An Adaptive Anomaly Labeling Method for DAS Signals

Abstract

This invention relates to the field of vibration signal analysis of fiber optic acoustic wave sensors (DAS), and provides an adaptive anomaly labeling method for DAS signals. The method includes: signal data preprocessing, extracting vibration features from the acquired DAS signal data X by frame division. Firstly, each channel needs to be framed: the original signal is framed using frame overlap. Let the frame length be l and the frame shift be q, then the data of the j-th channel in the k-th frame is represented as s. kj =[x q(k‑1)+1,j ,x q(k‑1)+2,j ,...,x q(k‑1)+l,j The beneficial effect of this invention is that it achieves automatic labeling of vibration samples in a more efficient and accurate manner.

Description

Technical Field

[0001] This invention belongs to the field of vibration signal analysis of fiber optic acoustic wave sensors, and particularly relates to an adaptive abnormal signal annotation method for vibration signals of fiber optic distributed acoustic wave sensors. Background Technology

[0002] As a cutting-edge field in distributed optical fiber monitoring technology, distributed optical fiber acoustic sensors (DAS) possess advantages such as strong resistance to electromagnetic and radio frequency interference, flexible deployment, and wide monitoring range. They hold immense potential in areas such as pipeline leak detection, perimeter protection, and structural monitoring. DAS senses the intensity of vibration events along the optical fiber, detects the phase change caused by vibration based on the Rayleigh scattering principle, and acquires the vibration acoustic signal by measuring the intensity of the scattered coherent light. Simultaneously, due to the constancy of the speed of light, the disturbance point can be spatially located based on the time difference between the scattered and incident light. A single optical fiber contains multiple signal acquisition channels, and the spacing between these channels constitutes the spatial resolution, which typically ranges from 1 meter to 10 meters in accuracy. Therefore, the vibration signal measured by DAS is a two-dimensional data matrix that simultaneously includes both temporal and spatial information.

[0003] Before analyzing vibration signal data, it is necessary to first collect and label vibration event samples. Early labeling of vibration events primarily relied on manual identification, using observation of the waterfall plot fluctuations of DAS-acquired signals to mark the time of occurrence and the range of channels involved. However, when faced with labeling scenarios involving diverse data types and massive amounts of data, manual labeling revealed drawbacks such as long processing times, low efficiency, and high rates of mislabeling and omissions. Summary of the Invention

[0004] The purpose of this invention is to provide an adaptive anomaly labeling method for vibration signals of fiber optic distributed acoustic wave sensors, namely, an adaptive anomaly labeling method for DAS signals, which aims to solve the problems mentioned in the background art.

[0005] The present invention is implemented as follows: an adaptive anomaly signal labeling method for vibration signals of an optical fiber distributed acoustic wave sensor, the method comprising the following steps:

[0006] The method includes:

[0007] Signal data preprocessing involves extracting vibration features from the acquired DAS signal data X by framing it. First, each channel needs to be framed: the original signal is framed using frame overlap. Let the frame length be l and the frame shift be q, then the data of the k-th frame of the j-th channel is represented as s. kj =[x q(k-1)+1,j ,xq(k-1)+2,j ,...,x q(k-1)+l,j If 1≤k≤t, 1≤j≤m, then the acquired signal matrix X is converted into a frame matrix X'.

[0008]

[0009] Windowing is applied to each frame of signal to convert the acquired signal matrix X into a frame energy matrix X”;

[0010] Valid frame selection: Based on the experimental scenario and the strength of abnormal signal energy, an average energy threshold θ for abnormal frames is set. Frames with energy greater than or equal to the threshold are marked as 1, and frames with energy less than the threshold are marked as 0. The frame energy matrix is ​​then transformed into a 0-1 matrix.

[0011]

[0012] in,

[0013] Sparse matrix location clustering converts the 0-1 matrix into a sparse matrix, performs DBSCAN clustering on the locations, and sets the parameters to eps = 1.6 and min_samples = 2. The smallest covering rectangular region in each cluster is selected and called the 0-1 rectangular cluster region. Let each 0-1 rectangular cluster region be:

[0014]

[0015] The center position of the rectangular cluster region is set by calculating the horizontal and vertical histograms of each 01 rectangular cluster region and selecting the position of the center point (loc_r, loc_c) as the maximum position of the horizontal and vertical histograms.

[0016]

[0017] The core rectangular cluster region is marked, starting from the center point and moving upwards by N. up Downward N down , to the left N left N to the right right Calculate the 1-norm of the differences between adjacent rows or columns in turn.

[0018]

[0019] The traversal is terminated based on the threshold δ. The starting row, ending row, starting column, and ending column of the core rectangular region of each rectangular cluster region are determined. Finally, the core rectangular region is traced back to the original data and the core abnormal region is marked.

[0020] As a further aspect of the present invention, the step of windowing each frame signal to convert the acquired signal matrix into a frame energy matrix specifically includes:

[0021] Calculate the average energy p of each frame of the signal. kj =mean(abs(FFT(s)) kj If 1≤k≤t, 1≤j≤m, then the acquired signal matrix X is converted into a frame energy matrix X”.

[0022]

[0023] This invention provides an adaptive anomaly signal annotation method for vibration signals from a fiber optic distributed acoustic sensor. First, the signal data undergoes preprocessing such as signal framing and windowing. Second, vibration features are extracted, and vibration event frames are distinguished from non-vibration event frames based on these features. Then, a low-density clustering algorithm is used to merge and classify adjacent event frames. Finally, the optimal data range is determined by calculating the 1-norm of adjacent channels and adjacent time points, and the start and end times of the vibration event and the start and end range matrix of the involved channels are output. This method provides an effective solution to the shortcomings of manual annotation. Attached Figure Description

[0024] Figure 1 This is the main flowchart of an adaptive abnormal signal annotation method for vibration signals from a fiber optic distributed acoustic sensor.

[0025] Figure 2 This is a schematic diagram illustrating the marking of the core abnormal region in a continuous ten-second fiber optic sensing data in an adaptive abnormal signal annotation method for vibration signals of a fiber optic distributed acoustic sensor.

[0026] Figure 3 This is a lateral histogram of the 01 rectangular cluster region in an adaptive abnormal signal annotation method for vibration signals of a fiber optic distributed acoustic sensor.

[0027] Figure 4 This is a longitudinal histogram of the 01 rectangular cluster region in an adaptive abnormal signal annotation method for vibration signals of a fiber optic distributed acoustic sensor. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0029] The specific implementation of the present invention will be described in detail below with reference to specific embodiments.

[0030] This invention provides an adaptive abnormal signal annotation method for vibration signals from fiber optic distributed acoustic sensors. Addressing the shortcomings of current manual annotation methods, this method combines vibration event feature extraction, low-density clustering algorithms, and the nearest neighbor 1 paradigm to determine the time and channel range of vibration events. Ultimately, it achieves automatic annotation of vibration samples in a more efficient and accurate manner.

[0031] like Figure 1 The diagram shown is a main flowchart of an adaptive anomaly signal annotation method for vibration signals of an optical fiber distributed acoustic wave sensor according to an embodiment of the present invention. The adaptive anomaly signal annotation method for vibration signals of an optical fiber distributed acoustic wave sensor includes:

[0032] Assume the raw data signal acquired by the DAS system is matrix X.

[0033]

[0034] Signal data preprocessing involves extracting vibration features from the acquired DAS signal data X by framing it. First, each channel needs to be framed: the original signal is framed using frame overlap. Let the frame length be l and the frame shift be q, then the data of the k-th frame of the j-th channel can be represented as s. kj =[x q(k-1)+1,j ,x q(k-1)+2,j ,...,x q(k-1)+l,j If 1≤k≤t, 1≤j≤m, then the acquired signal matrix X is converted into a frame matrix X'.

[0035]

[0036] To reduce the signal truncation effect caused by framing (the periodicity of the original signal is truncated, leading to spectral leakage in the vibration signal), windowing can be applied to each frame of the signal to reduce spectral leakage. The average energy p of each frame of the signal is then calculated. kj =mean(abs(FFT(s)) kj If 1≤k≤t, 1≤j≤m, then the acquired signal matrix X is converted into a frame energy matrix X”.

[0037]

[0038] For valid frame filtering, based on the experimental scenario and the strength of abnormal signal energy, an average energy threshold θ for abnormal frames is set. Frames with energy greater than or equal to the threshold are marked as 1, and frames with energy less than the threshold are marked as 0. Thus, the frame energy matrix can be transformed into a 0-1 matrix.

[0039]

[0040] in,

[0041] Sparse matrix location clustering. The 0-1 matrix is ​​converted to a sparse matrix, and DBSCAN clustering is performed on the locations with parameters set to eps = 1.6 and min_samples = 2. The smallest covering rectangle region (containing all elements within the cluster) of each cluster is selected, called the optimal 0-1 rectangle cluster region. Let each optimal 0-1 rectangle cluster region be:

[0042]

[0043] Setting the center position of the optimal rectangular cluster region. Calculate the horizontal and vertical histograms of each optimal 01 rectangular cluster region, and select the position of the maximum value in the horizontal and vertical histograms as the center point (loc_r, loc_c);

[0044]

[0045] The core rectangular cluster region is marked, starting from the center point and moving upwards by N. up Downward N down , to the left N left N to the right right Calculate the 1-norm of the differences between adjacent rows (or columns) in turn.

[0046]

[0047] The traversal is terminated based on a threshold δ, and the starting row, ending row, starting column, and ending column positions of the core rectangular region for each optimal rectangular cluster region are determined. Finally, the core rectangular regions are traced back to the original data, and the core abnormal regions are marked.

[0048] The above method will be described below with reference to specific embodiments:

[0049] The experimental data used were 10 consecutive seconds of signal data from channels 176-375 (0.1 billion data points) collected by a fiber optic vibration sensor with a sampling rate of 10000. Actual experimental labeling results can be found in the appendix. Figure 2 Due to the large number of data points, it is impossible to display all the real data in detail. Therefore, the following only uses a small sample of data to illustrate the core calculation process:

[0050] Assume the calculated frame average energy matrix is ​​shown in Table 1 below:

[0051]

[0052]

[0053]

[0054] (1) If the threshold for abnormal frames is set to 15, the 01 matrix after filtering and transformation of valid frames is shown in the table below.

[0055] As shown in Table 2:

[0056]

[0057]

[0058]

[0059] 2) After clustering the positions marked 1 in the table above using DBSCAN, the optimal 01 rectangular cluster regions obtained are shown in Table 3 below (the region marked 1 in bold):

[0060]

[0061]

[0062]

[0063] (3) Calculation of the optimal coordinates of the horizontal and vertical histograms of the optimal 01 rectangular cluster region, through methods such as... Figure 3 and Figure 4 The two histograms show that the optimal center position is (frame-3, channel-5).

[0064] (4) Traverse upward, downward, leftward, and rightward at the center position (frame-3, channel-5), set the threshold to 0.6, and mark the final core rectangular cluster area (the rectangular area formed by bold 1) as shown in Table 4 below.

[0065]

[0066]

[0067]

[0068] The above embodiments of the present invention provide an adaptive abnormal signal labeling method for vibration signals of fiber optic distributed acoustic wave sensors.

[0069] In order for the above methods and systems to operate smoothly, the system may include more or fewer components than those described above, or combine certain components, or different components, in addition to the various modules mentioned above. For example, it may include input / output devices, network access devices, buses, processors, and memory.

[0070] The processor referred to can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. This processor is the control center of the system, connecting various parts through various interfaces and lines.

[0071] The aforementioned memory can be used to store computer and system programs and / or modules. The aforementioned processor implements the various functions mentioned above by running or executing the computer programs and / or modules stored in the memory, and by calling the data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as information collection template display function, product information publishing function, etc.). The data storage area may store data created based on the use of the berth status display system (such as product information collection templates corresponding to different product types, product information that different product providers need to publish, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0072] It should be understood that although the steps in the flowcharts of the various embodiments of the present invention are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the various embodiments may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least a portion of the sub-steps or stages of other steps.

[0073] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0074] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

[0075] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An adaptive anomaly labeling method for DAS signals, characterized in that, The method includes: Signal data preprocessing involves extracting vibration features from the acquired DAS signal data X by framing it. First, each channel needs to be framed: the original signal is framed using frame overlap. Let the frame length be l and the frame shift be q, then the data of the k-th frame of the j-th channel is represented as s. kj =[x q(k-1)+1,j ,x q(k-1)+2,j ,...,x q(k-1)+l,j If 1≤k≤t, 1≤j≤m, then the acquired signal matrix X is converted into a frame matrix X', where, Windowing is applied to each frame of signal to convert the acquired signal matrix X into a frame energy matrix X”; Valid frame selection: Based on the experimental scenario and the strength of abnormal signal energy, an average energy threshold θ for abnormal frames is set. Frames with energy greater than or equal to the threshold are marked as 1, and frames with energy less than the threshold are marked as 0. The frame energy matrix is ​​then transformed into a 0-1 matrix. in, Sparse matrix location clustering converts the 0-1 matrix into a sparse matrix, performs DBSCAN clustering on the locations, and sets the parameters to eps = 1.6 and min_samples = 2. The smallest covering rectangular region in each cluster is selected and called the 0-1 rectangular cluster region. Let each 0-1 rectangular cluster region be: The center position of the rectangular cluster region is set by calculating the horizontal and vertical histograms of each 01 rectangular cluster region, and selecting the position of the maximum value in the horizontal and vertical histograms as the center point (loc_r, loc_c). The core rectangular cluster region is marked, and starting from the center point, the 1 norm of the differences between adjacent rows or columns is calculated sequentially upwards, downwards, leftwards, and rightwards. The traversal is terminated based on the threshold δ. The starting row, ending row, starting column, and ending column of the core rectangular region of each rectangular cluster region are determined. Finally, the core rectangular region is traced back to the original data and the core abnormal region is marked.

2. The adaptive anomaly labeling method for DAS signals according to claim 1, characterized in that, The step of windowing each frame of signal to convert the acquired signal matrix into a frame energy matrix specifically includes: Calculate the average energy p of each frame of the signal. kj =mean(abs(FFT(s)) kj If 1≤k≤t, 1≤j≤m, then the acquired signal matrix X is converted into a frame energy matrix X”, where,

Images