A preference-inspired multi-objective soft scheduling algorithm for the practical steelmaking-continuous casting production

Uncertainty is the most challenging problem for implementing scheduling algorithms under practical environments, since the schedule released into a shop floor with optimal objectives often deteriorates or even suffers infeasibility during its execution period. This paper focuses on the uncertain scheduling problem arising from the steelmaking-continuous casting (SCC) manufacturing system and proposes a multi-objective soft scheduling (MOSS) to overcome this challenge. In this study, a soft-form schedule including critical decisions and characteristic indicators is introduced to provide more flexibility against random disturbances. In the MOSS algorithm, we propose a preference-inspired chemical reaction optimization (PICRO) algorithm to solve the uncertain SCC scheduling problem with soft-form solutions, in which the objectives of waiting time, cast-break and over-waiting are tackled by the preference-inspired method. In the PICRO, a simulation-based T-test method is used to evaluate solutions, and a knowledge-based local search (KLS) is embedded to enhance the convergence of PICRO. Following this, a clean-up procedure is proposed for ranking and selecting the best solutions in the final population output by the PICRO. Computational experiments on the synthetic and real-world SCC scheduling instances demonstrate that the proposed MOSS algorithm can result in significantly better solutions compared to other algorithms under practical environments.