High-frequency circuit design using a neural space-mapping algorithm based on a two-layer perceptron with optimized nonlinearity
View/
Date
2006-11Author
Rayas-Sánchez, José E.
Gutiérrez-Ayala, Vladimir
Metadata
Show full item recordDescription
In this work we present an improved version of the Neural Space-Mapping algorithm with regulated nonlinearity. The new version uses a nonlinear two-layer perceptron (2LP), instead of a three layer perceptron (3LP), to train the space-mapping (SM)-based neuromodel. The 2LP mapping nonlinearity is automatically regulated with classical optimization algorithms. Additionally, the new algorithm uses a different optimization method to train the SM-based neuromodel. With these three main improvements we obtain a more efficient and faster algorithm. In order to verify the algorithm performance, we design a stopband microstrip filter with quarter-wave resonant opens stubs, and a microstrip notch filter with mitered bends. Both circuits use a full-wave electromagnetic simulator.ITESO, A.C.
Consejo Nacional de Ciencia y Tecnología