Fast hybrid integral equation-neural network method for the modeling of multiconductor transmission lines

Multiconductor transmission lines (MTL) have been modeled by distributed parameters R, L C and G in many commercial CAD packages, where most of the parameters are frequency independent or at most R/spl prop//spl radic/f. We present a new and fast technique based on neural network and integral equation methods (IEM) to accurately evaluate this frequency dependence, while dramatically reducing the computation time.