Joint optimization of opportunistic maintenance and speed control for continuous process manufacturing systems considering stochastic imperfect maintenance

For continuous process manufacturing systems (CPMSs) where production cannot be stopped, opportunistic maintenance and speed control are the main means to improve production completion probability. However, existing studies on the joint optimization of these two means ignored the stochastic characteristic of imperfect maintenance, which increases the risk of unplanned downtime for CPMSs. Therefore, a novel joint optimization method of opportunistic maintenance and speed control under stochastic imperfect maintenance is proposed. The opportunity time window (OTW) is introduced to characterize the production-constrained maintenance "opportunities" between two production batches. Based on the states and the number of imperfect maintenance times of machines at the end of the last production batch, the proposed method requires determining the maintenance schedule during the OTW with limited maintenance resources and the speed level for the next batch. Moreover, a stochastic flow manufacturing network is established to evaluate the weighted production completion probability under stochastic imperfect maintenance and production speed with different weights. The joint optimization problem to maximize the weighted sum of production completion probability over a finite operation horizon is formulated as a Markov decision process (MDP). Then, a tailored deep learning and knowledge based approximate dynamic programming algorithm, which incorporates the structural property of MDP, is developed to solve this optimization problem. Finally, a case study of the hot rolling manufacturing system is conducted to validate that the proposed method can improve production efficiency and reduce the negative impact of stochastic imperfect maintenance on stability.