1. Science
  2. Publications
  3. Scientific Works of Kharkiv National Air Force University
  4. 2(20)'2009
  5. Multivariate nonstationary time series prediction on the base of special artificial neural networks

Multivariate nonstationary time series prediction on the base of special artificial neural networks

S.V. Popov, T.E. Chepenko
Annotations languages:

The article is devoted development of the specialized artificial neuron network, intended for prognostication of multidimensional rows of indexes which describe the conduct of the difficult multiconnected systems and methods of studies of this network. The example of design of process of one-step prognostication of multidimensional row is considered on the basis of artificial neuron network with dynamic neurons and on the basis of artificial neuron network on neurons-filters. The task of construction forecasting the model of the system of the protective signaling is decided, taking into account the non-stationary forecast sequence. A forecasting artificial neuron network is synthesized on dynamic neurons.
Keywords: artificial neuron networks, sentinel row, multidimensional stochastic processes, dynamic neurons-filters, are specialized, eventual and endless impulsive descriptions