Short-Term Load Forecasting with an Ensemble Model Based on 1D-UCNN and Bi-LSTM

Short-term load forecasting (STLF), especially for regional aggregate forecasting, is essential in smart grid operation and control. However, the existing CNN-based methods cannot efficiently extract features from electricity load. The reason that basic requirement of using CNNs space invariance, which not satisfied by actual data. In addition, models multi-scale input representing tendency load, resulting a reduction performance. As solution, this paper proposes novel ensemble model, four-stage framework composed feature extraction module, densely connected residual block (DCRB), bidirectional long short-term memory layer (Bi-LSTM), thinking. model first extracts derived raw data module. comprise hourly average temperature features, can capture huge randomness trend characteristics DCRB effectively above compared with models. experiment results show proposed method provide higher performance than models, almost 0.9–3.5%.