In: Automatica, Vol. 27, No. 2, pages 239-245. March 1991.
Abstract: A new method of fault detection using dynamic measurement signals which is suitable for larger systems is presented. By modelling the system as a Petri net, failures with very slow time constants are detectable. The method is limited to the identification of sensor or process errors which are manifested in signals related to physical conservation quantities. After a fault is detected a prognosis of the future system's behavior can be provided. The method is applied to a nuclear power plant secondary cooling loop. The advantages and drawbacks are discussed in detail.
Keywords: fault detection (using) dynamic measurement signal(s); sensor error; process error; nuclear power plant secondary cooling.