Прогнозирование экстремальных ситуаций на основе анализа многомерных разнотипных временных рядов

In this paper, we suggest a method for prediction of extreme events using logical probabilistic models. We assume that these models are based on analysis of the multidimensional time series. In the proposed method, a weighed distance (similarity) in heterogeneous parameters space is used. An optimization method for multiple extremes on simplex is suggested for searching for the optimal weights