A novel grey prognostic model based on Markov process and grey incidence analysis for energy conversion equipment degradation

Maintenance strategy for energy conversion equipment degradation is now experiencing the transformation from fail-and-fix to predict-and-prevent due to the equipment complexity and the strict requirements for equipment reliability. Actually, the current situation of world class maintenance is providing never-before-seen opportunities and challenges for the maintenance specialists. For this problem, the essence is to optimize present PM (preventive maintenance) strategies, so as to avoid some common maintenance problems, such as insufficient proactive maintenance, frequent problem repetition, and unnecessary and conservative PM. Besides, accurate prognostic methodology is the core section of this optimization. Considering the data uncertainty and the requirements for long-term forecast, grey model serves as an attractive and effective prognostic model for equipment degradation prognosis. To compensate the limitation of traditional grey model resulting in the unfitness of fluctuant data, the Markov model is introduced into traditional grey model. In order to expand the dimension of the original data, the grey incidence model is adopted, so as to further employ the additional time series data similar to the target series. Then, the scheme of the novel grey prognostic model, based on the Markov process and the grey incidence analysis, is proposed. Finally, the fouling process of a gas turbine compressor is chosen as an instance to validate this novel model. In addition, the study has been conducted on the relationship between model parameters and the prognostic accuracy, and the best parameters for this case are suggested. Comparative study results of different prognostic models show that considering the prognostic accuracy and fluctuations, this novel model is better than some other prognostic models.