Deep learning for time series forecasting: The electric load case

Management and efficient operations in critical infrastructures such as smart grids take huge advantage of accurate power load forecasting, which, due to its non-linear nature, remains a challenging task. Recently, deep learning has emerged the machine field achieving impressive performance vast range tasks, from image classification translation. Applications models electric forecasting problem are gaining interest among researchers well industry, but comprehensive sound comparison different—also traditional—architectures is not yet available literature. This work aims at filling gap by reviewing experimentally evaluating four real world datasets on most recent trends contrasting architectures short-term forecast (one-day-ahead prediction). Specifically, focus feedforward recurrent neural networks, sequence-to-sequence temporal convolutional networks along with architectural variants, which known signal processing community novel one.