On-line diagnosis for shorted field-turns of synchronous generator based on artificial neural networks

A technique whereby synchro-generator shorted turns in the field winding is detected on-line using an artificial neural network (ANN) with fuzzified output. It is independent on mathematical models and parameters of the synchro-generator. The diagnostic ANN model, adaptive training algorithm and schematic circuit are proposed. Results of dynamic experiments show that the method is efficient and accurate.