Abstract: Fault Detection and Isolation (FDI) is an important aspect in designing any electrical power system, particularly those with high power due to the high cost of failure. This paper presents an approach to detect and isolate open circuit faults in Zero Voltage Switching (ZVS) full bridge isolated Buck converters, used in Systems of Multiple Sources of Energy (SMSEs). The estimation of the state variables obtained with an observer are compared to measured state variables in order to generate residuals. The generated residuals are able to detect the faults but unable to isolated them. Consequently, a Bayesian Belief Network (BBN) learned from those residuals is designed in order to isolate the occurring fault. The proposed technique is able to detect the studied faults regardless of disturbance, and isolate the fault type with 99.7 % accuracy.

Keywords: Fault detection and isolation, DC/DC converter, Observer, Bayesian Belief Network.