Direct-coupled cavity filters design using a hybrid feedforward neural network–finite elements procedure.

In many filtering applications, direct-coupled cavity filters are often used for their symmetry and handling easiness. In this paper, a method is described for the design of these filtering devices by using an artificial neural network (ANN). Starting from design requirements, the procedure discussed herein directly determines the geometric dimensions of the filtering device with good accuracy and very short processing time. ©1999 John Wiley & Sons, Inc. Int J RF and Microwave CAE 9: 287–296, 1999. [ABSTRACT FROM AUTHOR]