数据挖掘技术在控制器局域网总线中的应用研究

Application of Data mining in Controller Area Network System

提出了数据挖掘技术在汽车控制器局域网(CAN)总线通信数据中的应用。汽车中的各种电子控制系统通过CAN总线相互连接通信,其中产生的大量报文被存储在日志文件中。首先对CAN总线日志文件的中大量通信数据做预处理,再扩展使用关联分析、分类及聚类分析等方法对处理过的数据进行挖掘,从而找到各条报文之间及报文内部各位或字节之间可能存在的新规律或者关系,为进一步研究CAN总线系统连接的各个控制系统之间的通信提供了帮助。

This report presents application of data mining in the controller area networks（CAN） bus of real automotive systems. A set of analysis tools are applied in the experiment. The association rules are utilized to describe the relationship between different message IDs, classification method to find important bits or bytes from content of each message and clustering method to check whether these bits or bytes are important or not. Because of the lack of domain experts in the CAN bus field, the results do not refer to semantic meaning about message IDs and messages from different devices, but they only focus on interesting relationship or rules between different messages.