Automatic Processing of Ionograms on the Basis of the Artificial Neural Network Method

摘要

The Hopfield artifiecial neural network method (ANN) was used in processing vertical-incidence (VIS), oblique-incidence (OIS) and backscatter (BSS0 ionograms. Using a modified rotor model for ANN and the mean field theory (MFT) it was possible to implement an algorithm that ensures the convergence of ANN. An algorithm for data reduction is proposed to include the information density on the ionogram. The algorithm was tested against ionograms obtaiend with a chirp-sonde.