Drilling head knives degradation modelling based on stochastic diffusion processes backed up by state space models

Abstract

System quality requirements are typically formed by consideration of reliability and safety performance. Failures caused by system weakness, degradation or fatigue may cause undesired, and potentially dangerous, consequences. For various reasons, not all processes of system degradation are easily monitored in the lifecycle of a system. Degradation evolution leads to changes in both performance and reliability characteristics.In this article, we investigate a mining system consisting of dataset records on in-field operational characteristics of a drilling head. We work with these data in order to get a picture of system degradation and actual condition. For data assessment and modelling, we apply both improved and specific new mathematical models. We examine the data using extended and enhanced state space models, which are suitable for system state and condition investigation. Our time series approaches are based on a modified Kalman-type backpropagation recursion. The improved and modified state space models are accompanied by improved forms of selected stochastic diffusion processes. The diffusion processes are used both for degradation modelling and also for forecasting potential failure occurrence. All of these models are expected to help both with deterioration propagation assessment and with the indication of when the degradation of the system under investigation is predicted to reach the critical limit. Such a limit is represented by threshold performance characteristics that may lead to either soft or hard failure with related faults. The outcomes presented in this article may help with i) failure occurrence prediction, ii) residual useful life prognosis, iii) safer system operation, iv) system utilisation rationalisation and v) maintenance forecasting.