GUARANTEED NARMAX MODEL FOR THE PREDICTION OF GEOMAGNETIC D-st INDEX

Theoretical analysis of the guaranteed prediction model of the D-st index has been conducted. A new identification approach for the D-st index prediction model using experimental data and an evolutionary algorithm has been proposed. All models are based on an observational data set consisting of V B-s, the solar wind parameters as the input and the D-st index as the output. This method showed a number of advantages over the earlier proposed methods of space weather fore-casting. First, using the dynamic-information approach it is possible to automatically select the most significant regressors from the history of the experimental data. Secondly, it becomes possible to automatically reconstruct a mathematical model and search for optimal model parameters.