Applying regression diagnostics for identifying non-standard behaviour of a technical system

System condition is an important characteristic in the stage of operation and maintenance during a life cycle. The system condition in respective time periods usually correlates to system time deterioration. Since the degradation may lead to both soft and hard failure, reliability characteristics might be needed to describe each type of such failure. We concentrate on selected oil characteristics which are strongly correlated to both operating and calendar time elapse. In this paper we focus on soot particles and we model mathematically the effect of independent operating values [Mh] and [day] in relation to soot concentration [%]. Using the recorded data we are trying to find the records which could help us to change the oil concentration of soot particles significantly.