Agent Strategies in Smart Energy Markets - PowerTAC 2016

Abstract

The current energy models and infrastructures need to be restructured in order to face the changes in energy consumption, production and management. The adoption of renewable power sources combined with the capability of a more reasonable and autonomous participation on the grid lead to this energy revolution. These changes demand improvements in the way participants act, not only related to the physical electricity grid, but mainly regarding the related services, as energy markets. Since there is no real-life market to test new approaches for smart grid markets, simulations should be used. This work focus on the PowerTAC simulation framework, a top-of-the-art platform in which competitors develop broker agents to enact market companies. In this context, the tariff composition problem plays a fundamental role since customers (both real and simulated) interact with the market by selecting a tariff. Brokers should seek for creating and updating tariffs to be competitive and still profitable. Broker's performance is given by its market share and profit on the market. Current competitors use a centralized approach, with focus on single features to compose tariffs. In this work an alternative approach to this problem is presented. We propose the creation of a Broker that is inherently a Multi-Agent System - a broker composed by different specialist agents that evaluate different features to compose the final tariff. To validate the results from our work we will analyse the results from the participation of our broker on the PowerTAC, firstly in a local version of the competition against previous years public competitors and secondly, by trying to qualify to the annual competition.