Short-Term Power Load Forecasting Based on RRPNN-CEEMD-BiLSTM Model
With the continuous development of low-carbon economy,accurate power load forecasting is crucial for energy management.To improve the accuracy of short-term power load forecasting,short-term power load forecasting based on an improved ridge polynomial neural network(RRPNN),complementary ensemble empirical modal decom-position(CEEMD)and bi-directional long short-term memory network(BiLSTM)is proposed.Firstly,the output ac-curacy of the RRPNN is adjusted by autoregressive inputs to enable fast processing of non-linear load data and to a-chieve initial forecasts.Then,the CEEMD and BiLSTM methods are used to reduce modal confounding and inter-mo-dal interactions to obtain accurate modal component forecasting results.Finally,a simulation test is carried out to veri-fy the actual grid load in a region as an example.The experimental validation shows that the RRPNN-CEEMD-BiL-STM model can effectively achieve accurate short-term power load forecasting with high load forecasting accuracy compared with other models such as RRPNN-BiLSTM,CEEMD-BiLSTM and RRPNN-PSO-BiLSTM.
Short-term load forecastingRidge polynomial neural networksEmpirical modal decompositionBidi-rectional long and short term memory networks
万方数据
维普
