Short-term load forecasting for power grids based on MEEMD and neural networks

According to the non-linearity and non-smoothness of power load series, a short-term load forecasting method for power grid based on MEEMD and BP neural network is proposed. The collected load data are first processed by EMD, EEMD, and MEEMD to obtain several IMFs and a ¹trend term, respectively. Then the obtained components are reselected to construct a new signal and algorithmic correlation is introduced to evaluate the new signal. Four models are constructed with the help of BP neural network, which are EMD-BP model, EEMD-BP model, MEEMD-BP model and BP direct prediction model to predict the future load changes, and finally, simulation experiments are conducted on the load data of a region in Shaanxi Province. The simulation results show that the method can effectively improve the accuracy of load prediction.