Visual associative memory for power systems monitoring

This paper describes how the behaviour of a complex dynamic multi-variable system can be represented in an abstracted, but nevertheless meaningful, manner by points and trajectories in a special kind of internal-representation 2-D space. This 2-D space serves as an environment for knowledge acquisition, efficient compilation of experiences, and for associative interpretation of new episodes. The 2-D nature of it facilitates 'visualization' by humans and machines. This present discussion concentrates on exposition of basic underlying concepts and on demonstration of use of this methodology for monitoring of distributed multisensor systems.