A self-adaptive metaheuristic to minimize the total weighted tardiness for a single-machine scheduling problem with flexible and variable maintenance

Inspired by the chemical vapor deposition process in semiconductor manufacturing, this paper addresses the challenging NP-hard scheduling problem with flexible-variable maintenance in a single-machine environment. The problem also includes sequence-dependent setups and release times. The key aspect lies in maintenance operations, where the start times can vary within specified intervals and durations depending on working conditions. To tackle the problem at hand, a new metaheuristic algorithm, named Improved Self-Adaptive Harmony Search, is proposed. The algorithm introduces a dynamic calibration of parameters and three procedures to enhance the solution quality. The algorithm is compared with state-of-the-art algorithms using a full-factorial design of experiments, with total weighted tardiness as the objective function. The statistical analysis demonstrates the algorithm’s effectiveness, particularly for medium- and large-sized problems. This work not only studies a new variant of the scheduling problem but also provides a robust algorithmic solution, which can offer practical applications in manufacturing systems.