In stepped-impedance lowpass filters (SI-LPFs), electrically short high- and low-impedance sections are cascaded to approximate the corresponding ladder LC lumped-circuit prototype. Typically, such filters are designed using charts and tables. Hany Taher, who hails from Saudi Arabia's Umm Elqura University and Egypt's Electronics Research Institute, instead uses an artificial neural network (ANN) to learn the highly nonlinear relationship between the design specifications and physical dimensions of the microstrip SI-LPF.
This ANN was constructed by learning from a set of input/output data. After training, the ANN can generalize the relationship between the input and output. Essentially, ANN models were built to relate the physical dimensions of the target design to the design specificationswithout using any charts or tables. Using this technique, designers gain a fast and accurate computer-aided-design (CAD) tool to design a microstrip SI-LPF. See "A New Artificial Intelligent Technique for Designing of Microstrip Stepped Impedance Low-Pass Filter," Microwave And Optical Technology Letters, September 2010, p. 1946.
