Modular Neural Networks in Time Series Modelling

In this paper, modular neural networks, their key features and benefits need to conventional neural networks monolithic architecture. Also in this paper we describe a number of neural networks, which are based on self-organizing Kohonen maps, and that can be successfully applied to the identification of dynamic objects, and describes the new, developed and successfully applied to the identification of dynamic objects modular neural networks, their architecture, learning algorithms, and work, the article reviewed examples of neural networks, and conducted a comparative analysis with several other neural network algorithms of identification of dynamic objects. A brief overview of modern hybrid forecasting methods based on the ANFIS networks. And given the modular neural network model, which contains as modules ANUS.