Researches on Flexible Job-Shop Scheduling Problem

Evolutionary algorithms (EAs) prove to be powerful in solving combinatorial optimization problems. A symbiotic evolutionary algorithm is applied to deal with complex job-shop scheduling problem (JSP). An efficient crossover, Merge and split recombination crossover (MSX) which always produces feasible offspring and enhances population diversity and search efficiency, is introduced for the JSP. Evaluation of individual´s contribution in a combination is well important and difficult in symbiotic evolutionary algorithm. A new individual fitness function is defined according to the arithmetic average over previous solution combinations that the individual has taken part in. The improved symbiotic algorithm is tested on a set of standard instances taken from the literature and compared with original approach. Experimental results validate the effectiveness of the improved algorithm, improving the solution quality and decreasing the computation time.