Nonstationary wear trend analysis method in on-line ferrograph monitoring system

Based on 80 groups of experiment data and with the severity index of wear chosen as research object, the nonstationary wear course is analyzed. An auto regressive integrated moving average (ARIMA) model is adopted to analyze the nonstationary series, which can be turned into stationary series by using difference operator and standardization. ARIMA (4,1,1) model is deduced to fit the nonstationary wear course. Hodrick-Prescott filter analysis and experiment data show that the prediction precision can be improved by 9.8% than before, and a lower computing cost can be achieved. Therefore it is suitable to take this algorithm as an embedded software for the on-site application of on-line ferrograph.