Antenna design based on prior knowledge and BP neural network

The traditional BP(Back Propagation) neural network methods are intensively dependent on training sample data, and it takes the electromagnetic simulation software a great deal of time to generate the sample data. This paper presents a new design method based on neural network for alleviating the bottleneck. In this method, a small number of samples are selected to train the Support Vector Machine (SVM), and the predicted value obtained by the SVM model is taken as a priori knowledge, together with the original training samples as the input of BP neural network. In the design and optimization of inverted-F antenna, the higher accuracy of result is easily yielded within the reduced number of training samples.