Search

Browse

• Communities
• Authors
• Year
• Titles
• Subjects

My Space

• Login

Solving multiobjective vehicle routing problems and their variants using evolutionary algorithms

Psychas Iraklis-Dimitrios

Full record

---

Summary

In this Ph.D. thesis four new multiobjective energy vehicle routing problems were presented and solved. The first objective function of all problems has always as goal the minimization of the time needed for moving between the customers or the customers and the depot. The main innovation of these problems is in the second objective function (which concerns the calculation of the fuel consumption) where besides the distance and the weight of the cargo, some additional real life parameters of the routes are taken into account (slope of the road, direction and wind power as well as rpms of the engine of the vehicle). For the first two problems, the Multiobjective Symmetric Delivery Route based Fuel Consumption Vehicle Routing Problem and the Multiobjective Asymmetric Delivery Route based Fuel Consumption Vehicle Routing Problem, the second objective function is used for the minimization of the fuel consumption in a delivery problem. In the first problem, we assume there are no route parameters while in the second problem we assume that the conditions are not perfect thus, an asymmetric problem is produced. For the last two problems, the Multiobjective Symmetric Pick-up Route based Fuel Consumption Vehicle Routing Problem and the Multiobjective Asymmetric Pick-up Route based Fuel Consumption Vehicle Routing Problem, the second objective function is used for the minimization of the fuel consumption in a pick-up problem. In the first problem, we assume there are no route parameters while in the second problem we assume that the conditions are not perfect thus, an asymmetric problem is produced. The proposed problems were solved using Evolutionary Algorithms (Genetic Algorithms, Nature Inspired Algorithms and Artificial Immune Systems). The common characteristics of all algorithms that are the main innovative features of this Ph.D. thesis are referred to the creation of the initial population, to the local search method and to the parallel multi-start method. More innovations that are presented in the proposed algorithms are the addition of more steps and the modification of the main functions of the algorithms so that their efficiency to be improved. In order to estimate the efficiency of the algorithms four different evaluation measures are used. In the last chapter of this Ph.D. thesis a new method based on the utility theory was presented in order to classify the solutions of the Pareto front.

Available Files

Services