A principal component analysis dominance mechanism based many-objective scheduling optimization

Abstract

Many-objective scheduling optimization problems (MOSOPs) are NP hard problems. It is a great challenge for an algorithm to solve them. In order to solve a many-objective scheduling optimization model aiming at minimizing total production cost, makespan, earliness and tardiness penalty and carbon emissions in manufacturing processes, algorithms with high performances on convergences and diversities are needed. Currently most algorithms might suffer insufficient selection pressures and cause low search efficiency for MOSOPs. As a dominance mechanism based on principal component analysis has shown good performances on reducing dimensionalities of a data set with a large number of interrelated variables and sorting non-dominated individuals, a generally frame of algorithms based on a principal component analysis dominance mechanism has been proposed. Performances of the principal component analysis based algorithm is verified by benchmark problems. Results show that the proposed principal component analysis based algorithm outperforms other algorithms in both convergences and diversities.