Application of Fuzzy Cognitive Maps to Multi-Step Ahead Prediction of Electricity Consumption

This paper is devoted to the multi-step ahead prediction of multivariate time series with the use of fuzzy cognitive maps. The aim of the analysis is to predict the electricity consumption in a household. The learning process was performed with the developed Structure Optimization Genetic Algorithm (SOGA). A comparison with the well-known method for fuzzy cognitive maps learning: the Real-Coded Genetic Algorithm was done. The obtained results show that the SOGA algorithm enables simplifying the fuzzy cognitive map by selection of the most significant concepts and connections and forecasting the three next steps of electricity consumption with satisfactory accuracy.