2020/07/04 by Kevin Trebing, Trebing, Kevin, Siamak Mehrkanoon +1
Engineering · #Energy Load and Power Forecasting
paper · pdf · doi:10.48550/arxiv.2007.12567
Accurate wind speed forecasting is of great importance for many economic,\nbusiness and management sectors. This paper introduces a new model based on\nconvolutional neural networks (CNNs) for wind speed prediction tasks. In\nparticular, we show that compared to classical CNN-based models, the proposed\nmodel is able to better characterise the spatio-temporal evolution of the wind\ndata by learning the underlying complex input-output relationships from\nmultiple dimensions (views) of the input data. The proposed model exploits the\nspatio-temporal multivariate multidimensional historical weather data for\nlearning new representations used for wind forecasting. We conduct experiments\non two real-life weather datasets. The datasets are measurements from cities in\nDenmark and in the Netherlands. The proposed model is compared with traditional\n2- and 3-dimensional CNN models, a 2D-CNN model with an attention layer and a\n2D-CNN model equipped with upscaling and depthwise separable convolutions.\n