Multi agent based energy management control for commercial buildings

Abstract

This paper discusses the use of Multi-Agent- Systems to control various systems in a commercial building in order to achieve maximum energy efficiency while maintaining comfort for the occupants and allowing a possible interconnection with a smart-grid. An approximated optimal control is proposed in this paper, where on-line training of a Bayesian state-machine learns the system for a given utility function. Different aspects and challenges associated with the control of a building will be discussed, and a control scheme using Multi-Agent-Systems technology is proposed.