Microwave cylindrical cavity applicators modeling using knowledge based neural network

In this paper, the loaded cylindrical metallic cavity with circular cross-section is modeled using knowledge based neural networks (KBNN). The load in the form of a homogeneous dielectric slab with losses located on the bottom of the cavity is considered. The appropriate neural model is investigated in which knowledge about resonant frequency behaviour, defined in approximate approach, is integrated. In the aim of comparison, the considered cavity is modeled using classical multilayer perception (MLP) network, too. For the same training set the appropriate MLP model, giving best results, is investigated. The accuracy of both models as well as the advantage of using KBNN model is illustrated through the example of TM/sub 112/ mode.