For power load sequence is not smooth, strong randomness, input model directly will lead to poor fitting effect and low prediction accuracy, this paper proposes a add complementary white noise based on complementary ensemble empirical mode decomposition (CEEMD) and gated recurrent unit neural network (GRU) combination forecast method of power load short-term prediction precision is improved effectively. First of all, for the empirical mode decomposition (EMD) to deal with the sequence of large interference signals there is a modal aliasing problem, proposed CEEMD decomposition method, the addition of complementary white noise, the original sequence into different scales of the sub-sequence, then use GRU neural network, and optimize the network super parameters, so as to get the best prediction results. The experiment results show that the reconstruction error is small and the prediction effect is good.
