Development of forecasting theory and methods for developing forecasts in electroenergetics

Abstract

The article discusses several fundamental problems in forecasting problems that form the scientific novelty of the research. One of the problems is formulated as the need to determine the key parameters that form the basis of the forecasting model and allow determining the state of the subject area. The next key problem in the literature is formulated as the curse of dimensionality. Occurs when the researcher tries to take into account the maximum number of indicators and criteria for evaluating the subject area in the model, and this leads to the fact that the computer model required for its solution approaches the Turing limit. The third problem is described in the literature as the problem of supersystem. All elements of the predicted system or process can form higher-level systems that have their own unique properties. This makes it fundamentally impossible to describe a super-system mapping of target functions from the point of view of the systems and processes that make up the super-system. To develop the theory of forecasting and overcome these problems, we propose the use of such sections of modern mathematics as neuromathematics and deep machine learning, chaos theory, catastrophe theory, and the theory of self-organizing systems. It is believed that these methods will increase the depth of the forecast by identifying hidden patterns and relationships among poorly formalized conventional methods of predictive indicators