Data driven root cause analysis for intermittent connection faults in controller area networks

In networked automation systems, network plays a central role in information exchange. The error in local system will directly or indirectly affect other parts of the automation system, which may induce system or network faults. In this paper, a novel data driven root cause analysis method for the intermittent connection faults, a common but difficult troubleshooting problem in controller area network (CAN), is developed for networked automation systems. Using discrete random process theory and quasi-period pattern recognition method, a root cause reasoning algorithm that correlating high level system events with low level CAN intermittent faults has been developed. Experiments had been conducted based on laboratory test-bed for proof of concept, and the preliminary experiment results show that the proposed algorithm is able to find the root causes of the network problems.