1. UKM Learning and Research Repository
2. Master Thesis / Tesis Master
3. Faculty of Science and Technology / Fakulti Sains dan Teknologi

Title:	Tuning genetic algorithm with fuzzy logic controller in solving fixed-charge transportation problem
Authors:	Mohammadreza Rostamian Delavar (P50195)
Supervisor:	Zalinda Othman, Dr.
Keywords:	Genetic algorithm Fuzzy logic controller Transportation problem Transportation problems (Programming)
Issue Date:	10-Feb-2012
Description:	Since the competitiveness is dreadfully increasing in today's globalized market, supply chain design has been gaining importance and attention. Firms have to at least keep the same customer service level, while the market's competitiveness forces them to reduce their overall costs to maintain their profit margins. A supply chain is a network of facilities and distribution centers that is responsible for procuring the materials, making intermediate and finished products out of those materials, and distributing the products to customers, and supply chain management is the strategy of integrating these functions together in order to facilitate the synchronization of all parts of this chain. A supply chain network consists of five major chains: R&D chain, purchasing chain, production chain, quality chain and logistic chain. Among logistics activities, transportation network design provides a remarkable potential to reduce the overall costs and also to improve the service level. The transportation problem is a well-known basic network problem. In order to make the transportation problem more comprehensive and also practical, many research papers assume that a fixed cost is also incurred along with variable cost of each commodity. This problem is called Fixed Charge Transportation Problem (FCTP). In a FCTP, a single commodity is shipped from origin (source, supply) locations to destination (sink, demand) locations. The objective is to find the combination of routes that minimizes the total variable and fixed costs while satisfying the supply and demand requirements of each origin and destination. Since the problem is NP-hard, the computational time to obtain exact solution increases in a polynomial manner and very quickly becomes extremely difficult and quite time-consuming as the size of the problem increases. Thus the exact solutions are practically inefficient and the previously proposed heuristics also did not demonstrate an outstanding reduction in computational cost and time. Evolutionary algorithms (AI) such as genetic algorithm (GA) are shown to be better solutions with remarkably less time and cost in these kinds of problems. Some researches solved the FCTP with GA, but none of them tune the critical parameters of GA with a dynamic tuning method. Distinguishing the gap, this research utilized two Fuzzy Logic Controllers (FLCs) to automatically tune the crossover probability and mutation probability as two critical parameters in GA's performance. Also Taguchi parameter design technique is designed to specify the best crossover operator, the best mutation operator and the best population size among all addressed ones. Three algorithms (GA, GA-FLC-1, and GA-FLC-2) are applied and invesigated through two data sets (small-size data sets, and large-size data sets). Results demonstrated that two algorithms which are supported with FLC represent much better performance in comparison with simple conventional GA. It can be concluded that the GA performance is more effective and more efficient by tuning the parameters.,Master/Sarjana
Pages:	120
Call Number:	QA402.6.D438 2012 3
Publisher:	UKM, Bangi
Appears in Collections:	Faculty of Science and Technology / Fakulti Sains dan Teknologi

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.