Neural network and its ancillary techniques as applied to power systems

The layered perceptron neural net is receiving the most attention as a viable candidate for application to power systems. The layered perceptron is taught by example. Before neural networks can gain the necessary recognition as useful problem solving tools in the power industry, certain fundamental issues need to be addressed. Some of them are associated with neural network technology, and others are problem dependent. The author discusses the following issues: learning versus memorisation; best net size determination, network saturation, feature extraction, neural net inversion, genetic based neural nets, and fuzzified neural nets.