The major achievements on the research of fault detection in control systems are reviewed in detail from three aspects which including signal threshold based approaches, signal model based approaches and process model based approaches. Particular emphasis is put on the process model based approaches which is using the closed-loop monitoring information in control systems to establish the quantitative and qualitative process model, detecting and then isolating the main failures in sensors, actuators, and the controlled process. The corresponding application problems are stated within a purpose of acquiring small probability of both the false alarm rate and missed alarm rate. As a result of the growing demand for higher performance, efficiency, reliability, sensitivity and rapidity in fault diagnosis systems , the robust fault detection in the transition process, the knowledge acquisition for quantitative and qualitative diagnosis based on processing history data, and hybrid intelligent fault diagnosis system architecture are worthy of a deeper research. http://www.sj-ce.org/paperInfo.aspx?ID=1494
