Methods and systems for monitoring faults in wires and cables

Abstract

This invention relates to the field of wire and cable fault monitoring technology, and discloses a method and system for wire and cable fault monitoring. The method includes transmitting light pulses and recording the transmission time, and acquiring Rayleigh scattering signals for preprocessing to obtain a denoised signal waveform; performing frequency domain transformation on the denoised signal waveform to obtain a frequency domain signal, detecting abnormal peaks in the frequency domain signal, extracting the narrowband time domain signals corresponding to the abnormal peaks, and performing attenuation fitting to obtain an attenuation time constant; when the attenuation time constant is less than a preset time threshold, performing defect analysis based on the narrowband time domain signal and the transmission time to obtain the preliminary defect location; performing error correction based on the preliminary defect location to obtain a high-precision defect location; mapping the narrowband time domain signal of the high-precision defect location into an image matrix, and performing visualization rendering to generate a visual distribution map. This method can transform fault signals into accurate information about the internal structure of defects.

Description

Technical Field

[0001] This invention relates to the field of wire and cable fault monitoring technology, and in particular to a method and system for wire and cable fault monitoring. Background Technology

[0002] In a current-based fault monitoring technology for wires and cables, a pulsed current method is employed. This method applies a high-voltage pulse to the faulty cable, causing a breakdown discharge at the fault point. The weak traveling wave signal generated by the discharge is then captured using coil coupling. By measuring the time difference between the traveling wave and the fault point, and combining this with the known wave velocity, the distance to the fault point is calculated. However, the traveling wave signal suffers from severe attenuation and distortion over long distances, and its starting point is easily submerged by noise, leading to errors in the time difference measurement. This method can only provide a suspected range of the defect. In practice, personnel still need to conduct a secondary inspection within this range on-site and manually assess the severity and internal morphology of the defect before a repair plan can be developed and implemented.

[0003] Existing technologies have the problem of difficulty in converting fault signals into accurate information on the internal structure of defects, resulting in poor reliability of fault monitoring results. Summary of the Invention

[0004] This invention provides a method and system for monitoring faults in electric wires and cables, in order to solve the problem of difficulty in converting fault signals into accurate information on the internal structure location of defects.

[0005] In a first aspect, to solve the above-mentioned technical problems, the present invention provides a method for monitoring faults in wires and cables, comprising: An optical pulse is emitted into the optical fiber in the cable, the emission time is recorded, and the Rayleigh scattering signal is obtained. The Rayleigh scattering signal is then digitally filtered to obtain an initial signal sequence. The initial signal sequence is denoised to obtain the denoised signal waveform; The denoised signal waveform is transformed into a frequency domain to obtain a frequency domain signal. Peak detection is performed on the frequency domain signal to obtain abnormal peaks. The narrowband time domain signal corresponding to the abnormal peaks is extracted. The envelope of the narrowband time domain signal is attenuated and fitted to obtain the attenuation time constant. When the decay time constant is less than a preset time threshold, the signal distortion point is determined based on the narrowband time domain signal, the time difference between the transmission time and the signal distortion point is calculated, and the preliminary location of the defect is obtained based on the time difference and the signal propagation speed. Based on the initial location of the defect, an error correction is performed to obtain the high-precision defect location; The narrowband time-domain signal at the high-precision defect location is mapped into an image matrix; Visualization rendering is performed based on the image matrix to generate a visual distribution map.

[0006] In one optional implementation, the step of digitally filtering the Rayleigh scattering signal to obtain an initial signal sequence includes: The Rayleigh scattering signal is photoelectrically converted to obtain an analog voltage signal; The analog voltage signal is digitally processed to obtain time series data; The time series data is filtered for noise using a digital filter to obtain an initial signal sequence.

[0007] In one optional implementation, the step of denoising the initial signal sequence to obtain the denoised signal waveform includes: The initial signal sequence is subjected to interference signal separation to obtain the main signal components; The main signal components are decomposed into time-frequency features to obtain multi-scale features; The signal is reconstructed based on the multi-scale features to obtain the denoised signal waveform.

[0008] In one optional implementation, the step of obtaining abnormal peak values ​​based on the frequency domain signal by peak detection includes: If the power density of each frequency point of the frequency domain signal is greater than the power density of the two adjacent frequency points, then the frequency point is taken as a candidate peak. Calculate the average power density of all frequency points in the frequency domain signal, and iterate through each candidate peak to see if the power density corresponding to each candidate peak is greater than the average power density. If so, the candidate peak is taken as a high power peak. Perform Hilbert transform on the denoised signal waveform to extract the signal envelope; Determine the envelope amplitude of each frequency point based on the signal envelope, and calculate the reflection intensity ratio of each frequency point in the frequency domain signal. Calculate the average reflection intensity ratio of all frequency points in the frequency domain signal, iterate through each high-power peak to see if the reflection intensity ratio is greater than the average reflection intensity ratio, and if so, treat the high-power peak as an abnormal peak and generate an abnormal peak list.

[0009] In one optional implementation, the step of extracting the narrowband time-domain signal corresponding to the abnormal peak and performing attenuation fitting on the narrowband envelope of the narrowband time-domain signal to obtain the attenuation time constant includes: For each abnormal peak in the list of abnormal peaks, a narrowband time-domain signal is extracted from the denoised signal waveform, with the abnormal peak frequency as the center. Perform a Hilbert transform on the narrowband time-domain signal to extract the narrowband envelope; On the narrowband envelope, a data segment is selected from the peak value down to 1 / e of the peak value. The data segment is then fitted with an exponential decay function to obtain the corresponding narrowband decay time constant. If there are multiple abnormal peaks in the list of abnormal peaks, repeat the above extraction and fitting operations for each abnormal peak to obtain the decay time constant corresponding to each abnormal peak.

[0010] In one optional implementation, when the attenuation time constant is less than a preset time threshold, determining the signal distortion point based on the narrowband time-domain signal, calculating the time difference between the transmission time and the signal distortion point, and obtaining the preliminary location of the defect based on the time difference and the signal propagation speed, includes: The decay time constant corresponding to each abnormal peak is compared with a preset time threshold. If the decay time constant of an abnormal peak is less than the preset time threshold, the position where the signal amplitude drops the most in the narrowband time domain signal corresponding to the abnormal peak is marked as the distortion time point. Subtracting the transmission time from the distortion time point yields the signal propagation two-way time, and taking half of the signal propagation two-way time yields the one-way propagation time. The initial location of the defect is obtained by multiplying the one-way propagation time by the signal propagation speed.

[0011] In one optional implementation, the step of correcting the error based on the preliminary defect location to obtain a high-precision defect location includes: Continuously emit light pulses, and independently perform the above steps to obtain the preliminary location of the defect for each light pulse, and use the preliminary location of the defect obtained for each light pulse as the measurement value; The measured value is optimally estimated using a Kalman filter, and the output of the Kalman filter is used as the high-precision defect location.

[0012] In one optional implementation, the step of mapping the narrowband time-domain signal based on the high-precision defect location into an image matrix includes: The amplitude of the narrowband time-domain signal at the high-precision defect location is normalized to obtain the normalized amplitude. The normalized amplitude is binned to obtain the binning result for each sampling point of the narrowband time-domain signal; The number of bin transitions between adjacent sampling point pairs is counted to construct a Markov transition probability matrix, and an image matrix is ​​formed based on the Markov transition probability matrix.

[0013] In one optional implementation, the step of performing visualization rendering based on the image matrix to generate a visualization distribution map includes: The probability values ​​of the image matrix are linearly mapped to grayscale ranges to obtain a grayscale image matrix; Edge detection is performed on the grayscale image matrix to extract the geometric contours of the regions; Establish the mapping relationship between grayscale values ​​and colors in the grayscale image matrix; Each pixel in the grayscale image matrix is ​​traversed, filled with color according to the mapping relationship, and the geometric contour of the region is superimposed to generate a visual distribution map.

[0014] Secondly, the present invention provides a wire and cable fault monitoring system, comprising: A digital filtering module is used to transmit light pulses into the optical fiber in the cable, record the transmission time and acquire the Rayleigh scattering signal, and perform digital filtering on the Rayleigh scattering signal to obtain an initial signal sequence. The signal denoising module is used to denoise the initial signal sequence to obtain the denoised signal waveform. The attenuation parameter extraction module is used to perform frequency domain transformation on the denoised signal waveform to obtain a frequency domain signal, perform peak detection on the frequency domain signal to obtain abnormal peaks, extract the narrowband time domain signal corresponding to the abnormal peaks, and perform attenuation fitting on the envelope of the narrowband time domain signal to obtain the attenuation time constant. The defect location module is used to determine the signal distortion point based on the narrowband time domain signal when the attenuation time constant is less than a preset time threshold, calculate the time difference between the transmission time and the signal distortion point, and obtain the preliminary location of the defect based on the time difference and the signal propagation speed. The defect image output module is used to perform error correction based on the preliminary defect position to obtain a high-precision defect position; map the narrowband time-domain signal of the high-precision defect position into an image matrix; and perform visualization rendering based on the image matrix to generate a visualization distribution map.

[0015] Compared with the prior art, the present invention has the following beneficial effects: (1) This invention acquires Rayleigh scattering signals and sequentially performs digital filtering, denoising, frequency domain analysis, and attenuation fitting. This enables the extraction of weak vibration signals from a strong noise background and the calculation of the attenuation time constant as a basis for defect judgment. Simultaneously, when the attenuation time constant exceeds a preset threshold, the localization process is automatically triggered. The above operations improve the reliability of the signal, realize the preliminary detection of defects, provide a high-quality data foundation for the subsequent generation of a visual distribution map of defects, and enhance the reliability of fault monitoring results.

[0016] (2) This invention identifies abnormal peaks through frequency domain peak detection, extracts narrowband time-domain signals from abnormal peaks for attenuation fitting and location calculation, achieving preliminary defect location. Furthermore, a Kalman filter is used to optimize the estimation of multiple measurement results, effectively suppressing random errors caused by factors such as system clock jitter and fiber refractive index fluctuations, significantly improving location accuracy. This achieves precise defect location, providing accurate defect location information for subsequent defect visualization distribution maps, further enhancing the reliability of fault monitoring results.

[0017] (3) This invention maps narrowband time-domain signals into a two-dimensional image matrix through Markov transfer field coding, transforming the temporal dynamic features of a one-dimensional signal into two-dimensional spatial features. Then, a visual distribution map is generated through grayscale mapping, edge detection, and color rendering. Through the above operations, Markov transfer field coding is based on probability statistics, making the extracted signal features stable and reliable. At the same time, the generated image integrates precise positioning information and signal feature information, which can transform abstract fault signals into intuitive color images. The resulting visual distribution map can clearly show the boundary range, depth level, etc. of defects, greatly improving the reliability of fault monitoring results. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the wire and cable fault monitoring method provided in the first embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of the wire and cable fault monitoring system provided in the second embodiment of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Reference Figure 1 The first embodiment of the present invention provides a method for monitoring wire and cable faults, including the following steps: S11, emit an optical pulse into the optical fiber in the cable, record the emission time and acquire the Rayleigh scattering signal, and perform digital filtering on the Rayleigh scattering signal to obtain an initial signal sequence; S12, the initial signal sequence is denoised to obtain the denoised signal waveform; S13, the denoised signal waveform is transformed into a frequency domain to obtain a frequency domain signal, peak detection is performed on the frequency domain signal to obtain abnormal peaks, the narrowband time domain signal corresponding to the abnormal peaks is extracted, and the envelope of the narrowband time domain signal is attenuated and fitted to obtain the attenuation time constant. S14, when the attenuation time constant is less than the preset time threshold, the signal distortion point is determined according to the narrowband time domain signal, the time difference between the transmission time and the signal distortion point is calculated, and the preliminary location of the defect is obtained according to the time difference and the signal propagation speed. S15, perform error correction based on the preliminary defect location to obtain a high-precision defect location; map the narrowband time-domain signal of the high-precision defect location into an image matrix; perform visualization rendering on the image matrix to generate a visualization distribution map.

[0021] In step S11, it should be noted that firstly, a laser with a wavelength of 1550 nanometers is set in the optical fiber as a light source. A specific wavelength of light signal is reflected by the periodic refractive index change of the fiber Bragg grating. Combined with an optical time-domain reflectometer, light pulses are transmitted at a pulse frequency of 1000 times per second. Simultaneously, the system records the transmission time of these light pulses using a high-precision clock. Next, Rayleigh scattering signals are obtained from inside the cable using the fiber Bragg grating sensing mechanism. Rayleigh scattering signals are backscattered light generated when light propagates in the optical fiber and encounters microscopic inhomogeneities in the fiber medium. When a defect exists at a certain location in the cable, the vibration at that location differs from that at a normal location. When a light pulse passes through this location, the intensity of its Rayleigh scattering signal will change accordingly. The obtained Rayleigh scattering signal data is the backscattered light power at each time point.

[0022] In one implementation, the Rayleigh scattering signal is digitally filtered to obtain an initial signal sequence, including: The Rayleigh scattering signal is photoelectrically converted to obtain an analog voltage signal; The analog voltage signal is digitally processed to obtain time series data; The time series data is filtered for noise using a digital filter to obtain an initial signal sequence.

[0023] It should be noted that the Rayleigh scattering signal is converted into an analog voltage signal using a photodetector. Next, a high-speed data acquisition card is used to sample the analog voltage signal at equal intervals at a set sampling frequency, recording the signal amplitude value at each sampling moment. The sampling frequency is determined based on the required spatial resolution; it should be greater than the ratio of the speed of light in the optical fiber to the spatial resolution. For example, when a spatial resolution of 1 meter is required, the sampling frequency should be no less than 200 MHz. Then, the amplitude values ​​of the sampled discrete-time samples are quantized using an analog-to-digital converter, mapped to corresponding digital codes, resulting in discrete digital quantities. These discrete digital quantities are arranged in chronological order of sampling time to generate time-series data.

[0024] It should be noted that the Butterworth low-pass digital filter can be used for noise filtering. The Butterworth filter has the characteristic of maximum flatness of the amplitude-frequency response within the passband, which can filter out high-frequency noise while preserving the effective signal to the greatest extent. First, the time series data is demodulated using orthogonal demodulation or Hilbert transform demodulation to extract the intensity envelope of the Rayleigh scattering signal, obtaining the envelope time series. Next, a cutoff frequency is set, and the filter coefficients are calculated based on the sampling frequency and the cutoff frequency to determine the filter order, ensuring that the transition band meets the system's requirements for signal distortion. The cutoff frequency is determined based on the frequency range of the weak vibration signal. To preserve the weak vibration signal while filtering out environmental noise above this frequency band to the greatest extent, since the effective vibration signal frequency caused by defects is usually concentrated in the range of 1Hz to 9Hz, this embodiment sets the cutoff frequency to twice the highest frequency of the weak vibration signal, ensuring the integrity of the effective signal while effectively suppressing high-frequency noise components. The envelope time series is input into the digital filter, and convolution is performed using a difference equation to filter out high-frequency noise components with frequencies higher than the set cutoff frequency, preserving the low-frequency effective signal, thus obtaining the initial signal sequence. In the filtered initial signal sequence, the characteristics of weak vibration signals are preserved, and the smoothness of the waveform is significantly improved, providing high-quality input data for subsequent time-frequency analysis.

[0025] In step S12, in one implementation, the initial signal sequence is denoised to obtain a denoised signal waveform, including: The initial signal sequence is subjected to interference signal separation to obtain the main signal components; The main signal components are decomposed into time-frequency features to obtain multi-scale features; The signal is reconstructed based on the multi-scale features to obtain the denoised signal waveform.

[0026] It should be noted that the initial signal sequence is divided into equal-length signal segments according to time, and these segments are arranged as row vectors to form a signal matrix. The mean of each column of the signal matrix is ​​calculated, and each element is subtracted from the mean of its column to obtain a centered matrix. This centering process can eliminate the DC component in the signal, making subsequent analysis more accurate. The centered matrix is ​​transposed, and then multiplied by the original centered matrix. Each element in the product matrix is ​​then divided by the number of signal segments minus one to obtain the covariance matrix. The elements of the covariance matrix reflect the correlation between signals at different time points. Next, eigenvalues ​​and corresponding eigenvectors are obtained from the covariance matrix. The eigenvalues ​​represent the energy magnitude in the direction of the principal component; the larger the eigenvalue, the greater the contribution of the principal component to the original signal. The eigenvectors represent the direction of the principal component. All eigenvalues ​​are arranged in descending order, and the eigenvectors are arranged according to the order of their corresponding eigenvalues.

[0027] Next, principal components are selected based on the cumulative contribution rate, and the sum of all eigenvalues ​​is calculated to obtain the total energy. Then, the eigenvalues ​​are summed sequentially in descending order, and the percentage of the sum to the total energy is calculated; this percentage is the cumulative contribution rate. The top k principal components with a cumulative contribution rate of 95% are retained as the main signal components, while the components corresponding to the remaining eigenvalues ​​are considered noise interference components. The 95% cumulative contribution rate is an empirical threshold, indicating that the retained principal components contain 95% of the energy information of the original signal, which is sufficient to characterize the main features of the signal, while the discarded 5% weak components usually correspond to noise. This threshold can be adjusted according to the actual signal-to-noise ratio: in a strong noise environment, it can be appropriately reduced to 90% to filter out more noise; in a weak noise environment, it can be increased to 98% to retain more details. The centering matrix is ​​multiplied by the eigenvector matrix to obtain the projection coefficient matrix. Then, the projection coefficient matrix is ​​multiplied by the transpose of the retained eigenvector matrix to obtain the reconstructed centering matrix. Finally, the average value of the column previously subtracted is added to each column of the reconstructed centering matrix to recover the actual signal amplitude value. Finally, the reconstructed matrix is ​​expanded row-wise to restore it to a one-dimensional sequence, yielding the main signal components.

[0028] It should be noted that time-frequency decomposition first performs a discrete cosine transform on the main signal components, obtaining a transform coefficient sequence of the same length as the original signal. Each coefficient corresponds to a specific frequency component, and these coefficients are arranged from low to high frequency, with low-frequency coefficients first and high-frequency coefficients last. Next, the time-frequency decomposition scale is set, including the number of decomposition layers and the frequency range corresponding to each layer. Through statistical analysis of a large number of measured weak vibration signals from cables, it was found that three layers are sufficient to distinguish differences in energy distribution, and the signal energy is mainly concentrated in the low-frequency band below 9 Hz. The frequency components above 9 Hz have extremely low energy and have negligible impact on defect detection. Therefore, in this embodiment, the decomposition scale is set to three layers, with the characteristic frequencies of the first layer concentrated between 1 Hz and 3 Hz, the second layer between 3 Hz and 6 Hz, and the third layer between 6 Hz and 9 Hz.

[0029] Next, based on the defined frequency ranges, coefficients belonging to each layer are extracted from the transform coefficient sequence. For a given layer's transform coefficients, its energy is equal to the sum of the squares of all transform coefficients within that layer. Energy reflects the intensity of that frequency component in the original signal; higher energy indicates a more significant frequency component. The percentage of energy in each layer relative to the total energy is then calculated. The frequency range and energy percentage of each layer constitute the multi-scale features. For example, when processing a signal with a sampling rate of 1000 Hz, the calculated energy percentages are: first layer 20%, second layer 60%, and third layer 20%. This identifies the second layer as the primary vibration signal feature, providing a basis for subsequent adaptive thresholding denoising.

[0030] It is worth noting that the layer with the highest energy proportion is identified from the multi-scale features, and the frequency range corresponding to this layer is the frequency band where the main vibration signal is located. The coefficients corresponding to this frequency band are extracted from the transform coefficient sequence, the remaining coefficients are set to zero, and then an inverse discrete cosine transform is performed to obtain the time-domain signal component of this frequency band. Next, sampling points whose amplitude of the time-domain signal component is lower than a preset amplitude threshold are identified as noise components, and the amplitude values ​​of these sampling points are set to zero. Sampling points with amplitudes higher than or equal to the threshold are retained as valid signal components. The processed signal is the preliminary signal waveform after denoising. The preset amplitude threshold is determined based on statistical analysis of a large number of measured weak vibration signals from cables. Statistical analysis of signal amplitudes under different defect types and different noise environments revealed that the amplitude of the valid vibration signal is usually greater than 0.05 volts, while the amplitude of environmental noise is usually less than 0.05 volts. Therefore, this embodiment sets the preset amplitude threshold to 0.05 volts to retain the valid vibration signal while filtering out environmental noise.

[0031] In step S13, it should be noted that the denoised signal waveform is converted into a frequency domain signal through a fast Fourier transform. The frequency domain signal is a data set with frequency as the horizontal axis and power density as the vertical axis, reflecting the distribution of signal energy at different frequencies.

[0032] In one implementation, abnormal peak values ​​are obtained by peak detection based on the frequency domain signal, including: If the power density of each frequency point of the frequency domain signal is greater than the power density of the two adjacent frequency points, then the frequency point is taken as a candidate peak. Calculate the average power density of all frequency points in the frequency domain signal, and iterate through each candidate peak to see if the power density corresponding to each candidate peak is greater than the average power density. If so, the candidate peak is taken as a high power peak. Perform Hilbert transform on the denoised signal waveform to extract the signal envelope; Determine the envelope amplitude of each frequency point based on the signal envelope, and calculate the reflection intensity ratio of each frequency point in the frequency domain signal. Calculate the average reflection intensity ratio of all frequency points in the frequency domain signal, iterate through each high-power peak to see if the reflection intensity ratio is greater than the average reflection intensity ratio, and if so, treat the high-power peak as an abnormal peak and generate an abnormal peak list.

[0033] It should be noted that, firstly, the power density of each frequency point in the frequency domain signal is determined. If the power density of any frequency point is greater than that of the two adjacent frequency points, then that frequency point is considered a candidate peak. This process is repeated for each frequency point in the frequency domain signal to obtain a list of candidate peaks. Candidate peaks are all local maxima in the frequency domain, representing frequency components with relatively concentrated energy in the signal. These frequency components may originate from vibration signals caused by cable defects or from environmental noise, providing a basis for further screening. Next, the power densities of all frequency points in the frequency domain signal are summed and averaged to obtain the average power density. The power density of each candidate peak in the candidate peak list is then determined. If the power density of a candidate peak in the list is greater than the average power density, then that candidate peak is considered a high-power peak. This process is repeated for each candidate peak in the list to obtain a list of high-power peaks. The average power density reflects the overall energy level of the frequency domain signal. Candidate peaks with a power density higher than the average value have energy higher than the average level of the signal and are more likely to originate from effective vibration signals caused by cable defects.

[0034] It should be noted that a Hilbert transform is performed on the denoised signal waveform to obtain a new signal with the same frequency and amplitude as the original signal, but with a 90-degree phase lag. Using the original signal as the real part and the Hilbert-transformed new signal as the imaginary part, an analytic signal is constructed, and its magnitude is calculated to obtain the signal envelope. The signal envelope is a smooth curve connecting the peaks of each oscillation cycle of the signal, reflecting the overall intensity change of the signal. A short-time Fourier transform is performed on the denoised signal waveform to obtain a time-frequency spectrum, determining the time position corresponding to each frequency point. The envelope amplitude at that time point is read from the signal envelope to obtain the envelope amplitude at each frequency point. The ratio of the envelope amplitude to the reference intensity is the reflection intensity ratio. The reference intensity is a normal signal intensity benchmark, which in this embodiment is determined by measuring the signal envelope amplitude in a normal, defect-free area and calculating the average value.

[0035] Next, the average reflection intensity ratio is obtained by summing the reflection intensity ratios of all frequency points in the frequency domain signal. Then, the reflection intensity ratio of each high-power peak in the high-power peak list is analyzed. If the reflection intensity ratio of a high-power peak in the high-power peak list is greater than the average reflection intensity ratio, that high-power peak is considered an abnormal peak. This process is repeated for each high-power peak in the high-power peak list to obtain an abnormal peak list. When a defect exists, it will cause signal reflection, increasing the reflection intensity ratio at that location. The average reflection intensity ratio reflects the overall level of signal reflection. Peaks with a reflection intensity ratio higher than the average indicate that the signal reflection intensity at that frequency point is higher than normal, potentially indicating a cable defect. Finally, each abnormal peak in the abnormal peak list corresponds to a possible cable defect.

[0036] In one implementation, the narrowband time-domain signal corresponding to the abnormal peak is extracted, and the attenuation time constant is obtained by attenuating and fitting the narrowband envelope of the narrowband time-domain signal, including: For each abnormal peak in the list of abnormal peaks, a narrowband time-domain signal is extracted from the denoised signal waveform, with the abnormal peak frequency as the center. Perform a Hilbert transform on the narrowband time-domain signal to extract the narrowband envelope; On the narrowband envelope, a data segment is selected from the peak value down to 1 / e of the peak value. The data segment is then fitted with an exponential decay function to obtain the corresponding narrowband decay time constant. If there are multiple abnormal peaks in the list of abnormal peaks, repeat the above extraction and fitting operations for each abnormal peak to obtain the decay time constant corresponding to each abnormal peak.

[0037] It should be noted that, using the frequency of each abnormal peak in the abnormal peak list as the center frequency, a common 4th-order Butterworth bandpass filter is used, with the filter bandwidth set to 1 / 10 of the center frequency. This filters the denoised signal waveform, extracting a narrow-band time-domain signal containing only that frequency component. A Hilbert transform is then performed to obtain a new signal with the same frequency and amplitude as the original signal, but with a 90-degree phase lag. Using the original signal as the real part and the Hilbert-transformed signal as the imaginary part, an analytic signal is constructed. The magnitude of this analytic signal is calculated as the narrow-band envelope, reflecting the amplitude change of that specific frequency component over time. On this narrow-band envelope, a data segment from the peak value down to 1 / e of the peak value is selected, and a least-squares fit is performed on this envelope data segment using an exponential decay function. .in, This corresponds to the peak value of the narrowband envelope. The peak amplitude of the narrowband envelope. For time, for The envelope amplitude at time, The solution is the decay time constant. If there are multiple abnormal peaks in the list of abnormal peaks, repeat the above fitting operation for each abnormal peak to obtain the decay time constant corresponding to each abnormal peak.

[0038] It should be noted that the attenuation time constant reflects the rate of energy loss of the signal at the defect location. When defects exist inside the cable, the defect interface causes signal reflection, scattering, and energy absorption, resulting in signal strength attenuation as the propagation distance increases. The more severe the defect, the stronger the attenuation effect on the signal, the faster the signal energy loss, and the smaller the corresponding attenuation time constant.

[0039] In step S14, in one implementation, when the attenuation time constant is less than a preset time threshold, the signal distortion point is determined based on the narrowband time-domain signal, the time difference between the transmission time and the signal distortion point is calculated, and the preliminary location of the defect is obtained based on the time difference and the signal propagation speed, including: The decay time constant corresponding to each abnormal peak is compared with a preset time threshold. If the decay time constant of an abnormal peak is less than the preset time threshold, the position where the signal amplitude drops the most in the narrowband time domain signal corresponding to the abnormal peak is marked as the distortion time point. Subtracting the signal transmission time from the distortion time point yields the two-way signal propagation time, and taking half of the two-way signal propagation time yields the one-way propagation time. The initial location of the defect is obtained by multiplying the one-way propagation time by the signal propagation speed.

[0040] It should be noted that the attenuation time constant corresponding to each abnormal peak is compared with a preset time threshold. When the attenuation time constant is greater than the preset time threshold, it indicates that the signal attenuation is normal, no significant energy loss is detected, and it is determined that there is no defect of concern at the location corresponding to the abnormal peak. In this case, the positioning process is not initiated, and the system continues monitoring. When the attenuation time is less than the preset time threshold, it indicates that the signal attenuation is too fast, with significant energy loss or abnormal reflection. It is determined that there may be a defect of concern at the location corresponding to the abnormal peak, and the positioning process for the abnormal peak is initiated. The preset time threshold is determined experimentally. A defect-free standard cable segment of the same model and specification as the cable under test is selected. Under the same ambient temperature and operating conditions as the actual monitoring, the Rayleigh scattering signal of the standard cable is acquired, and the standard attenuation time is calculated. The average and standard deviation of the standard attenuation time obtained from 10 measurements are calculated repeatedly, and the preset time threshold is set as the average standard attenuation time minus twice the standard deviation.

[0041] Then, for each abnormal peak whose attenuation time constant is less than a preset time threshold, the amplitude change of the narrowband time-domain signal corresponding to the abnormal peak is analyzed to determine the distortion time point. The time-domain signal is scanned point by point in chronological order, and the position where the signal amplitude drops the most is marked as the distortion time point. The distortion time point is the time when the defect reflection signal is received. The signal propagates along the cable and is reflected or scattered after encountering the defect, causing the amplitude of the returned signal to deviate from the normal amplitude. Subtracting the transmission time from the distortion time point gives the two-way propagation time of the signal from transmission to the defect location and back. Dividing the two-way propagation time by 2 gives the one-way propagation time. Multiplying the one-way propagation time by the signal propagation speed in the cable gives the preliminary location of the defect.

[0042] In step S15, in one implementation, error correction is performed based on the preliminary defect location to obtain a high-precision defect location, including: Continuously emit light pulses, and independently perform the above steps to obtain the preliminary location of the defect for each light pulse, and use the preliminary location of the defect obtained for each light pulse as the measurement value; The measured value is optimally estimated using a Kalman filter, and the output of the Kalman filter is used as the high-precision defect location.

[0043] It should be noted that due to factors such as environmental noise, system clock jitter, and changes in fiber refractive index, the preliminary defect location obtained from a single measurement contains random errors. To suppress this error, this embodiment employs a Kalman filter algorithm to perform optimal estimation of multiple measurement results, fusing multiple measurements into an estimate that is closer to the true value. The system continuously emits at least 10 light pulses, each of which independently executes steps S11 to S14, independently calculating the corresponding preliminary defect location measurement value, thus obtaining a sequence of measurement values. The fluctuations in these measurement values ​​reflect the random error of a single location measurement. Next, the measurement noise covariance and process noise covariance are defined.

[0044] The measurement noise covariance represents the magnitude of the random error in a single location measurement. It needs to be determined through calibration experiments before the system is put into use. This is achieved by selecting a standard cable segment of the same model and specifications as the cable being tested, repeating steps S11 to S14 to obtain 10 defect location measurements, and then calculating the variance between the measured defect location values ​​and the actual defect location values ​​as the measurement noise covariance. Since the measurement noise covariance reflects the system's own measurement error characteristics and is independent of the specific defect type and location, the calibrated measurement noise covariance value can be applied to the location measurement of other defects at any location on the same type of cable. The process noise covariance represents the possible actual variation range of the defect location. Since the cable defect location can be considered constant in a short time, the process noise covariance can be taken as 0.01.

[0045] Next, the first measurement value is used as the initial estimate, and the measurement noise covariance is used as the initial estimate error covariance for initialization. Then, the Kalman filter recursively calculates the optimal estimate through two steps: prediction and update. The optimal estimate from the previous update is used as the current prediction value, and the estimation error covariance from the previous update is added to the process noise covariance to obtain the current prediction error covariance. Then, the Kalman gain is calculated using the current prediction error covariance and the measurement noise variance, using the following formula: ,in, For Kalman gain, This is the prediction error covariance. To measure the noise variance, the position is then updated using the Kalman gain, as shown in the formula: And update the estimated error covariance, as shown in the formula. .in, This is the optimal estimate obtained after the update. This is the predicted value for this time. For Kalman gain, These are preliminary measurements of the defect location. This is the prediction error covariance. This represents the updated estimation error covariance. After iteration, the optimal estimate from the initial output is used as the high-precision defect location.

[0046] In one implementation, mapping the narrowband time-domain signal of the high-precision defect location into an image matrix includes: The amplitude of the narrowband time-domain signal at the high-precision defect location is normalized to obtain the normalized amplitude. The normalized amplitude is binned to obtain the binning result for each sampling point of the narrowband time-domain signal; The number of bin transitions between adjacent sampling point pairs is counted to construct a Markov transition probability matrix, and an image matrix is ​​formed based on the Markov transition probability matrix.

[0047] It should be noted that the narrowband time-domain signal corresponding to the high-precision defect location is subjected to amplitude normalization processing, linearly scaling the amplitude to the [0,1] interval. The amplitude range is obtained by subtracting the maximum and minimum amplitude values ​​in the narrowband time-domain signal. Then, the minimum amplitude value is subtracted from the amplitude value of each sampling point in the narrowband time-domain signal, and the difference is divided by the amplitude range to obtain the normalized amplitude. The [0,1] interval is divided into 64 equal sub-intervals, each sub-interval is called a bin, and the bins are numbered from 1 to 64. Each normalized sampling point falls into its corresponding bin based on its normalized amplitude. Next, the pairs of all adjacent sampling points are statistically analyzed. The number of bin transfers is used to construct a 64×64 Markov transition probability matrix, which is as follows: At the same time, when When it is 0, It is also 0. Among them, the sequence number... This is the box number. For the first Sorting to the first The transfer probability of bins, For the separation of boxes Transfer to separate containers Number of times, For the separation of boxes Total number of transfers from the starting point.

[0048] For example, suppose a narrowband time-domain signal, after normalization, has 4 sampling points, and the range [0,1] is divided into 4 equal bins. The amplitude of t1 is 0.8, belonging to bin 4; the amplitude of t2 is 0.9, belonging to bin 4; the amplitude of t3 is 0.3, belonging to bin 2; and the amplitude of t4 is 0.2, belonging to bin 1. Next, the number of bin transfers between adjacent time points is counted. From t1 to t2, it's from bin 4 to bin 4; from t2 to t3, it's from bin 4 to bin 2; and from t3 to t4, it's from bin 2 to bin 1. Therefore, the number of transfers from bin 1 to bin 2 is 0, and the total number of transfers originating from bin 1 is 0. It is 0, similarly, It is 0.5.

[0049] Then, based on the number T of narrowband time-domain signal sampling points, an MTF image matrix M of size T×T is generated, and Furthermore, the high-precision defect location is used as the spatial location label of the image matrix M and stored in conjunction with M.

[0050] In one implementation, visualization rendering is performed based on the image matrix to generate a visual distribution map, including: The probability values ​​of the image matrix are linearly mapped to grayscale ranges to obtain a grayscale image matrix; Edge detection is performed on the grayscale image matrix to extract the geometric contours of the regions; Establish the mapping relationship between grayscale values ​​and colors in the grayscale image matrix; Each pixel in the grayscale image matrix is ​​traversed, filled with color according to the mapping relationship, and the geometric contour of the region is superimposed to generate a visual distribution map.

[0051] It should be noted that each element in the image matrix is ​​a probability value, ranging from [0,1]. The probability value represents information about signal anomalies; a higher probability value indicates frequent signal transitions, potentially indicating a defect; a lower probability value indicates a low frequency of signal transitions, potentially indicating a normal state or random noise. Therefore, multiplying the probability value by 255 and linearly mapping it to the grayscale range [0,255] yields a grayscale image matrix. After mapping, elements with higher probability values ​​have higher grayscale values, appear brighter in the image, and have a higher probability of being defects; elements with lower probability values ​​have lower grayscale values, appear darker, and have a lower probability of being defects.

[0052] Next, Canny edge detection is used to extract the geometric contours of the region from the grayscale image matrix. For each pixel in the grayscale image matrix, the horizontal gradient is obtained by subtracting the grayscale value of the left-hand neighbor from the grayscale value of the right-hand neighbor, and the vertical gradient is obtained by subtracting the grayscale value of the top-hand neighbor from the bottom-hand neighbor. Based on the gradient values ​​in both directions, the gradient magnitude and gradient direction of the pixel are calculated. For each pixel, the gradient magnitude is compared with the gradient magnitude of the two adjacent pixels in its gradient direction. If the gradient magnitude is not a local maximum, it is set to zero; otherwise, it is retained. Pixels with gradient magnitudes higher than a high threshold are marked as strong edges and determined as final edges; pixels with gradient magnitudes lower than a low threshold are marked as non-edges and excluded; pixels with gradient magnitudes between the low and high thresholds are marked as weak edges. For weak edges, if they are adjacent to any strong edge, they are retained as edges; otherwise, they are excluded. Adjacent edge points are connected to form a continuous closed curve, i.e., the geometric contour of the region.

[0053] Then, the mapping relationship between grayscale image matrix grayscale values ​​and colors is established. In this example, grayscale values ​​from 0-120 are mapped to green, grayscale values ​​in the 120-200 range are linearly interpolated between green and red, and grayscale values ​​from 200-255 are mapped to red. For grayscale values ​​in the 120-200 range, the red component linearly increases from 0 to 255, the green component linearly decreases from 255 to 0, and the blue component remains 0. The calculation formula is: , , .in, The grayscale image matrix contains grayscale values, and 80 represents the length of the interpolation interval 120-200. Each pixel in the image matrix is ​​iterated through, and its corresponding color is found based on its grayscale value and filled into that pixel. Then, a defect boundary outline is overlaid on the filled image to generate a visual distribution map. This visual distribution map clearly shows the location, extent, and depth of the defects.

[0054] In summary, this invention discloses a method for monitoring faults in wires and cables. By acquiring Rayleigh scattering signals, digitally filtering and denoising the signals, and identifying abnormal features of the signals to calculate the preliminary location of defects, the method corrects the errors in the preliminary defect location and finally generates a visualized distribution map of internal defects in the cable. This method realizes a complete technical chain from weak signal extraction and precise defect location to defect structure visualization, significantly improving the reliability of fault monitoring results and providing reliable technical support for intelligent fault prediction and health management of wires and cables.

[0055] Reference Figure 2 The second embodiment of the present invention provides a wire and cable fault monitoring system, comprising: A digital filtering module is used to transmit light pulses into the optical fiber in the cable, record the transmission time and acquire the Rayleigh scattering signal, and perform digital filtering on the Rayleigh scattering signal to obtain an initial signal sequence. The signal denoising module is used to denoise the initial signal sequence to obtain the denoised signal waveform. The attenuation parameter extraction module is used to perform frequency domain transformation on the denoised signal waveform to obtain a frequency domain signal, perform peak detection on the frequency domain signal to obtain abnormal peaks, extract the narrowband time domain signal corresponding to the abnormal peaks, and perform attenuation fitting on the envelope of the narrowband time domain signal to obtain the attenuation time constant. The defect location module is used to determine the signal distortion point based on the narrowband time domain signal when the attenuation time constant is less than a preset time threshold, calculate the time difference between the transmission time and the signal distortion point, and obtain the preliminary location of the defect based on the time difference and the signal propagation speed. The defect image output module is used to perform error correction based on the preliminary defect position to obtain a high-precision defect position; map the narrowband time-domain signal of the high-precision defect position into an image matrix; and perform visualization rendering based on the image matrix to generate a visualization distribution map.

[0056] It should be noted that the wire and cable fault monitoring system provided in this embodiment of the invention is used to execute all the process steps of the wire and cable fault monitoring method in the above embodiment. The working principle and beneficial effects of the two are one-to-one, so they will not be described again.

[0057] It should be noted that the system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the system embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0058] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A method for monitoring faults in electric wires and cables, characterized in that, include: An optical pulse is emitted into the optical fiber in the cable, the emission time is recorded, and the Rayleigh scattering signal is obtained. The Rayleigh scattering signal is then digitally filtered to obtain an initial signal sequence. The initial signal sequence is denoised to obtain the denoised signal waveform; The denoised signal waveform is transformed into a frequency domain signal. The power density of each frequency point in the frequency domain signal is checked against the power densities of the two adjacent frequency points. If so, the frequency point is selected as a candidate peak. The average power density of all frequency points in the frequency domain signal is calculated. The power density of each candidate peak is checked against the average power density. If so, the candidate peak is selected as a high-power peak. The denoised signal waveform is subjected to a Hilbert transform to extract the signal envelope. The envelope amplitude of each frequency point is determined based on the signal envelope. The reflection intensity ratio of each frequency point in the frequency domain signal is calculated. The average reflection intensity ratio of all frequency points in the frequency domain signal is calculated. The reflection intensity ratio of each high-power peak is checked against the average reflection intensity ratio. If so, the high-power peak is selected as an abnormal peak, and an abnormal peak list is generated. For each abnormal peak in the abnormal peak list, a narrowband time-domain signal is extracted from the denoised signal waveform with the abnormal peak frequency as the center. A Hilbert transform is performed on the narrowband time-domain signal to extract the narrowband envelope. On the narrowband envelope, a data segment from the peak to 1 / e of the peak value is selected. An exponential decay function is used to fit this data segment to obtain the corresponding narrowband decay time constant. If there are multiple abnormal peaks in the abnormal peak list, the above extraction and fitting operations are repeated for each abnormal peak to obtain the decay time constant corresponding to each abnormal peak. The decay time constant corresponding to each abnormal peak is compared with a preset time threshold. When the decay time constant of the abnormal peak is less than the preset time threshold, the position where the signal amplitude drops the most in the narrowband time domain signal corresponding to the abnormal peak is marked as the distortion time point. The time difference between the transmission time and the distortion time point is calculated. Based on the time difference and the signal propagation speed, the preliminary location of the defect is obtained. The high-precision defect location is obtained by error correction based on the initial defect location, and the narrowband time-domain signal of the high-precision defect location is mapped into an image matrix; the image matrix is ​​then visualized and rendered to generate a visual distribution map.

2. The method for monitoring wire and cable faults according to claim 1, characterized in that, The process of digitally filtering the Rayleigh scattering signal to obtain an initial signal sequence includes: The Rayleigh scattering signal is photoelectrically converted to obtain an analog voltage signal; The analog voltage signal is digitally processed to obtain time series data; The time series data is filtered out of noise using a digital filter to obtain the initial signal sequence.

3. The method for monitoring wire and cable faults according to claim 1, characterized in that, The step of denoising the initial signal sequence to obtain the denoised signal waveform includes: The initial signal sequence is subjected to interference signal separation to obtain the main signal components; The main signal components are decomposed into time-frequency features to obtain multi-scale features; The signal is reconstructed based on the multi-scale features to obtain the denoised signal waveform.

4. The method for monitoring wire and cable faults according to claim 1, characterized in that, The step of obtaining the preliminary location of the defect based on the time difference and signal propagation speed includes: Subtracting the transmission time from the distortion time point yields the signal propagation two-way time, and taking half of the signal propagation two-way time yields the one-way propagation time. The initial location of the defect is obtained by multiplying the one-way propagation time by the signal propagation speed.

5. The method for monitoring wire and cable faults according to claim 4, characterized in that, The step of obtaining a high-precision defect location by error correction based on the preliminary defect location includes: Continuously emit light pulses, and independently perform the above steps to obtain the preliminary location of the defect for each light pulse, and use the preliminary location of the defect obtained for each light pulse as the measurement value; The measured value is optimally estimated using a Kalman filter, and the output of the Kalman filter is used as the high-precision defect location.

6. The method for monitoring wire and cable faults according to claim 1, characterized in that, The process of mapping the narrowband time-domain signal based on the high-precision defect location into an image matrix includes: The amplitude of the narrowband time-domain signal at the high-precision defect location is normalized to obtain the normalized amplitude. The normalized amplitude is binned to obtain the binning result for each sampling point of the narrowband time-domain signal; The number of bin transitions between adjacent sampling point pairs is counted to construct a Markov transition probability matrix, and an image matrix is ​​formed based on the Markov transition probability matrix.

7. The method for monitoring wire and cable faults according to any one of claims 1-6, characterized in that, The step of visualizing and rendering the image matrix to generate a visual distribution map includes: The probability values ​​of the image matrix are linearly mapped to a grayscale range to obtain a grayscale image matrix; Edge detection is performed on the grayscale image matrix to extract the geometric contours of the regions; Establish the mapping relationship between grayscale values ​​and colors in the grayscale image matrix; Each pixel in the grayscale image matrix is ​​traversed, filled with color according to the mapping relationship, and the geometric contour of the region is superimposed to generate a visual distribution map.

8. A wire and cable fault monitoring system, characterized in that, Used to implement the wire and cable fault monitoring method according to any one of claims 1-7.

Images