An Applied Energy Management Approach in Intelligent Environments based on a Hybrid Agent Architecture

This paper presents a framework for ambient sensing and managing a University building aimed at energy savings and user comfort. The system builds upon previous work, using a Semantic Web Service middleware for unifying the various heterogeneous sensor and actuator networks. Two applications are introduced in the framework, the Manager App and the Rule App. The latter incorporates a hybrid intelligent agent that enables both reactive and deliberate manipulation of the environment, based on user-configurable policies expressed in defeasible logic. The Manager App provides users with advanced control and renders the system sustainable. Specifically, it allows bypassing the policies and manually administrating the infrastructure, e.g. during exceptional or emergency cases. The framework is deployed and evaluated at a University course office, guaranteeing 20% daily energy savings on controlled devices, aggregated to 17% total per room savings.