System Condition Assessment Based on Mathematical Analysis

Abstract

When determining a system technical condition, it is possible to use multiple approaches. For practical reasons it is convenient to use an indirect diagnostic signal. In our article we focus on applying oil field data collected from a few tens of heavy vehicle engines. The aim is to get a picture of how quickly oil polluting particles are made and consequently how quickly the degradation progresses. This leads to system condition monitoring. When modelling the occurrence of the oil polluting particles, advanced linear regression methods are used. When analysing the diagnostic data, we use mainly a novel quantile regression approach. The aim is to estimate i) the course of trend in the development of polluting particles, ii) critical threshold time hitting iii) distribution of first hitting time of occurrence of soft failure.