DEEPSO to predict wind power and electricity market prices series in the short-term

Abstract

With the advent of restructuring electricity sector and smart grids, combined with the increased variability and uncertainty associated with electricity market prices (EMP) signals and players' behavior, together with the growing integration of renewable energy sources, enhancing prediction tools are required for players and different regulators agents to face the non-stationarity and stochastic nature of such time series, which must be capable of supporting decisions in a competitive environment with low prediction error, acceptable computational time and low computational complexity. Hybrid and evolutionary approaches are good candidates to surpass most of the previous concern considering time series prediction. In this sense, this work proposes a hybrid model composed by a novel combination of differential evolutionary particle swarm optimization (DEEPSO) and adaptive neuro-fuzzy inference system (ANFIS) to predict, in the short-term, the wind power and EMP, testing its results with real and published case studies, proving its superior performance within a robust prediction software tool.