Research of Marine Engine Failure Prediction Based on Random Forest Algorithm

• Main Authors: LIN, HO-MIN, 林鶴閔
• Other Authors: GUO, JUN-LIANG
• Format: Others
• Language: zh-TW
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/handle/mzr5qb

碩士 === 國防大學 === 運籌管理學系 === 107 === Due to the development of the shipping industry and the increased competition in the international shipping market, ship maintenance and repair is a topic that recently has been paid more attention, especially, the maintenance and repair of marine engines. With the development of technology, it is possible to create an optimized maintenance strategy, which is the predictive maintenance. This maintenance strategy is based on the condition of monitoring an equipment operation, to determine whether maintenance is required or not. This forecast can play a key role in improving the quality of maintenance and reduction of costs. With the development of big data, the random forest classification algorithm is used for prediction and classification in equipment failure prediction. Through this mode, equipment failure status can be detected, which can save downtime caused by equipment failure and failure. The validity and reliability of the random forest failure detection method are proved. The method obtains 99% correct rate. It can be known that the monitoring condition of the equipment used in the random forest algorithm can predict the failure condition of the equipment and Perform maintenance before the fault occurs.