The application of neural network techniques to adaptive autoreclosure in protection equipment

Autoreclosure schemes as applied to EHV systems have, by offering benefits such as maintenance of system stability and synchronism, been the major cause of a substantial improvement in the continuity of supply. It is however, well known that the present practice of automatic reclosure after a fixed dead time can pose problems. In this respect, adaptive autoreclosure techniques, whereby a control logic system ascertains whether (or precisely when) to reclose the circuit breakers, offer a very attractive alternative. This paper is concerned with describing one such technique in which neural networks are employed in designing a circuit breaker control system. It is shown that by using sufficient training examples from accurate simulations of fault situations, it is possible to create a neural network which can recognise certain situations and give a good decision of whether and when to reclose