Estimation of distribution evolution memetic algorithm for the unrelated parallel-machine green scheduling problem

With the increasing concern on greenhouse gas emissions, green scheduling decision in the manufacturing factory is gaining more and more attention. This paper addresses the unrelated parallel machine green scheduling problem (UPMGSP) with criteria of minimizing the makespan and the total carbon emission. To solve the problem, the estimation of distribution evolution memetic algorithm (EDEMA) is proposed. Firstly, based on the minimum machine load first principle, the initialization of the population is proposed. Second, a multi-objective non-dominated sorting approach and the crowding distance are adopted to improve the diversity of individual. Third, to estimate the probability distribution of the solution space, a probability model is presented to enhance the searching ability. Third, five neighbourhood searching operators are designed to handle the job-to-machine assignment. Moreover, the population catastrophe is used to maintain the sustainable diversity of the population. Finally, based on the randomly generated instances of the UPMGSP, extensive computational tests are carried out. The obtained computational results show that the EDEMA has the better searching capability and the better objective value than those of the non-dominated sorting genetic algorithm II and the estimation of distribution evolution algorithm (EDEA) in solving the UPMGSP.