Operating load based real-time rolling grey forecasting for machine health prognosis in dynamic maintenance schedule

Machine health prognosis plays an important role in the dynamic maintenance decision-making. For complex manufacturing systems, it is necessary to schedule a predictive maintenance program and avoid production losses by predicting machine degradations. This paper proposes a novel prognostic method, a real-time rolling grey forecasting method, to provide efficient and accurate machine health prediction, while effects of influencing factors such as operating load are considered and analyzed. In this grey forecasting model, generating coefficient \(W\) values corresponding to variable operating loads are dynamically generated to overcome the shortage of a static \(W\) value. It improves the forecast accuracy of the frequency of failures. A series data about increasing machine health states of failure frequency is used as the in-sample test data. Results of the out-of-sample predictive data show that the application of the proposed method leads to a noticeable increase in forecast accuracy. This indicates the improved rolling grey forecasting model offers a potential to predict the failure frequency trend for supporting the dynamic maintenance schedule.