Application of neural networks for fault diagnosis in a power plant

The paper deals with the diagnosis of an industrial plant about which the knowledge in the fault situations is incomplete. The approach is as follows: the intelligent system, whose main part consists of a number of artificial neural networks, makes use of the qualitative knowledge about the process flow in the fault-free case and in the presence of faults, and it concludes about the possible faults in the system examined. The experience in using the feedforward neural network is described.