Abstrait

Research on remote vehicle intelligent diagnosis based on KNN

Miu Kehua, Li xiaokun

This paper provides a remote vehicle diagnosis system, which is designed to locate the specific time when an occasional malfunction happened from the abundant vehicle’s ECU data flow. The system has been designed with an ability to learn by itself, using the wrong cases to retrain the classifier and raise system diagnosis rate. Through studying the occasional low-speed flameout, we come to a conclusion that 83.3% diagnosis rate and nanosecond-class diagnosis efficiency can totally meet requirement