Vibration Signal Recognition In Underground Cable Intrusion Monitoring

In recent years, with the increasingly common use of underground cables for powersupply in the power distribution system, cables in use have often been break by external forces.The advantage of cable power supply is severely affected. So protecting the cable fromexternal forces has become the urgent problem of power supply bureau. With the rapiddevelopment of distributed optical fiber sensing technology, people can use this technology toachieve the real-time measurement of the vibration signal around the transmission line. Withthese signals, we could get warning against external forces on transmission line and theexception place information.Based on the use of distributed fiber optic sensing system for real-time monitoring, wecan design program to classify and recognize the detected signals by different external eventscause vibration. It’s helpful for reducing the false alarm rate and providing more information,so that relevant staff could make better decisions. This article is committed to the algorithm ofsignal recognition for the data collected by the underground cable intrusion monitoringsystem using digital signal processing and data mining technology. The main research work isdescribed as follows:1. Design the preprocessing method for vibration signal, including bandpass filtering,wavelet denoising and segmentation of vibration fragment. According to the result ofexperiment, the preprocessing method could reduce the noise and separate the vibration signalfragment. It’s helpful for follow-up vibration signal recognition.2. Design the vibration signal recognition algorithm, including the feature extraction ofsignals and the recognition method of vibration fragment and vibration event. Taking intoaccount the diversity of the vibration signal and recognition algorithm’s scalability, we choosethe support vector machine (SVM) to construct the recognition algorithm model. For solvingthe small sample size, nonlinear and high dimensional pattern recognition problem, there areadvantages and excellent generalization ability of SVM.3. Design intelligent early warning mechanism for underground cable intrusionmonitoring. First, synthesize the short-term features of the vibration signal and the signalchange trend of a certain time period. This mechanism could provide vibration eventrecognition in a longer time dimension and effectively improve the recognition rate ofvibration events. Second, design incremental learning mechanism. The recognition algorithmcould use the sample signals which are submitted by the operator to complete incremental learning. So the system’s ability to recognize vibration events is improved through continuouslearning.This paper studies the vibration signal processing and recognition method inunderground cable intrusion monitoring system, and completes experiment according to thesample data. On six kinds of vibration event sample, fragment recognition rate up to85.84%,and the event recognition rate can reach92.69%, which can meet the application requirements.The vibration signal processing and recognition method is universal. In future, it could beextended to the perimeter security, long distance oil and gas pipeline monitoring and otherfields.