Method of ionospheric data analysis based on a combination of wavelet transform and neural networks

Abstract

The paper presents a hybrid system based on a combination of wavelet filtering operations and regression neural networks. The system is adapted to analyze the ionosphere data obtained at "Paratunka" station (Kamchatka). Testing of the system has shown its efficiency in the tasks of analysis of characteristic properties of ionospheric data and detection of anomalies occurring during disturbed periods. For a detailed analysis of anomalies, computing solutions based on the application of continuous wavelet transform and threshold functions were suggested. The developed computational tools were implemented in software environment.