Fault Detection Using Nonlinear Low-Dimensional Representation of Sensor Data

Abstract

Sensor data analysis plays a key role in health assessment of critical equipment. Such data are multivariate and exhibit nonlinear relationships. This paper describes how one can exploit nonlinear dimension reduction techniques, such as the t-distributed stochastic neighbor embedding (t-SNE) and kernel principal component analysis (KPCA) for fault detection. We show that using anomaly detection with low dimensional representations provides better interpretability and is conducive to edge processing in IoT applications.