Recurrent inception convolution neural network for multi short-term load forecasting

• Published in: Energy and buildings Vol. 194; pp. 328 - 341
• Main Authors: Kim, Junhong, Moon, Jihoon, Hwang, Eenjun, Kang, Pilsung
• Format: Journal Article
• Language: English
• Published: Lausanne Elsevier B.V 01.07.2019; Elsevier BV
• Subjects: Artificial neural networks; Climate change; Convolution; Convolution neural network; Deep learning; Distributed generation; Electricity consumption; Energy management; Energy management systems; Energy storage; Forecasting; Load forecasting; Mathematical models; Modules; Multilayers; Neural networks; Predictions; Recurrent inception convolution neural network; Recurrent neural network; Recurrent neural networks; Renewable energy; Smart grid; State vectors; Storage systems; DNN; PV; ELM; DL; RICNN; IPSO-ANN; Deep learning; DT; MLP; HVAC; MLR; RNN; VSTLF; Recurrent neural network; Convolution neural network; PDRNN; GA-ANFIS; RBM; KMA; SRWNN; LTLF; EMS; MAPE; PSO; RF; FCRBM; ReLU; STLF; ANN; CNN; SFOA; Recurrent inception convolution neural network; MTLF; GRNN; SVR; IoT; SNN; Load forecasting; NLP; WNN; MWD; MSE; 1-D CNN; KEPCO; LSTM; ESS; CRBM; RMSE; S2S; ARIMA; ICT
• ISSN: 0378-7788; 1872-6178
• DOI: 10.1016/j.enbuild.2019.04.034

•A new multi short-term load forecasting model named RICNN is proposed.•The proposed model combines an RNN and a 1-D CNN of inception module.•The proposed RICNN yields better forecasting performance than MLP, 1-D CNN and RNN.•The proposed RICNN is verified by actual power consumption data collected from three industrial complexes in South Korea. Smart grid and microgrid technology based on energy storage systems (ESS) and renewable energy are attracting significant attention in addressing the challenges associated with climate change and energy crises. In particular, building an accurate short-term load forecasting (STLF) model for energy management systems (EMS) is a key factor in the successful formulation of an appropriate energy management strategy. Recent recurrent neural network (RNN)-based models have demonstrated favorable performance in electric load forecasting. However, when forecasting electric load at a specific time, existing RNN-based forecasting models neither use a predicted future hidden state vector nor the fully available past information. Therefore, once a hidden state vector has been incorrectly generated at a specific prediction time, it cannot be corrected for enhanced forecasting of the following prediction times. To address these problems, we propose a recurrent inception convolution neural network (RICNN) that combines RNN and 1-dimensional CNN (1-D CNN). We use the 1-D convolution inception module to calibrate the prediction time and the hidden state vector values calculated from nearby time steps. By doing so, the inception module generates an optimized network via the prediction time generated in the RNN and the nearby hidden state vectors. The proposed RICNN model has been verified in terms of the power usage data of three large distribution complexes in South Korea. Experimental results demonstrate that the RICNN model outperforms the benchmarked multi-layer perception, RNN, and 1-D CNN in daily electric load forecasting (48-time steps with an interval of 30 min).