Two-stage prediction model research on condition-based maintenance

State prediction is a critical and difficult problem in condition-based maintenance decision making. Aimed at the typical two-stage failure process of a piece of equipment in maintenance practice, a two-stage state prediction model was designed based on condition monitoring information obtained using the concept of delaying time and stochastic filtering theory. The model overcomes the deficiencies of a one-stage state prediction model. Not only did the model dynamically predict the residual useful life of monitored equipment in the failure delaying stage, but also the two stages of its failure process combined together gave a closer description of the real operating process of the monitored equipment. The model was simulated by Matlab and the results of the simulation proved the effectiveness of the model.