Lifetime Prediction of a Hydraulic Pump Using ARIMA Model

In the present work, the Auto-Regressive Integrated Moving-Average (ARIMA) forecasting method is used to predict the remaining useful life (RUL) of an axial piston pump used in the hydraulic system applied to the sheet metal industry. The leakage volume from the piston pump is considered a significant parameter for the life prediction analysis using the ARIMA model, as the proposed model uses auto-regressive (AR) and moving-average (MA) modelling for forecasting with high prediction accuracy. The life estimation through the ARIMA model is based on the degradation method of remaining useful life prediction. The life prediction results were in an acceptable range with a forecasting root-mean-square error (RMSE) of 0.3. The investigation results concluded that out of the two forecasting models, i.e., ARIMA (1,1,2) and ARIMA (20,1,2), ARIMA (1,1,2) model has the least forecasting error and also shows better adjustment with the historical data. The RUL of the hydraulic pump considered in the present investigation is predicted to be 28 months. Additionally, the model developed can be used to determine the lifetime of the operational pump and plan its maintenance schedule for hydraulic-operated unmanned vehicles used for off-road applications like mining, agriculture, defence, etc. The proposed RUL forecasting technique can help avoid mechanical systems from uncertain breakdowns and will also play a pivotal role in maintenance.