A genetic algorithm for the mixed flow shop problem

Abstract

In this thesis we present a new interesting version of the mixed flow shop se-quencing problem, which at the same time is a version of the classic flow shop,a very common topic on operations research.We propose a genetic algorithm to solve it that we will compare at the endwith a simple initial genetic-based algorithm previously design. For that wefirst focus on the crossover operator as we consider it the most challenging parton a sequencing problem. We study and compare 5 different crossover operatorsand we choose the one that performs better. Finally we calibrate the populationsize, the weight of mutation and crossover operators on the algorithm and alsothe mutations operator itself.The goal of the thesis is to better understand the specific mixed flow shopproblem version presented and design a genetic algorithm that clearly improvesthe performance of the initial algorithm