An Evolutionary EMD-FA-ELM approach for Short Term Wind Power Prediction using Wind Speed as Input

A hybrid method combining Efficient Firefly Algorithm (EFA), Empirical Mode Decomposition (EMD) and Special Extreme Learning approach (SELM) is presented in this paper for wind power prediction. The time series wind speed data is decomposed in different Intrinsic Mode Functions (IMFs) by the help of EMD. The optimal IMFs can be selected by the help of the IFA to be given as input to the SELM. The absence of randomness in SELM makes it different from the conventional ELM approach. The optimal IMFs are considered as the input to the SELM to predict the wind speed. Here the wind speed data is considered for 10 minute prediction interval. The predicted wind power results from the proposed optimized EMD-SELM are compared with EMD-SELM and EMD-ELM through various performance metrics. The performance of the IF A is also assessed against other evolutionary approach like differential evolution (DE) and Genetic Algorithm (GA).