Fault detection in fossil electric power plant via neural networks

In this paper, the authors discuss fault detection for fossil electric power plants. It is intended to use recurrent neural networks to perform such a detection. A sceme for fault detection is proposed. This scheme is based on comparison bctween thc measures coming from the plant and thc predicted values generatcd by a neural network model. This research is in its beginning: right now a full scale simulator is being uscd to generate the required data base.