Conveyor carrier roller fault positioning and decoupling method based on DAS space coherence

Abstract

The invention discloses a conveyor carrier roller fault positioning and decoupling method based on DAS space coherence, and relates to the technical field of optical fiber sensing and industrial monitoring. The method comprises the following steps: acquiring a sound wave signal along a conveyor and analyzing the sound wave signal into a space-time two-dimensional matrix; performing energy integral calculation on the matrix to lock a high-energy primary screening area; selecting a point with maximum energy in the region as a candidate center and observation points on the left and right sides of the candidate center; calculating a cross-correlation function of the candidate center and the point locations on the two sides and extracting relative time delay; and if the candidate center signals are ahead of the observation point signals on the two sides in time, determining that the candidate center signals are real fault sources. By implementing the technical scheme provided by the invention, the fault positioning accuracy of the carrier roller of the conveyor is improved.

Description

Technical Field

[0001] This application relates to the field of fiber optic sensing and industrial monitoring technology, and in particular to a method for fault location and decoupling of conveyor idler rollers based on DAS spatial coherence. Background Technology

[0002] Currently, belt conveyors are key equipment for bulk material transport in industries such as coal mines, ports, and power plants. With the continuous improvement of industrial automation, the trend towards long-distance and high-capacity conveying systems makes safe operation extremely important. Idler rollers, as the most numerous and widely distributed rotating components in a conveyor, directly affect the stability and efficiency of the entire machine. Once idler rollers experience bearing damage, jamming, or other malfunctions, it will not only significantly increase operating energy consumption, but in severe cases, it may even lead to major safety accidents such as longitudinal belt tearing or fire.

[0003] In related technologies, distributed optical fiber acoustic sensing (DAS) technology is typically used to monitor the entire conveyor line. The process involves logically dividing the optical fiber laid along the frame into several discrete, independent monitoring channels (e.g., one measuring point every 1 meter), and processing the vibration signal collected at each measuring point independently. The time-domain energy amplitude, kurtosis, or spectral characteristics of a specific frequency band of the signal at each measuring point are calculated in real time. When the signal index value at a certain measuring point exceeds a preset alarm threshold, it is determined that there is a roller fault at the physical location corresponding to that measuring point, and an alarm message is output.

[0004] However, in related technologies, because the frame of a belt conveyor is usually a rigidly connected metal structure with vibration transmission characteristics, when a mechanical failure occurs at a certain idler roller, causing impact vibration, the vibration wave will propagate rapidly along the frame to both sides of the fault point and attenuate slowly. In this situation, normal measuring points within a range of several meters or even more than ten meters around the fault point will detect high-amplitude vibration signals exceeding the threshold, making it difficult to distinguish between the source of vibration energy and the passive transmission area. This will trigger a wide range of consecutive alarms, making it impossible for maintenance personnel to accurately identify the true fault location, ultimately resulting in a high false alarm rate and insufficient location accuracy. Summary of the Invention

[0005] This application provides a method for fault location and decoupling of conveyor idler rollers based on DAS spatial coherence, which can improve the accuracy of conveyor idler roller fault location.

[0006] The first aspect of this application provides a method for fault location and decoupling of conveyor idler rollers based on DAS spatial coherence, the method comprising: Acquire acoustic signals along the conveyor line and resolve them into a two-dimensional data matrix. Calculate the energy integral by performing a sum of squares and integral on the two-dimensional data matrix. Mark the set of continuously distributed spatial points whose energy integral values ​​exceed a preset initial screening threshold as the initial screening area. Within the initial screening area, select the point with the largest energy integral amplitude as the candidate center point. Select a left observation point at a preset distance to the left of the candidate center point and a right observation point at a preset distance to the right of the candidate center point. Calculate the first cross-correlation function between the candidate center point and the left observation point, and the second cross-correlation function between the candidate center point and the right observation point. Based on the peak positions of the first and second cross-correlation functions, obtain the first and second time delays of the candidate center point relative to the left and right observation points, respectively. If the signal of the candidate center point leads the signals of both the left and right observation points in time, the candidate center point is determined to be the actual fault source.

[0007] In the above embodiments, the spatial coherence sensed by distributed optical fibers is used to verify the wave propagation characteristics of suspected high-energy regions. The time delay is extracted by calculating the cross-correlation function to confirm whether the vibration signal conforms to the physical propagation law of diverging from the candidate center to both sides. The judgment logic based on spatiotemporal causality can distinguish between real mechanical fault sources and random background noise or unidirectional conducted interference, thereby avoiding false alarms caused by relying solely on energy amplitude and ultimately improving the accuracy of conveyor roller fault location.

[0008] In conjunction with some embodiments of the first aspect, in some embodiments, when the signal of the candidate center point is ahead of the signals of both the left and right observation points in time, the candidate center point is determined to be the actual fault source, specifically including: Obtain the first physical distance between the candidate center point and the left observation point, and the second physical distance between the candidate center point and the right observation point; divide the first physical distance by the first time delay to obtain the left propagation velocity, and divide the second physical distance by the second time delay to obtain the right propagation velocity; determine whether the left propagation velocity and the right propagation velocity are both within the preset effective range of rack sound velocity; if so, if the signal of the candidate center point is ahead of the signals of the left observation point and the right observation point in time, determine that the candidate center point is the real fault source.

[0009] In the above embodiments, a wave velocity consistency verification mechanism based on the physical characteristics of the rack is introduced. By comparing the calculated propagation speed with the theoretical sound speed range of the metal rack, it is possible to distinguish between real fault vibrations transmitted through rigid structures and non-structural transmission interference such as fiber optic flapping and air propagation from the perspective of physical propagation mechanisms. This eliminates false signals that do not conform to physical laws and improves the physical reliability and anti-interference capability of fault source identification.

[0010] In conjunction with some embodiments of the first aspect, in some embodiments, before determining that the candidate center point is the actual fault source, the method further includes: The acoustic signals from the candidate center point, the left observation point, and the right observation point are divided into multiple consecutive verification sub-windows on the time axis according to a preset sub-window size. For each verification sub-window, the cross-correlation function between the candidate center point and the left observation point, and between the candidate center point and the right observation point are calculated, and the corresponding sub-window time delay is extracted. The number of verification sub-windows that satisfy the condition that the signal of the candidate center point leads the signals of the left observation point and the right observation point in time is counted, and the persistence ratio of the candidate center point to the total number of verification sub-windows is calculated. If the persistence ratio exceeds a preset fault confidence threshold, the candidate center point is determined to be the real fault source.

[0011] In the above embodiments, a signal persistence verification mechanism based on time-domain slicing is introduced. Utilizing the difference between the temporal continuity of real mechanical faults and the transient nature of interference such as material impacts, occasional random noise interference is filtered out by statistically analyzing the proportion of sub-windows that satisfy wave propagation laws. This time-dimensional statistical filtering ensures that alarms are only triggered for persistent anomalies, improving the system's robustness in identifying real faults and its resistance to false alarms under dynamic operating conditions.

[0012] In conjunction with some embodiments of the first aspect, in some embodiments, a left observation point located at a predetermined distance to the left of the candidate center point and a right observation point located at a predetermined distance to the right of the candidate center point are selected respectively. A first cross-correlation function between the candidate center point and the left observation point, and a second cross-correlation function between the candidate center point and the right observation point are calculated, specifically including: The original acoustic signal sequences of the candidate center point, the left observation point, and the right observation point are extracted respectively. The center original acoustic signal sequence of the candidate center point is differentially processed with the adjacent signal sequences of the adjacent spatial points to obtain the center differential signal. The left observation point and the right observation point are differentially processed to obtain the left differential signal and the right differential signal respectively. The cross-correlation function between the center differential signal and the left differential signal is calculated as the first cross-correlation function, and the cross-correlation function between the center differential signal and the right differential signal is calculated as the second cross-correlation function.

[0013] In the above embodiments, the point sound source signal generated by the roller fault has differences between adjacent spatial channels, while the background noise such as belt running exhibits the physical characteristics of a highly correlated common-mode signal. The differential operation cancels out the common-mode interference distributed throughout the line and enhances the local fault characteristics, thereby improving the signal-to-noise ratio of the signal and ensuring that the cross-correlation analysis is not dominated by strong background noise. Thus, the capture and location of weak fault signals are achieved under harsh working conditions.

[0014] In conjunction with some embodiments of the first aspect, in some embodiments, when the signal of the candidate center point is ahead of the signals of both the left and right observation points in time, the candidate center point is determined to be the actual fault source, specifically including: The acoustic signal sequences of the candidate center point, the left observation point, and the right observation point are transformed from the time domain to the frequency domain to obtain the corresponding power spectral density functions. The spectral centroids of the candidate center point, the left observation point, and the right observation point are calculated using the power spectral density functions. The magnitude relationship between the spectral centroid of the candidate center point and the spectral centroids of the left and right observation points is compared. If the signal of the candidate center point leads the signals of the left and right observation points in time, and the frequency of the spectral centroid of the candidate center point is higher than the frequency of the spectral centroids of the left and right observation points, the candidate center point is determined to be the real fault source.

[0015] In the above embodiments, by utilizing the physical characteristic that the attenuation rate of high-frequency vibration signals is faster than that of low-frequency signals during rack transmission, the evolution law of the spectral centroid decreasing from the center to both sides is verified, thus constructing a frequency domain physical constraint independent of the time domain. The dual interlocking mechanism of time-domain divergence and frequency-domain attenuation eliminates false signals such as low-frequency resonance waves transmitted from distant locations, improving the rigor and accuracy of the system in determining the real fault source in complex noise environments.

[0016] In conjunction with some embodiments of the first aspect, in some embodiments, before determining that the candidate center point is the actual fault source, the method further includes: The first amplitude squared coherence function is calculated using the acoustic signal sequences from the candidate center point and the observation point on the left, and the second amplitude squared coherence function is calculated using the acoustic signal sequences from the candidate center point and the observation point on the right. The first coherence coefficient value of the first amplitude squared coherence function at the centroid frequency of the spectrum of the candidate center point and the second coherence coefficient value of the second amplitude squared coherence function at the centroid frequency of the spectrum of the candidate center point are extracted. If both the first coherence coefficient value and the second coherence coefficient value are greater than the preset waveform correlation threshold, the candidate center point is determined to be the real fault source.

[0017] In the above embodiments, by leveraging the physical difference that real mechanical fault signals maintain a high degree of linear correlation during transmission along the frame, while sensor loosening or local random noise is uncorrelated, coherence verification at the main energy frequency band (spectral centroid) distinguishes physically transmitted signals from local independent interference. Deep verification based on the linearity of the signal system eliminates false high-frequency alarms lacking physical propagation evidence, improving the physical reliability and anti-interference accuracy of fault determination.

[0018] In conjunction with some embodiments of the first aspect, in some embodiments, within the initial screening area, the point with the largest amplitude of the energy integral is selected as the candidate center point, specifically including: For each spatial point within the initial screening area, the temporal kurtosis value of the acoustic signal sequence is calculated. The energy integral value and temporal kurtosis value of each spatial point within the initial screening area are normalized to obtain a normalized energy sequence and a normalized kurtosis sequence. The normalized energy sequence and the normalized kurtosis sequence are then weighted and fused point-to-point to obtain the fault source confidence index of each spatial point within the initial screening area. The spatial point with the largest fault source confidence index is selected as the candidate center point.

[0019] In the above embodiments, by weighted fusion of the time-domain kurtosis characterizing the impact characteristics and the energy integral, the problem of point selection deviation caused by the single energy index being easily affected by low-frequency resonance of the frame or stable environmental noise is solved, ensuring that the candidate center point can be locked on the real physical impact source with significant impact characteristics, and improving the anti-interference capability and positioning accuracy of the initial screening of fault sources under complex working conditions.

[0020] Secondly, embodiments of this application provide a conveyor idler fault location and decoupling system, which includes: one or more processors and a memory; the memory is coupled to the one or more processors, and the memory is used to store computer program code, which includes computer instructions, and the one or more processors call the computer instructions to cause the conveyor idler fault location and decoupling system to perform the method described in the first aspect and any possible implementation thereof.

[0021] Thirdly, embodiments of this application provide a computer program product containing instructions that, when the computer program product is run on a conveyor idler roller fault location and decoupling system, cause the conveyor idler roller fault location and decoupling system to perform the method described in the first aspect and any possible implementation thereof.

[0022] Fourthly, embodiments of this application provide a computer-readable storage medium including instructions that, when executed on a conveyor idler roller fault location and decoupling system, cause the conveyor idler roller fault location and decoupling system to perform the method described in the first aspect and any possible implementation thereof.

[0023] Understandably, the conveyor idler fault location and decoupling system provided in the second aspect, the computer program product provided in the third aspect, and the computer storage medium provided in the fourth aspect are all used to execute the conveyor idler fault location and decoupling method based on DAS spatial coherence provided in the embodiments of this application. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects in the corresponding methods, and will not be repeated here.

[0024] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages: 1. This application verifies the wave propagation characteristics of suspected high-energy regions by utilizing the spatial coherence of distributed optical fiber sensing. By calculating the cross-correlation function to extract the time delay, it confirms whether the vibration signal conforms to the physical propagation law of diverging from the candidate center to both sides. Based on the spatiotemporal causality-based judgment logic, it can distinguish between genuine mechanical fault sources and random background noise or unidirectional conducted interference, thereby avoiding false alarms caused by relying solely on energy amplitude and ultimately improving the accuracy of conveyor roller fault location.

[0025] 2. This application introduces a wave velocity consistency verification mechanism based on the physical characteristics of the rack. By comparing the calculated propagation velocity with the theoretical sound velocity range of the metal rack, it can distinguish between real fault vibrations transmitted through rigid structures and non-structural transmission interference such as fiber optic flapping and air propagation from the perspective of physical propagation mechanism. This eliminates false signals that do not conform to physical laws and improves the physical reliability and anti-interference capability of fault source identification.

[0026] 3. This application introduces a signal persistence verification mechanism based on time-domain slicing. Utilizing the difference between the temporal continuity of real mechanical faults and the transient nature of interference such as material impacts, it filters out occasional random noise interference by statistically analyzing the proportion of sub-windows that satisfy wave propagation laws. This time-dimensional statistical filtering ensures that alarms are only triggered for persistent anomalies, improving the system's robustness and false alarm resistance in identifying real faults under dynamic operating conditions. Attached Figure Description

[0027] Figure 1 This is a flowchart illustrating a method for fault location and decoupling of conveyor idler rollers based on DAS spatial coherence in an embodiment of this application. Figure 2 This is another flowchart illustrating the conveyor idler roller fault location and decoupling method based on DAS spatial coherence in this application embodiment; Figure 3 This is an exemplary hardware structure diagram of the conveyor idler roller fault location and decoupling system in the embodiments of this application. Detailed Implementation

[0028] The terminology used in the following embodiments of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. As used in the specification and appended claims of this application, the singular expressions “a,” “an,” “the,” “the,” “the,” and “this” are intended to include the plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in this application refers to and includes any or all possible combinations of one or more of the listed items.

[0029] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature, and in the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more.

[0030] In related technologies, distributed fiber optic acoustic sensing technology is typically used to monitor the entire conveyor line. The processing logic involves dividing the optical fiber laid along the frame into several independent discrete monitoring channels, and performing isolated energy or spectrum analysis on the vibration signals collected at each measuring point. When the signal index at a measuring point exceeds a preset threshold, a fault is determined to exist at that point. However, because the conveyor frame is usually a rigidly connected metal structure with extremely strong vibration transmission characteristics, when a mechanical failure occurs at a roller, causing impact vibration, the vibration wave will propagate rapidly along the frame to both sides of the fault point and attenuate slowly. This physical characteristic means that normal measuring points within a range of several meters or even more than ten meters from the fault point can detect high-amplitude signals exceeding the threshold, making it difficult to distinguish the actual source of vibration energy from the passively transmitted area. This easily triggers widespread, continuous alarms, making it impossible for maintenance personnel to accurately identify the true fault location, ultimately resulting in a high false alarm rate and insufficient location accuracy.

[0031] In this embodiment, the acoustic signal is analyzed into a two-dimensional spatiotemporal matrix. First, a high-energy preliminary screening region and candidate center are identified based on energy integration. Then, the signal time delay between the candidate center and the observation points on both sides is calculated using a cross-correlation function. The physical propagation law is verified by checking whether the signal diverges from the center to both sides in time (i.e., the center signal leads the sides). This mechanism utilizes the physical characteristic that a real vibration source inevitably radiates energy outwards, enabling the differentiation between real mechanical fault sources and random background noise or unidirectional conducted interference. This avoids false alarms caused by relying solely on energy amplitude and improves the accuracy of conveyor roller fault location.

[0032] Figure 1 This is a flowchart illustrating the conveyor idler roller fault location and decoupling method based on DAS spatial coherence in the embodiments of this application, including the following steps: S101. Acquire the acoustic wave signal along the conveyor line and parse the acoustic wave signal into a two-dimensional data matrix.

[0033] Among them, the acoustic signal along the conveyor refers to the signal of phase or intensity change of Rayleigh scattered light caused by mechanical vibration, which is sensed by distributed optical fiber sensors (DAS) laid along the belt conveyor frame; the two-dimensional data matrix is ​​a digital storage structure used to represent a spatiotemporally continuous vibration field, usually represented as D(x,t), with the horizontal axis being the distance along the optical fiber and the vertical axis being the sampling time point; the distance along the optical fiber is used to represent the physical length of the signal source from the DAS host end in space, corresponding to the column index of the matrix; the sampling time point is used to represent the time of data acquisition, corresponding to the row index of the matrix.

[0034] Specifically, the distributed acoustic wave sensor host emits highly coherent laser pulses into the sensing optical fiber laid along the conveyor. When the optical pulses are transmitted in the optical fiber, the external acoustic wave vibrations will change the refractive index of the optical fiber or stretch the optical fiber, causing the phase of the backscattered Rayleigh light to be modulated.

[0035] The host receives the backscattered light carrying vibration information, converts it into electrical signals using a photodetector, and reconstructs the vibration waveforms at each point along the optical fiber using orthogonal demodulation or phase demodulation algorithms. To facilitate computer processing, the data collected over a period of time is arranged according to spatial channels and time frames to construct an M×N matrix, where M represents the number of spatial sampling points (covering the entire length of the conveyor) and N represents the number of time sampling points (corresponding to the sampling duration), thus digitizing the vibration field into a computable mathematical object.

[0036] In some embodiments, the analysis and matrix construction of acoustic signals can be achieved in various ways: Optionally, phase-sensitive optical time-domain reflectometry is used: First, a narrow-linewidth laser pulse is emitted into the optical fiber and the backscattered Rayleigh light is received; second, the optical signal is converted into an intermediate-frequency electrical signal using a heterodyne detector structure, and IQ demodulation is performed to extract phase information; finally, the demodulated phase data is filled into a memory buffer according to distance and time order to form a phase-space-time matrix.

[0037] Optionally, a dual-pulse heterodyne demodulation technique is employed: First, a pair of optical pulses with a fixed frequency difference and time difference are modulated and injected into the optical fiber; second, the interference signal generated by the two pulses is received, and the beat frequency component is separated by bandpass filtering; finally, the beat frequency signal is demodulated by arctangent operation to obtain the phase change, and mapped into a two-dimensional matrix format.

[0038] It is understandable that direct detection techniques based on intensity demodulation can also be used to analyze acoustic signals, and this is not a limitation here.

[0039] S102. Perform a sum of squares and integrals on the two-dimensional data matrix to obtain the energy integral. Mark the set of spatial points that are continuously distributed and whose energy integral values ​​exceed the preset initial screening threshold as the initial screening area.

[0040] Among them, the sum of squares integral calculation refers to a signal processing algorithm used to quantify the total energy of a signal within a specific time window, mathematically expressed as ∑x(t). 2 Energy integral is used to represent the intensity of vibration at a certain spatial point over a period of time; preset initial screening threshold refers to the energy boundary used to distinguish between normal background noise and abnormal vibration events. This threshold is usually set based on historical background noise statistics of the conveyor under no-load or normal operating conditions, combined with the experimental experience of those skilled in the art, for example, it is set to 3 times the average background noise; the initial screening area refers to the spatial range that the system initially determines to have suspected faults or strong vibrations.

[0041] Iterate through each spatial channel (i.e., each column of the matrix) of the two-dimensional matrix and extract all sampling point data of that channel within the current time window. Square each of these data points to remove negative values ​​and amplify the influence of large-amplitude signals. Then, sum the squared values ​​to obtain the short-time energy integral value of that spatial point.

[0042] After the calculation is completed, the energy values ​​of all spatial points are compared with the preset initial screening threshold. In order to avoid false alarms caused by single-point noise, the spatial continuity is further checked. Only when the energy integral values ​​of multiple adjacent spatial points (for example, points corresponding to consecutive 5 meters) exceed the threshold at the same time is this series of spatial points marked as the initial screening area, thereby locking the physical segment where the vibration energy is concentrated.

[0043] In some embodiments, energy integration and region marking can be implemented in a variety of ways: Optionally, an energy screening method based on frequency domain characteristics is used: First, a fast Fourier transform (FFT) is performed on the signal at each spatial point to obtain spectral data; second, the sum of spectral line energies within a specific fault frequency band (such as the fault characteristic frequency band of the idler roller) is calculated; finally, it is determined whether the energy of the frequency band exceeds the dynamic threshold, and points that meet the conditions and are spatially adjacent are merged into the initial screening area.

[0044] It is understandable that energy assessment can also be achieved using calculation methods based on root mean square (RMS) or peak factor, and this is not limited here.

[0045] S103. Within the initial screening area, select the point with the largest amplitude of energy integral as the candidate center point.

[0046] The amplitude of the energy integral refers to the numerical value that characterizes the vibration intensity, calculated in step S102; the candidate center point refers to the spatial sampling point within the initial screening area that is assumed by the algorithm to be the source of vibration or the location of the fault, and is used as a reference point for subsequent waveform analysis.

[0047] Specifically, since vibration waves attenuate as they propagate along the frame, theoretically, the closer a location is to the vibration source, the stronger the received vibration energy. Based on this physical law, an extreme value search is performed within each marked initial screening area.

[0048] The energy integral values ​​corresponding to all spatial points within the region are read, and the maximum value is found through a comparison algorithm. The spatial coordinates (fiber optic distance index) corresponding to this maximum value are then locked. This point is considered to be the core of the most intense vibration in the region and may be the physical point of origin of the idler roller failure. Therefore, it is defined as a candidate center point for subsequent wave propagation characteristic verification around this point.

[0049] In some embodiments, candidate center points can be selected in a variety of ways: Optionally, the direct extreme value search method is as follows: First, traverse the energy integral values ​​corresponding to all spatial indices within the initial screening region; second, sort the energy values ​​using bubble sort or quick sort algorithm, or directly use the max function to find the maximum value; finally, return the spatial index corresponding to the maximum value as the candidate center point.

[0050] It is understandable that the center selection can also be achieved by calculating the energy centroid using the centroid method, and this is not limited here.

[0051] S104. Select a left observation point located at a preset distance to the left of the candidate center point and a right observation point located at a preset distance to the right of the candidate center point, respectively. Calculate the first cross-correlation function between the candidate center point and the left observation point, and the second cross-correlation function between the candidate center point and the right observation point.

[0052] Among them, the left / right observation points refer to the auxiliary sampling points located upstream and downstream of the candidate center point in space, used to capture the signal through which the vibration wave propagates; the preset distance refers to the spatial interval between the observation point and the center point. This distance is preset by those skilled in the art based on the acoustic attenuation characteristics and sampling rate of the conveyor frame to ensure that a significant time difference can be observed while ensuring that the signal has sufficient coherence; the cross-correlation function is a mathematical function used to measure the similarity of two time series at different time displacements, and is expressed as R(v).

[0053] Using the candidate center point determined by S103 as the origin, select a point by offsetting a preset distance in the direction of decreasing fiber optic distance (left) and select another point by offsetting a preset distance in the direction of increasing fiber optic distance (right).

[0054] Next, the vibration waveform data sequence of these three points within the same time period is extracted from the two-dimensional data matrix.

[0055] Subsequently, using signal processing algorithms, the waveform sequences of the candidate center point and the observation points on the left are subjected to a sliding inner product operation to obtain the first cross-correlation function; similarly, the waveform sequences of the candidate center point and the observation points on the right are operated on to obtain the second cross-correlation function. These two function curves reflect the degree of matching of the signal waveforms under different time delays.

[0056] In some embodiments, the cross-correlation function can be calculated in multiple ways: Optionally, the direct cross-correlation method in the time domain is used: First, extract the signal sequence x(t) at the center point and the signal sequence y(t) at the left point; second, set the sliding time delay range v, and calculate ∑x(t)×y(t+v) at each delay point; finally, arrange the calculation results according to v to generate the cross-correlation function curve.

[0057] Optionally, the frequency domain fast cross-correlation method is used: First, fast Fourier transform (FFT) is performed on the center point signal and the observation point signal respectively; second, the conjugate of the spectrum of the center point signal and the spectrum of the observation point signal is multiplied to obtain the cross power spectrum; finally, inverse fast Fourier transform (IFFT) is performed on the cross power spectrum to obtain the cross-correlation function in the time domain.

[0058] Understandably, normalized cross-correlation (NCC) calculation can also be used to eliminate the influence of signal amplitude differences, and this is not a limitation here.

[0059] In some embodiments, when the conveyor is operating under complex conditions and there is strong common-mode background noise such as belt misalignment, friction, or overall vibration, the cross-correlation function can be calculated by spatial differential preprocessing to eliminate common-mode interference and highlight weak fault characteristics, thereby improving the signal-to-noise ratio and accuracy of positioning.

[0060] First, the original acoustic signal sequences of the candidate center point, the left observation point, and the right observation point within the same time window are extracted respectively. Then, a spatial differential filtering mechanism is introduced, utilizing the physical characteristics of roller faults as point sound sources with significant differences (differential mode signal) in adjacent spatial channels, while belt friction or overall vibration as surface sound sources with high similarity (common mode signal) in adjacent channels.

[0061] The signal sequences of the candidate center point and its adjacent spatial points (e.g., points with fiber distance indices of +1 or -1) are selected, and the two are subtracted point by point to obtain the center differential signal. Mathematically, this differential operation is equivalent to a high-pass spatial filter, which can cancel out low-frequency common-mode background noise that changes synchronously in adjacent channels, while preserving and enhancing the differentiated high-frequency vibration components caused by local faults. Similarly, the same differential operation is performed on the left and right observation points to obtain the left and right differential signals, respectively.

[0062] Finally, instead of using the original signal, a sliding inner product operation is performed between the processed center differential signal and the left differential signal to calculate the first cross-correlation function; then, the center differential signal and the right differential signal are used to calculate the second cross-correlation function. This avoids strong background noise dominating the cross-correlation peak and ensures that the calculated time delay truly reflects the propagation characteristics of the fault signal. The selection of adjacent spatial points is preset, typically based on the spatial resolution of the fiber optic sensor and the vibration guidance characteristics of the rack structure, and is determined by those skilled in the art through experimental testing; for example, adjacent channels within a 1-meter interval are selected.

[0063] By employing the above technical steps, the common-mode interference across the entire conveyor line can be suppressed during unsteady operation (such as misalignment or heavy load). Through spatial difference calculations, weak idler fault signals submerged in strong background noise are extracted, resulting in a sharper peak value and more accurate positioning of the cross-correlation function. This solves the problem of positioning failure or false alarms caused by low signal-to-noise ratio under harsh operating conditions, improving the system's environmental adaptability.

[0064] S105. Based on the peak positions of the first cross-correlation function and the second cross-correlation function, obtain the first time delay and the second time delay of the candidate center point relative to the left observation point and the right observation point, respectively.

[0065] The peak position refers to the time coordinate (horizontal axis) corresponding to the point with the largest amplitude on the cross-correlation function curve; the time delay refers to the time difference between two signals, used to represent the time required for a sound wave to propagate from one point to another, or the time interval between two points receiving the same vibration event.

[0066] Specifically, the complex waveform similarity curve is transformed into a concrete physical quantity. The physical meaning of the cross-correlation function is that the function value reaches its peak when the overlap between two signals is highest at a certain time displacement. Therefore, the first cross-correlation function calculated in step S104 is scanned to find the horizontal coordinate position of the maximum value. This coordinate value is the first time delay, representing the arrival time difference between the signal at the candidate center point and the observation point on the left. Similarly, a peak search is performed on the second cross-correlation function to extract the second time delay. These two time delay values ​​contain information about the direction and velocity of wave propagation.

[0067] S106. If the signal at the candidate center point is ahead of the signals at the left and right observation points in time, the candidate center point is determined to be the real fault source.

[0068] Among them, "signal leading in time" means that the candidate center point receives the vibration wave earlier than the observation point receives the vibration wave; "real fault source" refers to the location where mechanical impact or wear actually occurs, rather than the forced vibration area excited by the conducted wave.

[0069] Specifically, if the candidate center point is the actual source of vibration (such as a faulty idler roller), then the vibration wave will originate from that point and spread to both sides. Therefore, the center point will inevitably feel the vibration first, and then the wave will be transmitted to the observation points on the left and right sides. In terms of data, this means that the time of the candidate center point is earlier than that of the left and right sides.

[0070] Check the sign or numerical relationship between the first and second time delays extracted in step S105 (depending on the definition of the reference frame for cross-correlation calculation) to confirm whether the bidirectional divergence condition of the center being earlier than the left and the center being earlier than the right is met. Only when this condition is met simultaneously is the candidate point confirmed as the source of vibration, thus determining it as the real fault source and eliminating interference signals from unidirectional conduction.

[0071] In some embodiments, the fault source can be determined in multiple ways: Optionally, a logical judgment method based on the delay sign is used: First, the cross-correlation calculation direction is defined as the center relative to the observation point; second, it is checked whether the first delay is negative (assuming a negative value represents leading) and whether the second delay is negative; finally, if both are negative, the output judgment result is the real fault source.

[0072] It is understandable that a wave speed consistency check (i.e., verifying whether the speed corresponding to the delay matches the rack speed of sound) can be used to assist in the determination, but this is not limited here.

[0073] In some embodiments, in complex scenarios facing fiber optic self-striking interference (non-rack conduction) or transient material impact interference (non-persistent fault), the physical properties and temporal characteristics of the signal source can be further verified through wave velocity-based physical consistency verification and time-domain slicing-based persistence statistics, thereby achieving the effect of eliminating false alarms and locking in real mechanical faults.

[0074] Specifically, a physical consistency check of wave velocity is first introduced. The first physical distance and the second physical distance between the candidate center point and the observation points on the left and right sides are obtained (usually determined by fiber calibration parameters), and then divided by the first time delay and the second time delay obtained in step S105, respectively, to calculate the apparent velocity of the sound wave propagating to the left and right.

[0075] Subsequently, it is determined whether both propagation speeds are within the preset effective range of the frame's sound velocity. This range is preset by those skilled in the art based on the sound wave propagation characteristics of the conveyor frame's metal material (usually steel) (approximately 5000 m / s) and a certain tolerance range (such as ±10%). The advantage of this is that the beat vibration wave velocity of the optical fiber itself (approximately several hundred meters per second) is much lower than the sound velocity of steel, and false alarms caused by optical fiber jitter can be physically eliminated by setting a speed threshold.

[0076] Secondly, before determining the true source of the fault, a time-domain persistence check is introduced. The acoustic signal to be analyzed is divided into multiple consecutive verification sub-windows on the time axis according to a preset sub-window size (e.g., set according to the roller rotation cycle). For each sub-window, cross-correlation calculation is performed independently, and the sub-window time delay is extracted to determine whether the divergence characteristic of the center point leading the sides is satisfied within this small time slice. The number of sub-windows that satisfy this characteristic is counted, and their persistence ratio to the total number of windows is calculated. This ratio reflects the temporal continuity of the vibration event. The preset fault confidence threshold is set based on the statistical regularity of a large number of historical fault samples (e.g., 80%). This is done because roller wear is continuous, while material drop impact is transient, and persistence statistics can effectively distinguish between the two.

[0077] Finally, only when the propagation speed matches the rack characteristics, the signal characteristics have sufficient persistence in time (proportional over-limit), and the center point time advance condition is met, is the candidate center point ultimately determined to be the real fault source.

[0078] By employing the above technical steps, non-structural conducted interference such as fiber optic wind-induced vibration is eliminated by utilizing the difference in medium propagation characteristics through wave velocity verification. By utilizing the difference in signal persistence through time-domain windowing statistics, occasional transient noise such as material impact is filtered out. The rigor of fault determination is ensured from both physical mechanism and time-domain statistics perspectives, thereby improving the system's positioning accuracy in complex interference environments.

[0079] In the above embodiments, the spatial coherence sensed by distributed optical fibers is used to verify the wave propagation characteristics of suspected high-energy regions. The time delay is extracted by calculating the cross-correlation function to confirm whether the vibration signal conforms to the physical propagation law of diverging from the candidate center to both sides. The judgment logic based on spatiotemporal causality can distinguish between real mechanical fault sources and random background noise or unidirectional conducted interference, thereby avoiding false alarms caused by relying solely on energy amplitude and ultimately improving the accuracy of conveyor roller fault location.

[0080] In some other embodiments of this application, when structural resonance exists in the frame, positioning deviation may occur due to energy accumulation at the resonance point. The conveyor idler roller fault location and decoupling method based on DAS spatial coherence provided in this application, by fusing kurtosis features to lock the impact source and using spectral centroid attenuation to verify the propagation path, can eliminate false resonance sources and achieve accurate positioning.

[0081] like Figure 2 The diagram shown is another flowchart illustrating the conveyor idler roller fault location and decoupling method based on DAS spatial coherence provided in this application, including the following steps: S201. Acquire the acoustic wave signal along the conveyor line and parse the acoustic wave signal into a two-dimensional data matrix.

[0082] S202. Perform a sum of squares and integrals on the two-dimensional data matrix to obtain the energy integral. Mark the set of spatial points that are continuously distributed and whose energy integral values ​​exceed the preset initial screening threshold as the initial screening area.

[0083] Steps S201-S202 and Figure 1 Steps S101-S102 in the illustrated embodiment are similar and can be found in the descriptions of steps S101-S102, which will not be repeated here.

[0084] S203. For each spatial point within the initial screening area, calculate the time-domain kurtosis value of the acoustic signal sequence.

[0085] Among them, the time-domain kurtosis value is a statistical measure used in probability theory and statistics to measure the kurtosis of the probability distribution of real random variables. In signal processing, it is used to represent the significance of the impact component in vibration signals. Mathematically, it is defined as the ratio of the fourth central moment to the square of the second central moment, and is used to characterize the impact characteristics of the signal waveform. Impact characteristics refer to whether there are sharp pulses or abrupt changes in the signal waveform that deviate from the normal distribution, which usually correspond to the impact or fracture events of mechanical parts.

[0086] Specifically, for each spatial channel (i.e., each sampling point on the optical fiber) within the initial screening area, the acoustic vibration time series x(t) within the current time window is extracted.

[0087] First, calculate the mean u and standard deviation i of the sequence. Next, subtract the mean from each data point in the sequence and raise the result to the fourth power. Sum the results and average them to obtain the fourth central moment. Finally, divide the fourth central moment by the fourth power of the standard deviation (i.e., the square of the variance) to obtain the dimensionless kurtosis value.

[0088] For normally distributed random noise, the kurtosis value is close to 3; however, for periodic impact signals caused by early faults such as pitting and peeling of idler roller bearings, the probability density distribution exhibits a thick tail characteristic, resulting in a kurtosis value significantly greater than 3.

[0089] By calculating kurtosis, it is possible to distinguish between low-frequency resonant signals with high energy but flat waveforms (low kurtosis) and real impact signals with moderate energy but sharp waveforms (high kurtosis).

[0090] In some embodiments, the calculation of time-domain kurtosis values ​​can be achieved in a variety of ways: Optional, standard definition calculation method: First, calculate the mean and variance of the signal sequence; second, calculate the fourth moment after centering the sequence; finally, use the formula K={E[(xu)} 4 ]} / i 4 Calculate the kurtosis value.

[0091] Optionally, the sliding window kurtosis method is used: First, a sliding sub-window shorter than the total duration is set; second, the local kurtosis value is calculated in each sub-window; finally, the maximum or average value of the kurtosis values ​​of all sub-windows is taken as the representative kurtosis of the spatial point to capture transient impacts.

[0092] Understandably, spectral kurtosis calculation can also be used to analyze non-Gaussianity within a specific frequency band, and this is not limited here.

[0093] S204. Normalize the energy integral value and temporal kurtosis value of each spatial point in the initial screening area to obtain the normalized energy sequence and normalized kurtosis sequence.

[0094] Normalization refers to the mathematical transformation process of mapping physical quantities of different dimensions and orders of magnitude to the same numerical range (usually [0, 1]), which facilitates subsequent weighted fusion; normalized energy sequence is used to represent the numerical set of the relative magnitude of energy at each point in the initial screening area; normalized kurtosis sequence is used to represent the numerical set of the relative strength of the impact characteristics at each point in the initial screening area.

[0095] Specifically, this is to eliminate the dimensional difference between energy values ​​(which are usually large and measured in volts squared or radians squared) and kurtosis values ​​(which are usually small and dimensionless).

[0096] Traverse all spatial points within the initial screening area to find the maximum value E in the energy integral sequence. max and minimum value E min and the maximum value K in the kurtosis sequence max and minimum value K min Using the range transformation formula, the energy value of each point x is mapped to (E... x -E min ) / (E max -E min ), to obtain the normalized energy value; similarly, map the kurtosis value of each point to (K x -K min ) / (K max -K min ), thus obtaining the normalized kurtosis value.

[0097] After processing, the values ​​of both sequences are distributed between 0 and 1, and the relative distribution of the original data is preserved, providing a unified benchmark for subsequent fusion evaluation.

[0098] S205. Perform point-to-point weighted fusion calculation on the normalized energy sequence and the normalized kurtosis sequence to obtain the fault source confidence index of each spatial point in the initial screening area.

[0099] Among them, point-to-point weighted fusion calculation refers to summing two feature indicators of the same spatial location after assigning different weight coefficients to them; the fault source confidence index is a comprehensive evaluation index used to quantify the probability that a certain spatial location is the real source of a fault, and the larger the value, the higher the probability.

[0100] Specifically, an energy weighting coefficient 'a' and a kurtosis weighting coefficient 'b' (usually a+b=1) are pre-set. These two coefficients are determined by those skilled in the art based on the significance of fault characteristics under actual working conditions. For example, the kurtosis weighting coefficient 'b' can be appropriately increased in a high-noise environment. For each spatial point x within the initial screening area, the normalized energy value E' is read. x and normalized kurtosis value K' x Calculate the fusion index Ix =a×E' x +b×K x Through this fusion, the confidence index of points with extremely high kurtosis (typical early fault characteristics) even if the energy is not the maximum, or points with high energy and high kurtosis (typical severe fault characteristics) will be significantly improved; while the confidence index of points with high energy but low kurtosis (typical resonance or environmental noise) will be suppressed.

[0101] In some embodiments, weighted fusion computation can be implemented in a variety of ways: Optional, linear weighted summation method: First, determine the weight coefficients a and b; second, perform multiplication and addition operations for each point; finally, generate the confidence index sequence.

[0102] Optional, multiplicative fusion method: First, set the nonlinear exponential parameter; second, calculate I. x =(E' x ) a ×(K' x ) b The effect of both having high values ​​is amplified by multiplication; finally, a confidence index sequence is generated.

[0103] It is understandable that fusion evaluation can also be achieved using reasoning rules based on fuzzy logic, and this is not limited here.

[0104] S206. Select the spatial point with the highest confidence index of the fault source as the candidate center point.

[0105] Among them, the maximum confidence index of the fault source refers to the highest value among all the index values ​​of all points calculated in step S205; the candidate center point refers to the spatial location that is finally selected by the algorithm as the reference for subsequent waveform analysis.

[0106] Specifically, the fault source confidence index sequence is scanned across the entire initial screening area, and the index position of the maximum value is found through a comparison algorithm. The spatial point corresponding to this position, after considering both vibration energy and impact pattern, is considered to best match the physical characteristics of the actual fault source. This point is marked as the candidate center point, and subsequent operations such as cross-correlation calculations and time delay extraction will all revolve around this point.

[0107] Compared to relying solely on energy for point selection, this method effectively avoids misjudging resonance points (high energy but no impact) on the rack as the center, thereby improving the robustness of positioning.

[0108] S207. Select a left observation point located at a preset distance to the left of the candidate center point and a right observation point located at a preset distance to the right of the candidate center point, respectively, and calculate the first cross-correlation function between the candidate center point and the left observation point, and the second cross-correlation function between the candidate center point and the right observation point.

[0109] S208. Based on the peak positions of the first cross-correlation function and the second cross-correlation function, obtain the first time delay and the second time delay of the candidate center point relative to the left observation point and the right observation point, respectively.

[0110] Steps S207-S208 and Figure 1 Steps S104-S105 in the illustrated embodiment are similar and can be found in the descriptions of steps S104-S105, which will not be repeated here.

[0111] S209. Transform the acoustic signal sequences of the candidate center point, the left observation point, and the right observation point from the time domain to the frequency domain to obtain the corresponding power spectral density functions.

[0112] Transforming from the time domain to the frequency domain refers to the process of using mathematical tools such as Fourier transform to convert a signal from a function with time as the independent variable into a function with frequency as the independent variable; the power spectral density function is a function used to characterize the distribution of signal power on the frequency axis, denoted as S(f), which reflects the energy intensity of the signal at each frequency component.

[0113] Specifically, the original acoustic signal sequences of the candidate center point, the left observation point, and the right observation point are read respectively. For each sequence, the Fast Fourier Transform (FFT) algorithm is applied to convert it into a frequency domain complex sequence.

[0114] Next, the square of the modulus of the complex sequence is calculated and divided by the frequency resolution to obtain the power spectral density sequence. To reduce random fluctuations, periodogram estimation methods such as Welch's method are often used, which involves segmenting the signal, applying windowing, performing an FFT, and then averaging to obtain a smoother and more stable power spectral density function. This function shows the frequency components of the signal at each point.

[0115] In some embodiments, the power spectral density function can be obtained in a variety of ways: Alternatively, the direct FFT periodogram method is used: First, a Hanning window is added to the signal sequence to reduce spectral leakage; second, an N-point FFT operation is performed; finally, the power spectrum is obtained by calculating the square of the amplitude.

[0116] It is understandable that parametric spectral estimation methods such as autoregressive models (AR models) can also be used to obtain the power spectrum, and this is not limited here.

[0117] S210. Calculate the spectral centroids of the candidate center point, the left observation point, and the right observation point using the power spectral density function.

[0118] Among them, the centroid of the spectrum refers to the geometric center of the spectrum energy distribution, similar to the concept of the center of gravity in mechanics. It represents the centroid of the signal energy distribution on the frequency axis and is used to represent the average frequency of the signal. Mathematically, it is defined as the weighted average of the frequency and the corresponding amplitude. The centroid of the distribution refers to the frequency position on the frequency axis where the spectral energy moment to the left and the spectral energy moment to the right of the point are equal.

[0119] Specifically, for each point (center, left, right), the power spectral density function S(f) obtained in step S209 is used for calculation. Each frequency value f... k With the corresponding power spectral amplitude S(f k Multiply by each product, then sum all the multiplications to obtain the first moment of the spectrum. Simultaneously, calculate the magnitudes of all power spectra S(f). k By directly summing them up, we can obtain the zeroth moment of the spectrum (i.e., the total energy).

[0120] Finally, dividing the first moment by the zeroth moment yields the centroid frequency of the spectrum. This metric accurately reflects the proportion of high-frequency components in the signal; a higher centroid indicates richer high-frequency energy, resulting in a sharper-sounding signal.

[0121] In some embodiments, the calculation of the spectral centroid can be achieved in a variety of ways: Optional, discrete integration method: First, obtain the discrete frequency points f. k and the corresponding PSD value P k Secondly, calculate the numerator ∑f k ⋅P k And the denominator ∑P k Finally, the centroid frequency is obtained by dividing the two.

[0122] Optionally, the frequency band weighting method involves: first, dividing the spectrum into several sub-bands; second, calculating the energy proportion of each sub-band as a weight; and finally, performing a weighted average of the center frequencies of each sub-band.

[0123] It is understandable that the median frequency can also be used as an alternative indicator to characterize the centroid of the frequency distribution, and this is not limited here.

[0124] S211. Compare the magnitudes of the spectral centroids of the candidate center point with those of the left and right observation points.

[0125] Among them, the magnitude relationship refers to the comparison of high and low values, which is used to judge the changing trend of frequency components of a signal during propagation.

[0126] Specifically, due to the damping and scattering inherent in the steel frame structure, high-frequency vibration waves attenuate faster during propagation than low-frequency waves. Therefore, theoretically, the actual vibration source (impact point) contains the richest high-frequency components and has the highest spectral centroid. As the wave propagates to both sides, the high-frequency components are gradually lost, causing the spectral centroid at the observation point to decrease.

[0127] The system numerically compares the spectral centroid FC of the candidate center point with the spectral centroid FL of the left observation point and the spectral centroid FR of the right observation point to check for the presence of the characteristic that FC > FL and FC > FR. This comparison process is to verify whether the signal conforms to the physical law of propagation from the center to both sides accompanied by high-frequency attenuation.

[0128] S212. If the signal at the candidate center point is ahead of the signals at the left and right observation points in time, and the centroid frequency of the spectrum at the candidate center point is higher than that at the centroid frequency of the spectrum at the left and right observation points, then the candidate center point is determined to be the real fault source.

[0129] Among them, the real fault source refers to the spatial location that simultaneously satisfies the time-domain divergence characteristics (wave source attributes) and the frequency-domain attenuation characteristics (primary attributes).

[0130] Specifically, summarizing the previous analysis results: First, time lead information from time-domain cross-correlation confirms that the wave propagates from the center outwards; second, centroid descent information from frequency-domain analysis confirms that high-frequency energy is generated at the center and attenuates outwards. Only when both conditions are met simultaneously can low-frequency resonant waves from a distance (characterized by time lead but low centroid) and non-causal interference be ruled out. A logical AND operation is performed; if all conditions are true, the final judgment result is output, confirming the candidate center point as the actual location of the idler roller fault, and triggering the corresponding alarm or recording process.

[0131] In some embodiments, the determination of the true source of the fault can be achieved in multiple ways: Optional, AND gate decision method: First, obtain the time domain decision flag (0 or 1); second, obtain the frequency domain decision flag (0 or 1); finally, if both are 1, output the final decision result.

[0132] It is understandable that a coherence coefficient verification step could be introduced to further verify the linear correlation of the waveform before making a judgment; however, this is not limited here.

[0133] In some embodiments, when the sensor mounting fixture is loose or there is local nonlinear high-frequency noise interference, waveform linear correlation can be verified by calculating the amplitude squared coherence function to distinguish between local independent noise and conducted fault signals and eliminate false high-frequency false alarms.

[0134] Specifically, a frequency domain coherence analysis mechanism is first introduced. Using the acoustic signal sequences from the candidate center point and the observation point on the left, the first amplitude squared coherence function is obtained by dividing the square of the magnitude of the cross-power spectral density by the product of their respective power spectral densities. Similarly, the second amplitude squared coherence function between the candidate center point and the observation point on the right is calculated. This function, with a range of 0 to 1, is used to quantify the linear causal relationship between the two signals at different frequency components.

[0135] Next, targeted coherence value extraction is performed. Since broadband coherence may be reduced by background noise, the spectral centroid frequency of the candidate center point calculated in step S210 is selected as the characteristic frequency point. The values ​​corresponding to this characteristic frequency point, i.e., the first coherence coefficient value and the second coherence coefficient value, are read from the first and second amplitude squared coherence function curves, respectively. The advantage of this approach is that the spectral centroid represents the main energy concentration band of the suspected fault signal, and checking the coherence of this band can accurately verify whether the "main energy" has actually undergone physical transmission.

[0136] Finally, the two extracted coherence coefficient values ​​are compared with a preset waveform correlation threshold. This threshold, determined through extensive field experiments based on the acoustic transfer function characteristics of the conveyor frame (e.g., set to 0.6), represents the minimum linear correlation that the signal should retain after propagation over a preset distance. Only when the coherence coefficient values ​​on both sides are greater than this threshold does it indicate that the vibration at the center point has indeed been linearly propagated to both sides through the frame, rather than a localized random chaotic vibration (such as sensor loosening and impact), thus ultimately determining the candidate center point as the true fault source.

[0137] By employing the above technical steps, the linear causal relationship of real mechanical fault signals during frame transmission is utilized, while local noise such as sensor loosening is uncorrelated with the remote signal. Through coherence verification at the main energy frequency band, false signals that, although conforming to time-frequency characteristics, lack physical transmission evidence are identified and eliminated, thereby reducing the false alarm rate caused by equipment loosening.

[0138] In the above embodiments, kurtosis statistical features and frequency domain centroid analysis mechanisms were introduced based on the initial screening using time-domain energy. First, by utilizing the sensitivity of the kurtosis index to impact fault signals, the vulnerability of energy-based point selection to resonance interference was corrected, ensuring that the candidate center point was locked at the physical impact source rather than a passive resonance point. Second, by utilizing the attenuation characteristics of high-frequency signals in the frame transmission, a dual verification logic of time-domain divergence and frequency-domain attenuation was constructed by comparing the spectral centroids of the center and both sides. This multi-dimensional decoupling strategy eliminates common low-frequency resonance false sources and local loosening noise in long-distance conveyors, improving the accuracy and robustness of idler roller fault location under complex operating conditions. The following describes an exemplary conveyor idler roller fault location and decoupling system 300 provided in an embodiment of this application. Figure 3 This is an exemplary hardware structure diagram of the conveyor idler roller fault location and decoupling system 300 provided in the embodiments of this application.

[0139] In some embodiments, the conveyor idler roller fault location and decoupling system 300 is a computer device or includes a computer device in the system. The computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database stores data. The network interface communicates with other external terminals or servers via a network connection. In some embodiments, the network interface can be a wired network interface; in some embodiments, it can also be a wireless network interface. When the computer program is executed by the processor, it implements the methods described in the embodiments of this application.

[0140] Those skilled in the art will understand that Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0141] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit it. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

[0142] As used in the above embodiments, depending on the context, the term "when..." can be interpreted as "if...", "after...", "in response to determining...", or "in response to detecting...". Similarly, depending on the context, the phrase "when determining..." or "if (the stated condition or event) is interpreted as "if determining...", "in response to determining...", "when (the stated condition or event) is detected", or "in response to detecting (the stated condition or event)".

[0143] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive), etc.

[0144] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. This program can be stored in a computer-readable storage medium, and when executed, it can include the processes described in the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM or random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A method for fault location and decoupling of conveyor idler rollers based on DAS spatial coherence, characterized in that, include: Acquire acoustic signals along the conveyor line and parse the acoustic signals into a two-dimensional data matrix; The horizontal axis of the two-dimensional data matrix represents the distance along the optical fiber, and the vertical axis represents the sampling time point. The energy integral is obtained by performing a sum of squares and integral on the two-dimensional data matrix. The set of spatial points that are continuously distributed and whose energy integral values ​​exceed the preset initial screening threshold is marked as the initial screening area. Within the initial screening area, the point with the largest amplitude of the energy integral is selected as the candidate center point; Select a left observation point located at a preset distance to the left of the candidate center point and a right observation point located at a preset distance to the right of the candidate center point, respectively, and calculate the first cross-correlation function between the candidate center point and the left observation point, and the second cross-correlation function between the candidate center point and the right observation point; Based on the peak positions of the first cross-correlation function and the second cross-correlation function, the first time delay and the second time delay of the candidate center point relative to the left observation point and the right observation point are obtained respectively. If the signal at the candidate center point is ahead of the signals at the left and right observation points in time, the candidate center point is determined to be the actual fault source.

2. The method according to claim 1, characterized in that, When the signal at the candidate center point is time-advancing compared to the signals at the left and right observation points, the candidate center point is determined to be the actual fault source. This specifically includes: Obtain the first physical distance between the candidate center point and the left observation point, and the second physical distance between the candidate center point and the right observation point; The propagation speed on the left is obtained by dividing the first physical distance by the first time delay, and the propagation speed on the right is obtained by dividing the second physical distance by the second time delay. Determine whether the propagation speed on the left and the propagation speed on the right are both within the preset effective range of the frame sound velocity; the effective range of the frame sound velocity is set based on the sound wave propagation characteristics of the metal material of the conveyor frame; If so, then if the signal of the candidate center point is ahead of the signals of the left observation point and the right observation point in time, the candidate center point is determined to be the real fault source.

3. The method according to claim 2, characterized in that, Before determining that the candidate center point is the actual fault source, the method further includes: The acoustic signals of the candidate center point, the left observation point, and the right observation point are divided into multiple consecutive verification sub-windows on the time axis according to a preset sub-window size. For each verification sub-window, calculate the cross-correlation function between the candidate center point and the left observation point, and between the candidate center point and the right observation point, and extract the corresponding sub-window time delay; The number of verification sub-windows that satisfy the candidate center point's signal in time is ahead of the signals of the left observation point and the right observation point is counted, and the persistence ratio of the total number of verification sub-windows is calculated. If the persistence ratio exceeds a preset fault confidence threshold, the candidate center point is determined to be the actual fault source.

4. The method according to claim 1, characterized in that, The step of selecting a left observation point at a predetermined distance to the left of the candidate center point and a right observation point at a predetermined distance to the right of the candidate center point, respectively, and calculating a first cross-correlation function between the candidate center point and the left observation point, and a second cross-correlation function between the candidate center point and the right observation point, specifically includes: The original acoustic signal sequences of the candidate center point, the left observation point, and the right observation point are extracted respectively. The center original acoustic signal sequence of the candidate center point is differentially processed with the adjacent signal sequences of adjacent spatial points to obtain the center differential signal, and differential processing is performed on the left observation point and the right observation point to obtain the left differential signal and the right differential signal respectively. The cross-correlation function between the center differential signal and the left differential signal is calculated as the first cross-correlation function, and the cross-correlation function between the center differential signal and the right differential signal is calculated as the second cross-correlation function.

5. The method according to claim 1, characterized in that, When the signal at the candidate center point is time-advancing compared to the signals at the left and right observation points, the candidate center point is determined to be the actual fault source. This specifically includes: The acoustic signal sequences of the candidate center point, the left observation point, and the right observation point are transformed from the time domain to the frequency domain to obtain the corresponding power spectral density functions. The spectral centroids of the candidate center point, the left observation point, and the right observation point are calculated using the power spectral density function, respectively; the spectral centroids characterize the centroid of the signal energy distribution on the frequency axis. Compare the magnitudes of the spectral centroids of the candidate center point with those of the left and right observation points; If the signal at the candidate center point is ahead of the signals at the left and right observation points in time, and the centroid frequency of the spectrum at the candidate center point is higher than that at the left and right observation points, then the candidate center point is determined to be the actual fault source.

6. The method according to claim 5, characterized in that, Before determining that the candidate center point is the actual fault source, the method further includes: The first amplitude squared coherence function is calculated using the acoustic signal sequence of the candidate center point and the left observation point, and the second amplitude squared coherence function is calculated using the acoustic signal sequence of the candidate center point and the right observation point. Extract the first coherence coefficient value of the first amplitude squared coherence function at the centroid frequency of the spectrum at the candidate center point, and the second coherence coefficient value of the second amplitude squared coherence function at the centroid frequency of the spectrum at the candidate center point; If both the first coherence coefficient value and the second coherence coefficient value are greater than a preset waveform correlation threshold, the candidate center point is determined to be the real fault source.

7. The method according to claim 1, characterized in that, The step of selecting the point with the largest amplitude of the energy integral as the candidate center point within the initial screening area specifically includes: For each spatial point within the initial screening area, the time-domain kurtosis value of the acoustic signal sequence is calculated; the time-domain kurtosis value is used to characterize the impact characteristics of the signal waveform. The energy integral value and temporal kurtosis value of each spatial point in the initial screening area are normalized to obtain the normalized energy sequence and the normalized kurtosis sequence. The normalized energy sequence and the normalized kurtosis sequence are weighted and fused point-to-point to obtain the fault source confidence index of each spatial point in the initial screening area. The spatial point with the highest confidence index of the fault source is selected as the candidate center point.

8. A fault location and decoupling system for conveyor idlers, characterized in that, The conveyor idler roller fault location and decoupling system includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to cause the conveyor idler roller fault location and decoupling system to perform the method as described in any one of claims 1-7.

9. A computer program product containing instructions, characterized in that, When the computer program product is run on the conveyor idler fault location and decoupling system, the conveyor idler fault location and decoupling system performs the method as described in any one of claims 1-7.

10. A computer-readable storage medium comprising instructions, characterized in that, When the instruction is executed on the conveyor idler fault location and decoupling system, the conveyor idler fault location and decoupling system performs the method as described in any one of claims 1-7.

Images