Event Detection and Classification for Fiber Optic Perimeter Intrusion Detection System

Abstract

A perimeter intrusion detection system (PIDS) is critical for the security of a shale gas field. Among many technologies, the fiber optic sensor-based method is the most widely used, due to its passive, low-cost, long-life, and strong anti-interference ability and strong environmental adaptability. This article proposes an event detection and classification method for a fiber optic PIDS. In general, three types of features are extracted for an improved double-threshold method to improve the probability of detection. Also, the detected intrusion events are distinguished by a support vector machine with wavelet features to reduce the nuisance alarm rate. Experiments on the PIDS in Chongqing Fuling's shale gas field show that detection algorithms based on the feature of short-time energy and short-time wavelet coefficient energy are much better, and the performance of event classification is satisfactory.