An Associative-Classification Based Motorcycle Injection Engine Diagnosis Support System

• Main Authors: Kuo-ping Tsai, 蔡國彬
• Other Authors: Wen-Yang Lin
• Format: Others
• Language: zh-TW
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/66011978185175518976

碩士 === 國立高雄大學 === 資訊工程學系碩士班 === 99 === 　In the recent years, due to the discontinuation of the old type scooter of the fourth Environmental regulations, scooter manufacturers have to launch a new type of scooter　using the injection fuel supply system similar to the automobile. Comparing with the traditional engines, the diagnosis of injection engines are too difficult to only count on the experiences of mechanics. A diagnosis support system is necessary. The current procedure for repairing a scooter uses a diagnostic device to read the defect code diagnosed by ECU. However, due to the complication of the ECU data, engineers have to spend a lot of time adjusting the ECU's parameters based on their test experience. There is no complete set of systems to help the engineer make a prompt and correct diagnosis. In this thesis, we take some motorcycle manufacture in Taiwan as a case study, aiming to develop a motorcycle injection engine diagnosis support system. Our system is based on a data warehouse collecting data derived from the ECU information and the defect analyses by the manufacturer, and provides OLAP and our developed data mining tool. Our mining tool adopts our proposed R-CMAR algorithm, an associative-classification technique, which can effectively reduce the amount of stored data. In addition, we use the concept of prestore to store the candidate rules in DB in order to combine the index and inquiry processing function of the data management system to speed up the response time of the system. According to the experiments, our proposed algorithm is superior to other algorithms, through which our system can provide an on-line and interactive diagnosis environment.