stretching, including: If the currently calculated time-domain stretching factor If the value is not greater than 1, it indicates that the measurement signal is stretched. Then, a stretching transformation is performed on the currently calculated measurement signal, and the normalized energy mean square error between the stretched measurement signal and the currently calculated reference signal is obtained. If the currently calculated time-domain stretching factor If the value is greater than 1, it indicates that the measurement signal is compressed. In this case, a stretching transformation is performed on the currently calculated reference signal, and the normalized energy mean square error between the currently calculated measurement signal and the stretched reference signal is calculated.

9. The method for processing ultrasonic guided wave signals of steel rails under variable temperature conditions according to claim 8, characterized in that, The process continues until the minimum normalized energy mean square error is obtained through iterative calculation. min and the optimal time-domain stretching factor, including, Determine the normalized energy mean square error obtained in each iteration of the calculation. Is it greater than the minimum normalized energy mean square error defined in the initialization? min : If the normalized energy mean square error after stretching calculation The minimum normalized energy mean square error greater than the initial definition min Then use step size Traversing the time-domain stretching factor The value of, i.e. Then, it returns to the previous step to determine the currently calculated time-domain stretching factor. Is it greater than the maximum value of the time-domain stretching factor β defined during initialization? max ; If the normalized energy mean square error after stretching calculation Not greater than the minimum normalized energy mean square error defined in the initialization. min Then the normalized mean square error of energy at this time The minimum normalized mean square error of energy, i.e. min = Then, it returns to the previous step to determine the currently calculated time-domain stretching factor. Is it greater than the maximum value of the time-domain stretching factor β defined during initialization? max .

10. A method for processing ultrasonic guided wave signals of steel rails under variable temperature conditions according to claim 1 or 9, characterized in that, The mean-reduced reference signal and the mean-reduced measurement signal are obtained using the following formula: in, This represents the baseline signal after mean removal processing. This represents the measurement signal after mean removal processing. This represents the reference signal after time-domain stretching. This represents the measurement signal after time-domain stretching, where N is the length of the acquired digital signal and n is the number of sampling sequence points.

11. A method for processing ultrasonic guided wave signals of steel rails under variable temperature conditions according to claim 1 or 9, characterized in that, The normalized reference signal and the normalized measurement signal are obtained using the following formula: in, This represents the normalized reference signal. This represents the normalized measurement signal. This represents the baseline signal after mean removal processing. This represents the measurement signal after mean removal processing, where N is the length of the acquired digital signal and n is the number of sampling sequence points.

12. A method for processing ultrasonic guided wave signals of steel rails under variable temperature conditions according to claim 1 or 9, characterized in that, The normalized energy mean square error is obtained by the following formula: in, This represents the normalized mean squared energy. This represents the normalized reference signal. This represents the normalized measurement signal, where N is the length of the acquired digital signal and n is the number of sampling sequence points.

13. A rail ultrasonic guided wave signal processing system for variable temperature environments, characterized in that, It includes a data acquisition unit and a computational acquisition unit. The acquisition unit is used to acquire the reference signal in the rail under non-destructive conditions and the measurement signal in the rail under the current conditions. The calculation and acquisition unit is used to acquire the minimum normalized energy mean square error and the optimal time-domain stretching factor between the reference signal and the measurement signal in order to determine whether there is damage to the rail. The acquisition of the minimum normalized energy mean square error and the optimal time-domain stretching factor between the reference signal and the measurement signal includes: Time-domain stretching factor The extreme range of [β] min ,β max and minimum normalized energy mean square error min Perform initialization definition; define the time-domain stretching factor. The extreme range of [β] min ,β max Within [the range], the time-domain stretching factor is traversed using a step size. The value is used to perform time-domain stretching on the reference signal or measurement signal and calculate the normalized energy mean square error of both. This continues until the minimum normalized energy mean square error is obtained through iterative calculation. min and the optimal time-domain stretching factor; where the optimal time-domain stretching factor is the minimum normalized energy mean square error. min Corresponding time-domain stretching factor ; in, The reference signal or measurement signal is subjected to time-domain stretching and the normalized energy mean square error of both is calculated. ,include, Time-domain stretching: performing a cubic spline interpolation stretching transformation on a reference signal or measurement signal; Mean removal processing: The mean removal processing is performed on the reference signal and the measurement signal after time-domain stretching respectively; Normalization processing: Energy normalization processing is performed on the mean-removed reference signal and the measurement signal respectively to obtain the normalized reference signal and the normalized measurement signal; The normalized energy mean square error is obtained based on the normalized reference signal and the normalized measurement signal. The measurement signal after the stretching transformation is: In the formula, This indicates the measurement signal after stretching. This indicates the measurement signal after stretching. The time-domain stretching factor. Represents a time variable. Indicates the first The reflection coefficient of the echo from the characteristic structure of a rail. This indicates that the transducer sends a sinusoidal signal modulated by a Hanning window. Indicates the first The arrival time of each characteristic echo; The reference signal after stretching transformation is: In the formula, This is represented as the reference signal after stretching. Represented as a reference signal, The time-domain stretching factor. Represents a time variable.

14. The ultrasonic guided wave signal processing system for rails under variable temperature conditions according to claim 13, characterized in that, The acquisition unit is used to acquire reference signals from the rail under non-destructive conditions and measurement signals from the rail under the current conditions, including: The acquisition unit is used to excite the transducer to generate an ultrasonic guided wave signal by using a Hanning window modulation signal based on the transducer installed on the rail. The acquisition unit is used to receive echo signals from the rail using ultrasonic sensors installed on the rail; wherein the echo signal includes a reference signal and a measurement signal, the reference signal representing the temperature... The echo signal obtained at that time, the measurement signal representing the change in temperature The echo signal obtained afterward.

Images