Neural network analysis of dynamically nonequilibrium situations of economy sectors interaction by using simulating model

We propose hierarchical structure for economy sectors interaction management system as a part of multisector macroeconomic system. This structure contains three management levels. We used principles of feedback, adaptation and situation management. Developed structure differs by utilization of simulating models for scenario experiments, and application of neural networks and production models for solution the weakly formalized problems of multi-path analysis for dynamically nonequilibrium situations. We propose procedure for design of intelligent algorithms of economy sector interaction process situation management. The procedure provides multi-stage clustering of dynamically nonequilibrium situations, which is based on Kohonen neural networks. Neural net clustering of situations implies coherent decomposition of situations set over several classification criterions differing by selected levels of economy sector interaction process analysis. We developed Kohonen neural networks for multi-stage clustering of situations and designed self-organizing maps for clusters of dynamically nonequilibrium situations.