Design and Development of an Integrated Optimization Platform (for Production Planning and Vehicle Routing The Developed Software Offers a Unified Graphical Interface (GUI)).

Abstract

This thesis focuses on the design and development of an integrated optimization platform capable of tackling two core combinatorial problems in operations research namely the Flexible Job-Shop Scheduling Problem (FJSP) and the Capacitated Vehicle Routing Problem (CVRP). The implementation is developed as a combined GUI in Python with Tkinter enabling users to define problem instances, setup metaheuristic parameters, run algorithm executions and display results in Gantt charts and log-reports of the routes generated. A sophisticated Enhanced Hybrid Genetic Algorithm that includes dynamic mutation, multi neighborhood Tabu search and a diversity maintenance mechanism is developed to solve the FJSP and obtains a makespan of 41 time slots (20.5 h) for a case-study of woven-labels from a factory having 16 jobs and 5 machines. The approach adapted to solve the VRP involves a Simulated Annealing solver incorporating the aspect of heterogeneous fleet, customers partitioning and multi-trip trips, providing reduction of over 23% in the total cost of transportation when compared to the greedy solution obtained for the problem. In this thesis the theoretical base of the methods is explained, the software structure(UML, Merise), algorithms logic, and relative experimental results highlighting the real life applications of metaheuristics to scheduling problems.

Description

Citation