Software implementation of space mapping based neuromodels of microwave components
View/ Open
Date
2000-04Author
Bandler, John W.
Rayas-Sánchez, José E.
Zhang, Qi J.
Metadata
Show full item recordDescription
We present novel realizations of SM based neuromodels of microwave components using available software. In the SM based neuromodeling techniques a neural network is used to implement a mapping from the electromagnetic to the circuit-theoretic input space. The implicit knowledge in the circuit model allows us to decrease not only the number of learning points needed, but also the complexity of the neural network and to improve the generalization performance. A Frequency Space Mapped Neuromodel (FSMN) of a microstrip right angle bend is implemented using NeuroModeler, and entered into ADS as a library component through an ADS plug-in module.Consejo Nacional de Ciencia y Tecnología
Carleton University