A Genetic Algorithms Approach to Determining the Parameters of Repairable Series and Parallel Systems

• Main Authors: Tzu-Lin Liu, 劉子麟
• Other Authors: Shui-Shun Lin
• Format: Others
• Language: zh-TW
• Published: 2003
• Online Access: http://ndltd.ncl.edu.tw/handle/55644137198464688308

碩士 === 東海大學 === 工業工程學系 === 91 === As the system structure becomes more and more complicated, failure of system function will always result in huge damage. It goes along with increasing cost while trying to improve system reliability or availability. Therefore, how to make an adequate decision to fit the real functional requirement is of importance. The main purpose of this study is to conduct a decision support method for repairable system designers. This method attempts to substitute the traditional experience-based means. Two important parameters of system parts are to be determined in the phase of parameter design for a series and parallel system: Mean Time Between Failure (MTBF) and Mean Time To Repair (MTTR). This study proposes a two-step, genetic algorithms based, method to determine the parameters. The step one is to list the approximate expressions of the system availability and summarize the whole cost. The step two is to take availability/TC as the objective function and thereafter use Genetic Algorithms to solve the optimization problem. The results show that using this method is capable of choosing a proper system parameters and producing reasonable repair policies.