A Multi-Objective Genetic Algorithm to Solve a Real Life Travelling Salesman Problem

摘要

This research aims at solving a Real-Life optimization problem. In fact, the Electricity Company in Bahrain receives many daily requests to perform maintenance or survey tasks at different locations. A team of technicians and workers has to execute these tasks in a given order and return back to their initial location. This problem can be modeled as a symmetric Travelling Salesman Problem (TSP). This research provides the description of Electricity Company's Travelling Salesman proposed scenario including objectives and constraints. A method based on genetic algorithms was proposed and successfully tested using number of Bahrain's locations to solve the mentioned problem. A new crossover technique called “Complete Min sub-tour Exchange Crossover” was introduced. Two objectives were included one minimizing the total travelled distance and the second minimizing the total delay in completing all the tasks. We used the Pareto Optimal technique to compare the solutions generated by the genetic algorithm.