Memetic algorithm used in a flow shop scheduling problem

Abstract

Scheduling activities in flow shops involves generating a sequence in which the jobs must be processed. To generate the sequence, some criteria are taken into account, such as the completion time of all the jobs, delay time in delivery, idle time, cost of processing the jobs, work in process, among others. In this case, completion time of all jobs and idle time are taken as the objective function. To generate the sequence, a Memetic Algorithm (MA) is used that combines Simulated Annealing (SA) and Genetic Algorithms (GA) to solve the problem. A permutation type decoding was used for the vectors that make up the MA population. The SA was used for the generation of the initial population. Selection, recombination and mutation processes are generated in a similar way to GA. In this case there are 6 parameters to be set; temperature, z parameter, recombination probability, mutation probability, cycles and initial population. To set these parameters, the Response Surface Methodology is used for two objectives. Achieving improvements in the algorithm result of at least 2%. These results help to minimize processing times which impacts with the economics of the enterprise. Using the MA in an interface that helps the user to make a decisión about the Schedule of the Jobs.