A method for time synchronization processing of ultrasonic guided wave signals of rail

Abstract

The present invention belongs to the technical field of ultrasonic guided wave structural health monitoring, and particularly relates to a method for time synchronization processing of ultrasonic guided wave signals of rail. The present invention adopts GCC-PHAT, that is, the generalized cross-correlation method with the weight of PHAT, adds a weight function on the basis of the cross-correlation function, weights the signal in the power spectral domain, sharpens the peak of the correlation function, improves the effectiveness and accuracy of calculating the time delay amount between the reference signal and the measurement signal, and uses this time delay amount to perform time domain translation on the measurement signal, realizing the time synchronization processing of the ultrasonic guided wave signals of rail, and avoiding the misdetection of rail damage caused by the time delay amount.

Description

Technical Field

[0001] The present invention belongs to the technical field of ultrasonic guided wave structural health monitoring, and particularly relates to a method for time synchronization processing of ultrasonic guided wave signals of rail. Background Art

[0002] With the development of railway towards high-speed passenger transport and heavy-haul freight transport, the requirements for transport safety are getting higher and higher, and the structural health monitoring of rail becomes an important basis for the safe operation of railway. The ultrasonic guided wave structural health monitoring technology is a new type of monitoring method, which can realize on-line real-time monitoring and crack detection of the entire cross-section of rail, and has received extensive attention.

[0003] When using ultrasonic guided wave technology for structural health monitoring of rail, the baseline subtraction method is often used, that is, the measured signal is subtracted from the reference signal, and whether there is damage to the measured rail is judged according to the magnitude of the residual. This puts forward very high requirements for the stability and consistency of the guided wave received signal. However, in practical applications, there is a time delay in the communication between ultrasonic guided wave excitation / reception devices, and the wave velocity of rail guided wave is also easily affected by temperature changes, resulting in unstable time delay in each collected signal. When performing baseline subtraction, due to the difference in the time domain positions of the waveforms of the measured signal and the reference signal, the residual after subtraction is too large, resulting in false detection of rail damage. Therefore, in order to improve the reliability of the ultrasonic guided wave rail structural health monitoring system, the signal time synchronization processing method is a technical problem to be solved urgently.

[0004] At present, for this technical problem, there are mainly two solutions, namely the threshold method and the cross-correlation function method. The threshold method is to trigger monitoring when the amplitude of the received signal exceeds a certain predetermined threshold. The advantage is that no complex calculation is required, but it is easily affected by the on-site environment and equipment noise. The cross-correlation method calculates the cross-correlation function of two signals to find the time offset when the two signals reach the highest similarity. However, when the signal spectrum is narrow, there are easily multiple peaks close to each other in the correlation function and the main peak is not obvious, resulting in errors in the calculated time delay.

[0005] In order to process the problem of signal time asynchronization, the commonly used method in guided wave signal processing is the cross-correlation function method, which provides a measure for the similarity of two signals. For two similar signals, the offset corresponding to the maximum value of the cross-correlation function is the time delay between the two signals. Liu Kangchi from North University of China in his doctoral thesis "Theoretical Research and Equipment Development of Magnetostrictive Ultrasonic Guided Wave Distributed Temperature Measurement" calculated the time delay between the transmitted-reflected signals through the correlation function to obtain the sound velocity of the ultrasonic guided wave propagating in the measured object, and measured the temperature of the object by using the corresponding relationship between temperature and sound velocity in the object. However, when the signal spectrum is narrow, there are multiple peaks in the correlation function and the main peak is not obvious, reducing the effectiveness and reliability of the time delay calculation. Summary of the Invention

[0006] In view of the problems existing in the above-mentioned ultrasonic guided wave rail health monitoring, the present invention provides a method for time synchronization processing of ultrasonic guided wave signals of rails, so as to improve the effectiveness and accuracy of calculating the time delay between the reference signal and the measurement signal, realize the time synchronization processing of ultrasonic guided wave signals of rails, and avoid misdetection of rail damage caused by the time delay.

[0007] The object of the present invention is achieved as follows. It relates to a method for time synchronization processing of ultrasonic guided wave signals of rails, which is characterized by including the following steps:

[0008] Step 1: When conducting rail structural health monitoring, the transmitting module excites an ultrasonic transducer installed at the rail web to transmit an ultrasonic signal modulated by a Hanning window. At the same time, the transmitting module sends an enabling signal to the receiving module, and the receiving module starts to collect the output signal of the ultrasonic sensor installed at the rail web. Among them, the transducer transmits a sinusoidal signal s(t) modulated by a Hanning window, and this signal can be expressed as:

[0009] s(t) = A0w(t)sin(2πft)

[0010] where w(t) is the Hanning window function:

[0011]

[0012] In the formula, A0 is the amplitude of the excitation signal, f is the frequency of the excitation signal, and k is the number of sine signal periods in the modulation signal.

[0013] Step 2: When the rail structure is intact and the system starts to work, the receiving module collects the rail guided wave signal u0(n) as the reference signal. After a period of time, it collects the rail guided wave signal u1(n) at the current moment as the measurement signal.

[0014] Step 3: In order to avoid the interference of the DC component in the signal and the external environmental noise, first preprocess the reference signal and the measurement signal. Remove the DC component through de-mean processing, and use a high-pass Chebyshev type II filter to filter out low-frequency interference. After preprocessing, the reference signal and the measurement signal are u′0(n) and u′1(n).

[0015] Step 4: Calculate the Generalized Cross Correlation (GCC) function of the preprocessed reference signal and measurement signal:

[0016]

[0017] where ω is the angular frequency, U0(ω) and U1(ω) respectively represent the Fourier transforms of the signals u′0(n) and u′1(n), is the conjugate of U1(ω). Ψ0，1 (ω) is the Phase Transform (PHAT) weight function and can be expressed as:

[0018]

[0019] Step 5: Find the maximum value of the generalized cross-correlation function, and the corresponding independent variable τ m is the time delay amount between the signals u′0(n) and u′1(n). By shifting the measured signal in the time domain by τ m , time calibration is performed to achieve synchronization of the reference signal and the measured signal in the time domain. The shifted signal can be expressed as:

[0020] u′ 1t (n) = u′1(n + τ m )

[0021] Step 6: For the calibrated reference signal and measured signal, in order to avoid excessive residuals between the two signals caused by local waveform phase differences, perform Hilbert transform to obtain the envelopes of the two signals, denoted as H0(n) and H1(n). After subtracting the signal envelopes, the residual function can be expressed as:

[0022] Res(n) = H1(n) - H0(n)

[0023] Find the maximum value R m of the residual function, and compare it with the rail damage threshold. If it is greater than the threshold, it is considered that the rail has damage; if it is less than the threshold, it is considered that the rail has no damage.

[0024] Adopt GCC-PHAT, that is, the generalized cross-correlation method with the weight of PHAT. Add a weight function to the cross-correlation function to sharpen the peak of the correlation function, and use this time delay amount to perform time domain translation on the measured signal, realizing the time synchronization processing of the rail ultrasonic guided wave signal.

[0025] In the existing methods, for two similar signals, the offset corresponding to the maximum value of the cross-correlation function is the time delay between the two signals. However, for signals with a narrow spectrum, there will be multiple similar peaks in the result, resulting in errors in the calculation results. In the present invention, GCC-PHAT is used to process the reference signal and the measured signal, and the signal is weighted in the power spectral domain, making the peak of the correlation function at the time delay more prominent, improving the accuracy and reliability of the time delay amount calculation.

[0026] The advantages of the present invention are as follows: The present invention adopts GCC-PHAT, that is, the generalized cross-correlation method with the weight of PHAT, adds a weight function on the basis of the cross-correlation function, weights the signal in the power spectral domain, sharpens the peak of the correlation function, improves the effectiveness and accuracy of calculating the time delay between the reference signal and the measurement signal, and uses this time delay to perform time domain translation on the measurement signal, realizing the time synchronization processing of the rail ultrasonic guided wave signal and avoiding the misdetection of rail damage caused by the time delay. Description of the Drawings

[0027] The present invention will be further described below in conjunction with the embodiments and the drawings:

[0028] Figure 1 is the flowchart of the embodiment of the present invention;

[0029] Figure 2 is the time domain waveform of the reference signal and the measurement signal;

[0030] Figure 3 is the cross-correlation function waveform of the reference signal and the measurement signal;

[0031] Figure 4 is the GCC-PHAT of the reference signal and the measurement signal. Detailed Embodiment

[0032] To further elaborate on the technical means adopted by the invention to achieve the predetermined purpose, the specific implementation manners, structural features and their effects of the present invention will be described in detail below in conjunction with the drawings and embodiments.

[0033] As Figure 1 shown, the present invention relates to a method for time synchronization processing of rail ultrasonic guided wave signals, which is characterized by including the following steps:

[0034] Step 1: When performing rail structural health monitoring, the transmitting module excites the ultrasonic transducer installed at the rail web to send an ultrasonic signal modulated by a Hanning window; at the same time, the transmitting module sends an enabling signal to the receiving module, and the receiving module starts to collect the output signal of the ultrasonic sensor installed at the rail web.

[0035] Among them, the transducer sends a sine signal s(t) modulated by a Hanning window, and this signal can be expressed as:

[0036] s(t) = A0w(t)sin(2πft)

[0037] Among them, w(t) is the Hanning window function:

[0038]

[0039] Wherein, A0 is the amplitude of the excitation signal, f is the frequency of the excitation signal, and k is the number of sine signal periods in the modulation signal.

[0040] Step 2: When the rail structure is intact and the system starts to work, the receiving module acquires the rail guided wave signal u0(n) as the reference signal. After a period of time, the rail guided wave signal u1(n) at the current moment is collected as the measurement signal.

[0041] Step 3: To avoid the interference of the DC component in the signal and the external environmental noise, the reference signal and the measurement signal are preprocessed first. The DC component is removed through mean removal processing, and the low-frequency interference is filtered out by using a high-pass Chebyshev type II filter. The preprocessed reference signal and measurement signal are u′0(n) and u′1(n).

[0042] Step 4: Calculate the Generalized Cross Correlation (GCC) function of the preprocessed reference signal and measurement signal:

[0043]

[0044] Wherein, ω is the angular frequency, U0(ω) and U1(ω) respectively represent the Fourier transforms of the signals u′0(n) and u′1(n), is the conjugate of U1(ω). Ψ 0，1 (ω) is the Phase Transform (PHAT) weight function and can be expressed as:

[0045]

[0046] Step 5: Find the maximum value of the generalized cross correlation function, and the corresponding independent variable τ m is the time delay amount between the signals u′0(n) and u′1(n). By translating the measurement signal in the time domain by τ m , time calibration is performed, and the synchronization of the reference signal and the measurement signal in the time domain can be achieved. The translated signal can be expressed as:

[0047] u′ 1t (n) = u′1(n + τ m )

[0048] Step 6: For the calibrated reference signal and measurement signal, to avoid too large a residual between the two signals caused by the local waveform phase difference, perform the Hilbert transform to obtain the envelopes of the two signals, denoted as H0(n) and H1(n). After subtracting the signal envelopes, the residual function can be expressed as:

[0049] Res(n) = H1(n) - H0(n)

[0050] Find the maximum value \(R\) of the residual function m , and compare it with the rail damage threshold. If it is greater than the threshold, it is considered that the rail has damage; if it is less than the threshold, it is considered that the rail has no damage.

[0051] Adopt GCC-PHAT, that is, the generalized cross-correlation method with the weight of PHAT, add a weight function on the basis of the cross-correlation function to sharpen the peak of the correlation function, and use this time delay amount to perform time domain translation on the measurement signal, realizing the time synchronization processing of the rail ultrasonic guided wave signal.

[0052] For two similar signals, the offset corresponding to the maximum value of the cross-correlation function is the time delay between the two signals.

[0053] However, for signals with a narrow spectrum, there will be multiple similar peaks in the result of this method, resulting in errors in the calculation result.

[0054] In the present invention, GCC-PHAT is used to process the reference signal and the measurement signal, and the signal is weighted in the power spectral domain, making the peak of the correlation function at the time delay more prominent, and improving the accuracy and reliability of the time delay amount calculation.

[0055] As Figure 2 shown, the time synchronization processing method of the rail ultrasonic guided wave signal proposed in the present invention is experimentally verified. Transducers and sensors are installed on the rail web. The transmitting module excites a guided wave signal with a frequency of 40 kHz and Hanning window modulation through the transducer, and the receiving module acquires the reference signal and the measurement signal.

[0056] It can be seen that the starting positions of the waveforms between the reference signal and the measurement signal are different, there is a delay amount between them, and the corresponding number of delay points is 231 points.

[0057] After preprocessing, directly calculate the cross-correlation function \(R_0(τ)\) of the two signals, and the result is as Figure 3 shown. In the figure, although the value of the independent variable \(τ\) corresponding to the maximum value of \(R_0(τ)\) is 231, which is consistent with the actual delay amount, its peak is not obvious, and there is also a peak with a relatively close magnitude at 210 points, which will reduce the reliability of the time delay amount calculation.

[0058] Calculate the generalized cross-correlation function (GCC-PHAT) with the weight of PHAT, and the result is as Figure 4 shown. In the figure, the value of the independent variable \(τ\) corresponding to the maximum value of \(R(τ)\) is also 231. Compared with the cross-correlation function, the peak of GCC-PHAT is more obvious, and there is only a large main peak at 231 points, improving the effectiveness and reliability of the time delay amount calculation.

[0059] The above content is a further detailed description of the present invention in combination with specific implementation manners. It cannot be determined that the specific implementation of the present invention is only limited to these descriptions. For those of ordinary skill in the technical field to which the present invention pertains, without departing from the concept of the present invention, several simple deductions or substitutions can still be made, and all should be regarded as belonging to the protection scope of the present invention.

Claims

1. A method for time synchronization processing of ultrasonic guided wave signals of steel rails, characterized in that: It includes the following steps: Step 1: The transmitting module stimulates the ultrasonic transducer to transmit an ultrasonic signal modulated by a Hanning window; Meanwhile, the transmitting module sends an enabling signal to the receiving module; The receiving module starts to collect the output signal of the ultrasonic transducer; Step 2: When the rail structure is intact and the system starts to work, the receiving module acquires the rail guided wave signal as a reference signal; After a period of time delay, the receiving module collects the rail guided wave signal at the current moment as the measurement signal; Step 3: First, preprocess the reference signal and the measurement signal, and remove the DC component through de-mean processing; The low-frequency interference is filtered out by using a high-pass Chebyshev type-II filter. After preprocessing, the reference signal and the measurement signal are and ; Step 4: Calculate the generalized cross-correlation function of the preprocessed reference signal and measurement signal; Step 5: Find the maximum value of the generalized cross-correlation function, and the corresponding independent variable is the signal and the time delay between them. By translating the measured signal in the time domain by , time calibration can be performed to achieve synchronization of the reference signal and the measured signal in the time domain. The translated signal can be expressed as: Step 6: Perform Hilbert transform on the calibrated reference signal and measurement signal to obtain the envelopes of the two signals, denoted as and ; After subtracting the signal envelopes, a residual function is obtained, which can be expressed as: Find the maximum value of the residual function , compare it with the rail damage threshold. If it is greater than the threshold, it is considered that the rail has damage; if it is less than the threshold, it is considered that the rail has no damage; In the described step 1, the transducer transmits a sine signal modulated by a Hanning window , which can be expressed as: Among them, is the Hanning window function: wherein, is the amplitude of the excitation signal, is the frequency of the excitation signal, is the number of sine signal periods in the modulation signal; The generalized cross-correlation function in Step 4; Among them, is the angular frequency, and respectively represent the Fourier transforms of signals and , is 's conjugate, is the phase conversion weight function and can be expressed as: .

2. A method for time synchronization processing of rail ultrasonic guided wave signals according to claim 1, characterized in that: The time delay for a period corresponds to 231 delay points.

Images