Multiple adaptive neuro-fuzzy inference systems for accurate microwave CAD applications

An approach for applying fuzzy logic for accurate CAD of microwave circuits is presented. Our proposed method combines space-mapping (SM) technology and multiple adaptive neuro-fuzzy inference systems (MANFIS) for the modeling of microwave devices. MANFIS is trained to predict a nonlinear vector multidimensional mapping function, which is obtained from SM approach. Optimization by micro-genetic algorithm is used to find nonlinear vector multidimensional mapping function for singular systems. This approach is applied to a shielded microstrip line within a region of interest. The parameter values (¿reff(f), Zc(f)) computed with our proposed method are in excellent agreement with those obtained from electromagnetic simulations. © 2007 IEEE.