Electromagnetic device performance identification using knowledge based neural networks
Frédéric Dandurand, Duncan Lowther · IEEE Transactions on Magnetics · 1999
A knowledge based artificial neural network which represents the design equations of an electromagnetic device is described. The network architecture is based on a rule set developed from a simple algebraic model of the device, The system is then revised by using numerical solutions as training sets to remove the assumptions built in by the algebraic system.