CN-121997170-A - DAS road cross-correlation risk evaluation method based on SOM segmentation

CN121997170ACN 121997170 ACN121997170 ACN 121997170ACN-121997170-A

Abstract

The invention belongs to the technical field of urban road structure safety monitoring, and discloses a DAS road cross-correlation risk evaluation method based on SOM segmentation, which specifically comprises the following steps of receiving DAS signals of roads along optical fibers collected by a distributed optical fiber acoustic wave sensing system, and carrying out noise reduction and enhancement treatment on the DAS signals; the method comprises the steps of extracting characteristic parameter vectors of signal channels in a distributed optical fiber acoustic wave sensing system based on silence period data in processed DAS signals, clustering the signal channels with adjacent spatial positions and similar characteristic parameter vectors through a self-organizing mapping network, dividing the optical fiber into a plurality of independent segments with continuous space according to clustering results, respectively calculating multi-mode cross-correlation functions among the signal channels in each independent segment, extracting and integrating peak sharpness indexes in each cross-correlation function, and effectively solving the problem of poor road risk assessment accuracy caused by uneven DAS signal channel response in the prior art.

Inventors

HOU SHITONG
JIANG ANGUO
WU DONG
FAN JIANHUA
LI YAOJIE
DING YOULIANG
WU JING
ZENG YIHUA

Assignees

东南大学

Dates

Publication Date: 20260508
Application Date: 20260119

Claims (8)

1. A DAS road cross-correlation risk evaluation method based on SOM segmentation is characterized by comprising the following steps: Receiving DAS signals along the optical fibers of a road collected by a distributed optical fiber acoustic wave sensing system, and filtering and wave selecting the DAS signals; Extracting characteristic parameter vectors of all signal channels in the distributed optical fiber acoustic wave sensing system based on silence period data in the processed DAS signals; Clustering signal channels with adjacent spatial positions and similar characteristic parameter vectors through a self-organizing map network, and dividing the optical fiber into a plurality of independent segments with continuous space according to clustering results; Based on effective excitation data in the DAS signals after filtering and wave selecting processing, multi-mode cross-correlation functions among signal channels are respectively calculated in each independent segment, peak sharpness indexes in the cross-correlation functions are extracted and integrated, and a channel-level reliability index is generated for each signal channel; Based on the spatial distribution of the reliability index along the optical fiber, a risk optical fiber region corresponding to a signal channel with abnormal reliability index is determined by adopting an identification method combining global statistical characteristics and local mutation, and a risk visual map is generated.
2. The method for evaluating the cross-correlation risk of a DAS road based on SOM segmentation according to claim 1, wherein the feature parameter vector of each signal channel in the distributed optical fiber acoustic wave sensor system is extracted based on the silence period data in the processed DAS signal, specifically comprising the following steps: respectively calculating and extracting steady-state characteristic parameters of each signal channel in the distributed optical fiber acoustic wave sensing system in the data of the silent period, wherein the steady-state characteristic parameters comprise power spectral density Variance of Kurtosis degree Time average The specific calculation formula is as follows: Wherein, the Represent the first The signal channels are at time DAS signal of (a); Sampling length for silence periods; The steady-state characteristic parameters together form a characteristic parameter vector of a corresponding signal channel, and the specific expression is as follows: Wherein, the Represent the first And the characteristic parameter vectors corresponding to the signal channels.
3. The method for evaluating the cross-correlation risk of the DAS road based on the SOM segmentation according to claim 1, wherein the signal channels with adjacent spatial positions and similar characteristic parameter vectors are clustered through a self-organizing map network, and the optical fiber is divided into a plurality of independent segments with continuous space according to the clustering result, and the method specifically comprises the following steps: Inputting the characteristic parameter vectors of each signal channel into a self-organizing mapping network for training, and outputting a group of weight vectors representing different types of characteristic parameter vectors by the self-organizing mapping network; Matching the characteristic parameter vector and the weight vector of each signal channel, attributing the characteristic parameter vector to the category corresponding to the weight vector most similar to the characteristic parameter vector, and respectively distributing a clustering label to each signal channel; and merging the corresponding areas of the signal channels with the same clustering labels and adjacent spatial positions on the optical fibers, dividing the areas into independent sections, and dividing the optical fibers into a plurality of independent sections with continuous space according to the clustering result.
4. The method for evaluating the cross-correlation risk of a DAS road based on SOM segmentation according to claim 3, wherein in each independent segmentation, a multi-mode cross-correlation function between signal channels is calculated, peak sharpness indexes in the cross-correlation functions are extracted and integrated, and a channel-level reliability index is generated for each signal channel, comprising the following steps: For any signal channel i in the independent segment, calculating a multi-mode cross-correlation function between the signal channel i and the rest signal channels in the independent segment, wherein the multi-mode cross-correlation function comprises a phase cross-correlation function, an amplitude cross-correlation function and a generalized cross-correlation; extracting peak sharpness indexes in various cross-correlation functions corresponding to the signal channel i, wherein the peak sharpness indexes comprise peak energy ratio, peak side lobe ratio and peak root mean square ratio; and fusing the peak sharpness indexes corresponding to the signal channel i to generate the reliability index corresponding to the signal channel i.
5. The method of claim 4, wherein the step of merging the peak sharpness indicators corresponding to the signal channel i includes calculating a root mean square of the peak sharpness indicators.
6. The method for evaluating the cross-correlation risk of the DAS road based on SOM segmentation according to claim 4, wherein the risk fiber region corresponding to the signal channel with abnormal reliability index is determined by adopting an identification method combining global statistical features and local mutation based on the spatial distribution of the reliability index along the optical fiber, and a risk visualization map is generated, and specifically comprises the following steps: Calculating global statistical characteristics of all signal channel reliability indexes, wherein the global statistical characteristics comprise global mean values And global standard deviation ; And carrying out normalization processing on the reliability indexes corresponding to each signal channel, wherein the calculation formula of the normalization processing is as follows: Wherein, the The reliability index corresponding to the signal channel i is represented; a normalized reliability index representing the signal channel i; setting a global threshold based on global statistical features The specific calculation formula is as follows: Wherein, the Is an experience coefficient; when the reliability index corresponding to the signal channel is smaller than the global threshold value When the corresponding signal channel is judged to be a global low-reliability channel; the local mutation identification method adopting the valley method is used for determining a local low-reliability channel and comprises the following steps: Defining the spatial rate of change of the reliability index at signal path i The specific mathematical expression is as follows: when the following determination condition is satisfied, the determination signal path i is a local low reliability path, and the expression of the determination condition is as follows: Wherein, the Is a local threshold; The optical fiber areas corresponding to the global low-reliability channel and the local low-reliability channel jointly form a risk optical fiber area; And generating a risk visualization map according to the signal channel positions of the global low-reliability channel and the local low-reliability channel in the risk optical fiber region and the corresponding reliability indexes.
7. The method for DAS road cross-correlation risk assessment based on SOM segmentation of claim 1, further comprising classifying and identifying risk events based on reliability indicators.
8. The method for evaluating the cross-correlation risk of the DAS road based on SOM segmentation according to claim 7, wherein the risk event is classified and identified based on the reliability index, specifically comprising the steps of: Fitting various probability distributions based on the numerical distribution of the reliability indexes aiming at different risk events to be identified and classified, and extracting the mean value and variance of each fit distribution; The method comprises the steps of analyzing the mean value and variance of the numerical distribution of the reliability indexes of different risk events, and realizing the preliminary identification and classification of the data of different risk events; For data of different risk events, selecting a maximum reliability channel in each reliability index curve as a representative channel, and acquiring corresponding characteristic parameter vectors; performing dimension reduction treatment on the characteristic parameter vector by adopting a principal component analysis method; And carrying out cluster analysis on the feature parameter vector after the dimension reduction through an unsupervised learning algorithm k-means to realize the identification of different types of risk events.

Description

DAS road cross-correlation risk evaluation method based on SOM segmentation Technical Field The invention belongs to the technical field of urban road structure safety monitoring, and particularly relates to a DAS road cross-correlation risk evaluation method based on SOM segmentation. Background The long-term performance of urban roads is affected by a variety of external factors, leading to degradation and potential safety hazards. Some underground cavities or ground risks are particularly serious, and irreversible damage can be caused to traffic facilities and lives and properties of people. The existing urban road structure monitoring method is dependent on point sensors or manual detection, and large-scale, continuous and high-sensitivity risk identification is difficult to realize. The distributed optical fiber acoustic wave sensing (DAS) technology can realize real-time vibration sensing along the whole optical fiber, but due to stronger signal noise, uneven channel response and huge data dimension, the consistency change among optical fiber channels is difficult to accurately reflect in the traditional single-channel or fixed window cross-correlation analysis, and the early recognition capability of underground cavities and abnormal road structures is limited, so that the problem of poor road risk assessment accuracy caused by uneven DAS signal channel response exists in the prior art. Disclosure of Invention Aiming at the defects of the prior art, the invention aims to provide a DAS road cross-correlation risk evaluation method based on SOM segmentation, which solves the problem of poor road risk evaluation accuracy caused by uneven DAS signal channel response in the prior art. The aim of the invention can be achieved by the following technical scheme: a DAS road cross-correlation risk evaluation method based on SOM segmentation specifically comprises the following steps: Receiving DAS signals along the optical fibers of a road collected by a distributed optical fiber acoustic wave sensing system, and filtering and wave selecting the DAS signals; Extracting characteristic parameter vectors of all signal channels in the distributed optical fiber acoustic wave sensing system based on silence period data in the processed DAS signals; Clustering signal channels with adjacent spatial positions and similar characteristic parameter vectors through a self-organizing map network, and dividing the optical fiber into a plurality of independent segments with continuous space according to clustering results; Based on effective excitation data in the DAS signals after filtering and wave selecting processing, multi-mode cross-correlation functions among signal channels are respectively calculated in each independent segment, peak sharpness indexes in the cross-correlation functions are extracted and integrated, and a channel-level reliability index is generated for each signal channel; Based on the spatial distribution of the reliability index along the optical fiber, a risk optical fiber region corresponding to a signal channel with abnormal reliability index is determined by adopting an identification method combining global statistical characteristics and local mutation, and a risk visual map is generated. Further, based on silence period data in the processed DAS signal, extracting feature parameter vectors of each signal channel in the distributed optical fiber acoustic wave sensing system, specifically comprising the following steps: respectively calculating and extracting steady-state characteristic parameters of each signal channel in the distributed optical fiber acoustic wave sensing system in the data of the silent period, wherein the steady-state characteristic parameters comprise power spectral density Variance ofKurtosis degreeTime averageThe specific calculation formula is as follows: Wherein, the Represent the firstThe signal channels are at timeDAS signal of (a); Sampling length for silence periods; The steady-state characteristic parameters together form a characteristic parameter vector of a corresponding signal channel, and the specific expression is as follows: Wherein, the Represent the firstAnd the characteristic parameter vectors corresponding to the signal channels. Further, clustering signal channels with adjacent spatial positions and similar characteristic parameter vectors through a self-organizing map network, and dividing the optical fiber into a plurality of spatially continuous independent segments according to a clustering result, wherein the method specifically comprises the following steps: Inputting the characteristic parameter vectors of each signal channel into a self-organizing mapping network for training, and outputting a group of weight vectors representing different types of characteristic parameter vectors by the self-organizing mapping network; Matching the characteristic parameter vector and the weight vector of each signal channel, attributing the characteristic p