Pareto analysis was used to investigate the failures database of a Hitachi EX1900 hydraulic shovel over one year of continuous operation. The hydraulic system and hydraulic links, hoses and piping were identified as the most critical system and subsystem, respectively. A three-parameter gamma distribution function was identified as the best fit for a probabilistic modeling of the time between failures (TBFs). This distribution was used in a Monte Carlo reliability analysis of the hydraulic system. Autoregressive integrated moving average (ARIMA), within the context of a time series analysis of the TBFs, was proposed as a viable alternative to the traditional technique of distribution fitting. The mean absolute percentage error (MAPE) metric was used to compare the two models. The MAPE value that resulted from the ARIMA was 66 percent smaller than that from the traditional technique in predicting the TBFs.
展开▼
