Real-time Pulse System Emergency Forecasting Through Time-series Intelligent Processing

Pulse energy conversion system (PECS) safe and reliable operation depends on timely forecasting of emergencies, which are connected with PECS dynamics evolution in the direction of undesirable dynamic regimes (such as sub-harmonic, quasi-periodic and chaotic processes). The main deficiency of the known approaches regarding this problem solving is an absence of the guarantee, that correct decision on the next stationary process in PECS dynamics is made before this process have settled. The approach offered is directed to the problem solution by PECS dynamics mapping from multi-D spaces into their small dimensional projections without useful information loss. In the paper the experimental investigation results, the purpose of which was an estimation of the possibility of practical use of the approach to real-time emergency forecasting in PECS dynamics, are presented.