Short-term micro-grid load forecast method based on EMD-KELM-EKF

Short-term load forecasting is an important part of micro-grid economic dispatch, and the forecasting error would directly affect the economical efficiency of operation. With respect to large power grid environment, micro-grid is more difficult to realize the short-term load forecasting on the user side. This paper proposes a combined short-term load forecasting model based on Empirical Mode Decomposition (EMD), Extended Kalman Filter (EKF) and Extreme Learning Machine with Kernel (KELM). The time-series data of micro-grid load with high randomness is gradually decomposed into a number of Intrinsic Mode Function (IMF) components through EMD. Two typical different prediction models - EKF and KELM - are adopted to predict different kinds of IMF components. The model prediction accuracy, the stability of period updating and the calculation efficiency is verified through examples analysis of micro-grid of the user side with different types and capacity.