An application of learning agents to smart energy domains

The main requirement for building an Internet of Things is the definition of smart objects in which it needs to put intelligence. The pervasive deployment of smart objects will add value to applications by capabilities of communication, negotiation, learning and distributed reasoning. In this paper we investigate how the paradigm shift from objects to agents is the driver for developing these capabilities by a case study in the context of Smart Energy application domain. In fact the paradigm shift we are seeing in these years is to consider the electricity network like an Internet of Energy, where each and every electrical device and generator will be connected in a network and able to communicate data and receive and react in real time to events and stimuli that arrive from other devices or from the grid: a scattered network of sensors, actuators, communication nodes, systems control and monitoring. Here we present the learning-based approach for power management in smart grids providing an agent-oriented modeling of the energy market. The main issue we focus on is a reasonable compromise between the resolution of the consuming profile representation and the performance and real time requirements of the system.