Development of precursors recognition methods in vector signals

Precursor recognition methods in vector signals of plasma diagnostics are presented. Their requirements and possible options for their development are considered. In particular, the variants of using symbolic regression for building a plasma disruption prediction system are discussed. The initial data preparation using correlation analysis and symbolic regression is discussed. Special attention is paid to the possibility of using algorithms in real time.