PCA Classification of vibration signals in WSN based oil pipeline monitoring system

Abstract

       Using wireless sensor network technology in structure health monitoring applications results in generating large amount of data. To sift through this data and extract useful information an extensive data analysis should be applied. In this paper, a Wireless Sensor Network (WSNs) is proposed for the oil pipeline monitoring system with proposed method for event detection and classification. The method depends on the Principal Component Analysis (PCA). It applied to features extracted from vibration signals of the monitored pipeline. These vibration signals are collected while applying damage events (knocking and drilling) to the oil pipeline. PCA is applied to features extracted from both time domain and frequency domain. The results manifest that this method is able to detect the existence of damage and also to distinguish between the different levels of harmful events applied to the pipeline.