A CASE-BASED APPROACH TO ALTERNATOR MODELLING

A case-based intelligent system for predicting alternator dynamics is described, in which the appropriate knowledge is accumulated from off-hne simulation using an accurate digital model. The approach is detailed in terms of the dynamic performance of a one-machine infinite-bus system following a three-phase short-circuit. The predictions are compared with those generated by direct use of the benchmark model. The level of accuracy indicated by this comparison and the speed of solution obtained both suggest that the case-based method could be suitable for the on-line study of transient stability when used in conjunction with a fault-identification system. The cpu time required is of the order of one quarter of the real time in the predicted response.