Statistical tools for anomaly detection as a part of predictive maintenance in the mining industry

Abstract

We present new achievements in the area of anomaly detection related to predictive maintenance in the mining industry. The main focus is on the problem of local damage detection based on vibration signals analysis. The vibration signals acquired from machines usually have a complex spectral structure. As the signal of interest (SOI) is weak (especially at an early stage of damage) and covers some frequency range, it must be extracted from raw observations. Up to now, most the techniques assumed the presence of Gaussian noise. However, there are cases in which the non-informative part of the signal (considered as the noise) is non-Gaussian due to random disturbances or to the nature of the process executed by the machine. In such cases, the problem can be formulated as the extraction of the SOI from the non-Gaussian noise. Recently, the importance of this problem has been recognised by several authors, and some new ideas have been developed. We present here a comparison of the new techniques for benchmark signals. Our analysis will cover classical approaches and recently introduced algorithms based on the stochastic analysis of the vibration signals with non-Gaussian distribution.