A Hybrid Memetic Algorithm for the Parallel Machine Scheduling Problem With Job Deteriorating Effects

This paper presents a hybrid memetic algorithm (HMA) for makespan minimization of parallel machine schedules with job deteriorating effects. The proposed HMA integrates two distinguished features - a novel compound neighborhood structure that extends the basic swap and insertion operators, as well as a divide-and-conquer based local search procedure. Unlike traditional local search methods, the divide-and-conquer based local search divides the original problem into subproblems so as to speed up the exploration of local optima. Experimental comparisons with the current state-of-the-art approaches show highly competitive performance of HMA in terms of both solution quality and computational efficiency. In particular, HMA obtains optimal solutions for all of the 900 small benchmark instances with an average computing time of 0.013 seconds and a success rate of 100%. Besides, it improves on the previous best-known results for 86 out of the 900 large benchmark instances, while matching the best-known results for the remaining cases. Additionally, we provide an analysis of the key algorithm features to identify its critical success factors.