Research on Short-Term Load Forecasting Based on Optimized GRU Neural Network

Accurate short-term load forecasting can ensure the safe and stable operation of power grids, but nonlinear increases complexity forecasting. In order to solve problem modal aliasing in historical data, fully explore relationship between time series characteristics this paper proposes a gated cyclic network model (SSA–GRU) based on sparrow algorithm optimization. Firstly, complementary sets empirical mode decomposition (EMD) are used decompose original data obtain characteristic components. The SSA–GRU combined is predict components, finally prediction results, complete Taking real company as an example, compares CEEMD–SSA–GRU with EMD–SSA–GRU, SSA–GRU, GRU models. Experimental results show that has better effect than other