An integrated framework for reliability prediction and condition-based maintenance policy for a hydropower generation unit using GPHM and SMDP

Condition-based maintenance (CBM) of hydropower generation unit (HPGU) is of great significance for the intelligent operation and maintenance of hydropower station. The accurate reliability assessment of equipment is the foundation of the maintenance. However, there has not yet been a thorough study of an integrated theoretical framework for CBM of HPGU based on reliability assessment. Here, we propose a theoretical framework integrates generalized proportional hazard model and semi-Markov decision process to balance reliability prediction and optimization. Firstly, the method described the impact of covariates, which was determined by the operating conditions, on equipment hazard rates during maintenance events. Secondly, the unit reliability was fitted through maximum likelihood estimation and used as one-step transition probability between the states in SMDP. Then, the cost and maintenance epoch within the framework were optimized by the discount criterion and utilized to finish the feedback of the maintenance threshold. Ultimately, the effectiveness of the method was verified by comparing with the average time cost model through actual cases, the results avoided an operating cost of 87.2 $/d and a maintenance cycle error of 15 days, respectively. Besides, the performance of the model under variable operating conditions was discussed, further enhancing the completeness of the framework.