Developing an Adaptive Neuro-Fuzzy Inference System (ANFIS) based CAD Tool for Designing a Corrugated Horn Antenna

Abstract

Corrugated horns are extensively used as feeds for the reflector antennas. The conventional techniques for designing corrugated horns suffer from a serious limitation in terms of processing time. In other words, such methods consume high time in arriving at the optimum solution. In this thesis, development of Adaptive Neuro-Fuzzy Inference System (ANFIS) based CAD tool is proposed for designing corrugated horns. The proposed CAD tool will accurately predict the horn design parameters with minimum computation time. For designing a corrugated horn using this CAD tool, the horn specifications (frequency of operation, bandwidth, beam width, cross-polar level, etc.) will be given as inputs, while the horn physical parameters will be obtained as the output. The CAD tool will be developed based on the Adaptive Neuro-Fuzzy Inference System (ANFIS), which is a very powerful soft-computing technique. It combines the best features of Fuzzy Inference Systems (FISs) and Artificial Neural Networks (ANNs). In order to develop an ANFIS based model, the required training data will be generated using the commercially available full-wave solver. Once trained, the model will be very easy to implement and will find out the optimum design parameters with high degree of accuracy. In addition to this, the model will take very less processing time as compared to the conventional methods.