An Integrated MOGA Approach to Determine the Optimal Kanban Number for a JIT System

• Main Authors: Wei-Chung Hu, 扈偉中
• Other Authors: Tung-Hsu Hou
• Format: Others
• Language: zh-TW
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/15932846031932483888

碩士 === 雲林科技大學 === 工業工程與管理研究所碩士班 === 96 === Two of the major pillars of TPS are “autonomation” and “just-in-time (JIT)”. Kanbans are information tools in a JIT system to control the production of the necessary products in the necessary quantities at the necessary time. The number of Kanbans represents the inventory quantity of work-in-process (WIP) or purchasing parts. It is an important issue to determine the optimal Kanban number. The shop floor engineer usually uses empirical rules or simple equations to determine the Kanban number. However, these rules or equations only take the WIP as the only objective into consideration. Some researches have taken multiple objectives into consideration to determine the Kanban number. However, these researches usually combine multiple objectives into one objective by using a transfer function or giving different weights for different objectives. These researches could not generate Pareto-optimal Kanban number for managers to make decisions. In this research, an integrated multiple objective genetic algorithm (MOGA) will be developed to determine the Pareto-optimal Kanban number, and be applied in a TPS-oriented manufacturing company to demonstrate its feasibility. First of all, a simulation model will be created to simulate the multi-stage JIT production system of the company. After an experiment layout of different Kanban number for different production stages being applied to test the production performances, regression models will be built based on the experiment design and simulation results. The regression models will be used to represent the relationships between the Kanban numbers of different production stages and the production performance, and will be used in genetic algorithms to generate the performance for chromosomes. Finally, the MOGA, by using the generalized Parato-based scale independent fitness function (GPSIFF) as the fitness function to evaluate the multiple objectives for chromosomes, will be used to find the Pareto-optimal Kanban number for multiple objectives, i.e., maximizing mean throughput rate and minimizing mean total WIP inventory. Four Kanban number combinations of the Pareto-optimal front have better production performances at both objectives than current Kanban number. The results of the proposed integrated approach show at least 6.9% improvement comparing with current mean throughput rate, and at least 5.5% improvement comparing with current mean total WIP inventory. However, there are no solutions better then each other when the both production performances are considered at the same time, which makes these Kanban numbers satisfy Pareto-optimal solutions.