Ant Colony Algorithm Approach Toward Decision-Support in Maintenance Scheduling of Oil Tanks

• Main Authors: Jing-Mao Lin, 林敬貿
• Other Authors: Sheng-Tun Li
• Format: Others
• Language: zh-TW
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/37442441220524092961

碩士 === 國立成功大學 === 資訊管理研究所 === 93 === 　The oil market in Taiwan is becoming very competitive today than before, which is due to her recent entry into WTO and liberalized Petroleum Management Law, and the utilization ratio of oil tanks has already become the important topic in administrative decision. An optimized oil tank maintenance schedule not only increases system-operating reliability, but also reduces operation cost, and increases enterprise's profit and competitiveness. Maintenance problem is a classical NP-Hard problem, and heuristic algorithms has already succeeded in applying to solve various kinds of NP-Hard problem, which including the Traveling Salesman's Problem, limited resources allocation problem, and so on. Recently, Ant Colony Optimization (ACO) becomes another promising methodology in solving NP-Hard problem, which mimics the ethology behavior of collecting food quickly by the cooperative blind ants in different colonies seeking to the shortest path. Therefore, this research adopts the ACO approach to solve the maintenance scheduling problem of oil tanks, and compares the efficiency and effectiveness with Genetic Algorithm. In addition to finding out the best solution, we also investigate the knowledge behind the solutions further, and help businesses to archive the goal of Knowledge Management (KM). Therefore, we collect the solutions passing the threshold values into one set, and use ACO-based model of knowledge discovery in databases to uncover the interesting hidden patterns. Some classification rules are built up, and the domain expert will examine and refine these rules through his plentiful practical experience. The knowledge rules discovered can be further applied to the development of an intelligent decision support system for helping decision makers refine their decision knowledge in oil-tank maintenance scheduling.