Multi-Product Kanban System Based on Modified Genetic Algorithm
Kanban system plays an important role in many manufacturing systems. The design of a Kanban system addresses the selection of two important parameters, i.e., the number of Kanbans and the lot size. This problem has been tackled in a number of studies usin
- PDF / 1,678,240 Bytes
- 8 Pages / 439.37 x 666.142 pts Page_size
- 67 Downloads / 228 Views
Multi-Product Kanban System Based on Modified Genetic Algorithm Liang Huang
Abstract Kanban system plays an important role in many manufacturing systems. The design of a Kanban system addresses the selection of two important parameters, i.e., the number of Kanbans and the lot size. This problem has been tackled in a number of studies using simulation models. But in the absence of an efficient gradient analysis method of the objective function, it is time-consuming in solving large-scale problems using a simulation model coupled with a meta-heuristic algorithm. In this chapter, a gradient-based heuristic is applied to a genetic algorithm for the design of a multi-product Kanban system. Several case studies in different sizes have been tried out and solutions from the modified genetic algorithm were compared to those from the classical genetic algorithm. Notable improvements in computing times or solutions by the modified genetic algorithm can be observed. Keywords Kanban system heuristic
Simulation Genetic algorithm Gradient-based
100.1 Introduction Kanban system has been widely used in many manufacturing systems. In a Kanban system, the main design parameters are the number of Kanbans and the lot size [1]. A variety of methods by means of analytical or simulation modeling have been used to tackle the problem.
L. Huang (&) Northeastern University at Qinhuangdao, Qinhuangdao 066004, People’s Republic of China e-mail: [email protected]
Z. Zhong (ed.), Proceedings of the International Conference on Information Engineering and Applications (IEA) 2012, Lecture Notes in Electrical Engineering 219, DOI: 10.1007/978-1-4471-4853-1_100, Springer-Verlag London 2013
817
818
L. Huang
Stochastic process models have been used to model many Kanban systems. Yoichi et al. [2] have designed a single-stage Kanban system based on a queuing model. The objective of this work is to determine the number of Kanbans, when a change of load to the system is planned. Nori and Sarker [3] have modeled a Kanban system using Markov Chains to determine the optimum number of Kanbans between adjacent work stations. The methods based on queuing models or Markov Chains can effectively simplify the description of Kanban systems, but the assumptions of the queuing rules, such as first come first service (FCFS), limit the scope of application of these papers in practice. For finding the required number of Kanbans and lot sizes in a complex Kanban system such as a multi-product Kanban system, simulation models can offer a number of advantages. Berkley [4] has simulated a two-card Kanban system with multiple part types to determine the effect of container size on average inventory and customer service levels. Shahabudeen et al. [5] have set the number of Kanbans as well as lot size at each work station using genetic algorithm (GA). In another work of them [6], they have set similar parameters using simulated annealing algorithm (SAA). GA or SAA start with an initial solution and tried to reach the optimum solution using neighborhood
Data Loading...
