Researchers: Zheng Wang
Cyber-Physical Systems (CPS) consists of computational elements, communication networks and physical systems. The integration of both cyber systems and physical systems significantly improves the performance and functionality of many critical infrastructures such as manufacturing, power, traffic and health-care networks and smart cities. However, because of the integration with public networks, a fault or an attack on one of these critical systems, may affect a large portion of society with lethal consequences. As such, the safety and reliability of CPS becomes more and more crucial every day. Fault, attack and anomaly detection mechanisms play a vital role in providing such reliability and safety to CPS. Observer-based fault diagnosis mechanism has been studied for decades to address traditional faults, such as sensor failure. However, traditional methods may fail to detect cross-domain faults, such as faults caused by intelligent attacks. In this project, we proposed a novel framework which consists of multiple different types of observers, considering both the continuous and discrete dynamics of CPS, for fault diagnosis. Based this framework, different methods are proposed and validated to extend the types of the faults that can be diagnosed, including faults caused by smart attackers.
Publications:
- Zheng Wang, D. M. Anand, J. Moyne & D. M. Tilbury (2017) Improved sensor fault detection, isolation, and mitigation using multiple observers approach, Systems Science & Control Engineering, 5:1, 70-96, DOI: 10.1080/21642583.2016.1278410
- Wang, Zheng, et al. “Conflict-driven Hybrid Observer-based Anomaly Detection.” ACC 2018, submitted