Application of ANNs in Evaluation of Microwave Pyramidal Absorber Performance

摘要

To evaluate the overall anechoic chamber performance it is necessary to determine reflectivity of the absorbers. As manufacturer specifications usually give only information about frequency dependent reflection coefficient at normal incidence of EM waves, a time-consuming electromagnetic analysis is necessary to calculate the reflection coefficient at off-normal incident angles. In this paper, an efficient alternative approach to obtain the reflection coefficient at off-normal incidence is proposed. It is based on artificial neural networks trained to model the absorber reflectivity dependence on the frequency and incident angle of horizontally and vertically polarized electromagnetic waves. The model has been developed for pyramidal absorbers at low microwave frequencies (0.4 GHz-1 GHz).