A Fully Probabilistic Decentralised Control Design for Complex Stochastic Systems

Abstract

Computational and communication complexity call for robustness of controlled systems as well as for distributed control. The proposed technical solutions in this paper are bottom up solutions where simple controllers are designed to care about individual nodes either completely independently or within various structures like cascade control. Cheap computational resources allow now the improvement of the overall behaviour of the network of such controlled loops by allowing the individual ”nodes” to share information with their neighbours without aiming at hopeless global solution. The current paper inspects this proposed method on a linearised version of coupled map lattice with spatiotemporal chaos yielding close to linear quadratic design which gives insight into possible behaviours of such networks.