A failure time series prediction method based on UML model

Due to significant industrial demands toward flight safety and airplane maintenance quality, improving airplane´s reliability in usage stage has become an important activity and the research domain is rapidly evolving. In this paper eighteen years´ field data, gathered from the maintenance phase of a Boeing 737 aircraft, is prepared as time-to-failure series. Then automatic processing models based on unified modeling language (UML) are presented to cope with this data, which incorporate three methods of Holt-Winters, autoregressive integrated moving average (ARIMA), and singular spectrum analysis (SSA). Each method´s modeling and forecasting process is analyzed, as well as SSA´s parameter optimization. Furthermore, a hybrid processing model is built to take advantage of each method. The results are compared and evaluated by root mean square error (RMSE) and show that hybrid methods are more adaptive than single methods, and valid that the proposed processing models are feasible and efficient to deal with the failure time series.