Application of genetic algorithms to chemical flowshop sequencing

The purpose of this paper is to show that methods of AI, genetic algorithms in particular, are very effective at solving difficult, important real-wor ld problems, specifically the optimization of serial multiproduct batch pla nt sequencing, and to present the results of our work in this field. This w ork deals with the problem of finding a sequence of batches that minimizes the makespan, and discusses the application of different genetic algorithms to find such an optimum sequence. To create such an application, the autho r must select, create, and modify the appropriate algorithm and set the alg orithm's parameters, so that the result is well suited to the specific prob lem type while remaining flexible enough to be applicable to different prob lems in the target group. This paper presents the analysis of performance o f different algorithm configurations and parameter values and, in addition, proposes a new crossover operator that offers an improvement over older on es. The results obtained by using genetic algorithms are compared to those obtained by using MINLP.