Multi-sensor monitoring information based decision support method for optimal predictive maintenance policy

For a class of multi-sensor dynamic systems subject to the latent degradation, a decision support method for condition-based optimal predictive maintenance is proposed in this paper. First, by adopting the distributed filtering and expectation-maximization estimation algorithm, the remaining useful life (RUL) is on-line predicted in accordance with the identified hidden degradation process. Then, a predictive maintenance policy is introduced based on the prediction results. Furthermore, the optimal predictive maintenance time is given by minimizing the maintenance cost. Our main results are verified by a practical case study of the milling machine experiment.