Multi-agent based metalearner using genetic algorithm for decision support in electricity markets

The continuous changes in electricity markets' mechanisms and operations turn this environment into a challenging domain for the participating entities. Simulation tools are increasingly being used for decision support purposes of such entities. In particular, multi-agent based simulation, which facilitates the modeling of different types of mechanisms and players, is being fruitfully applied to the study of worldwide electricity markets. An effective decision support to market players' negotiations is, however, still not properly reached due to the uncertainty that results from the increasing penetration of renewable generation and the complexity of market mechanisms themselves. In this scope, this paper proposes a novel metalearner that provides decision support to market players in their negotiations. The proposed metalearner uses as input the output of several other market negotiation strategies, which are used to create a new, enhanced response. The final result is achieved through the combination and evolution of the strategies' learning results by applying a genetic algorithm.