Spatiotemporal Attention Networks for Wind Power Forecasting

Xingbo Fu,Feng Gao,Jiang Wu,Xinyu Wei,Fangwei Duan

DOI: https://doi.org/10.48550/arXiv.1909.07369

2019-11-13

Abstract:Wind power is one of the most important renewable energy sources and accurate wind power forecasting is very significant for reliable and economic power system operation and control strategies. This paper proposes a novel framework with spatiotemporal attention networks (STAN) for wind power forecasting. This model captures spatial correlations among wind farms and temporal dependencies of wind power time series. First of all, we employ a multi-head self-attention mechanism to extract spatial correlations among wind farms. Then, temporal dependencies are captured by the Sequence-to-Sequence (Seq2Seq) model with a global attention mechanism. Finally, experimental results demonstrate that our model achieves better performance than other baseline approaches. Our work provides useful insights to capture non-Euclidean spatial correlations.

Machine Learning

What problem does this paper attempt to address?

This paper attempts to solve two main problems in wind power generation prediction: temporal dependence and spatial correlation. Specifically, due to its volatility and randomness, wind power generation is difficult to predict accurately. In order to improve the prediction accuracy, it is necessary not only to capture the temporal dependence in the time series but also to construct the spatial correlation between the target wind farm and other wind farms. Existing methods such as statistical models (e.g., historical averaging method and autoregressive moving average model ARMA) have limitations when dealing with non - stationary time - series data, while traditional neural network models (such as recurrent neural network RNN, long - short - term memory network LSTM and gated recurrent unit GRU) can extract explicit temporal dependence but do not consider the spatial dependence between wind farms sufficiently. In addition, although convolutional neural network (CNN) performs well in dealing with data with grid topology, it is not effective in modeling non - Euclidean correlations between different wind farms. To solve these problems, the paper proposes a new framework based on Spatiotemporal Attention Networks (STAN). This framework extracts the spatial correlation between wind farms through the multi - head self - attention mechanism and captures the temporal dependence through the sequence - to - sequence (Seq2Seq) model with a global attention mechanism. The experimental results show that, compared with the existing baseline methods, this model performs better in predicting wind power generation, and in particular, provides useful insights when dealing with non - Euclidean spatial correlations.

Spatiotemporal Attention Networks for Wind Power Forecasting

Multi-Source and Temporal Attention Network for Probabilistic Wind Power Prediction

A Self-Attention Based Method for Wind Speed Forecasting in High-Speed Railway System

An ultra-short-term wind power prediction method based on spatial-temporal attention graph convolutional model

Enhancing Short-Term Wind Speed Forecasting using Graph Attention and Frequency-Enhanced Mechanisms

Integrating spatio-positional series attention to deep network for multi-turbine short-term wind power prediction

Spatio-Temporal- Transformation-based Method for Hour-Ahead Wind Power Forecasting

Spatial correlation learning based on graph neural network for medium-term wind power forecasting

Spatio-temporal Graph Cross-Correlation Auto-Encoding Network for Wind Power Prediction

Spatio-temporal Markov Chain Model for Very-Short-term Wind Power Forecasting

Long-term Wind Power Forecasting with Hierarchical Spatial-Temporal Transformer

A multi-factor driven spatiotemporal wind power prediction model based on ensemble deep graph attention reinforcement learning networks

A novel automatic wind power prediction framework based on multi-time scale and temporal attention mechanisms

Short-term wind power prediction framework using numerical weather predictions and residual convolutional long short-term memory attention network

Wind power forecasting using attention-based gated recurrent unit network

Series-wise Attention Network for Wind Power Forecasting Considering Temporal Lag of Numerical Weather Prediction

Interpretable multi-graph convolution network integrating spatio-temporal attention and dynamic combination for wind power forecasting

Weather Forecasting Using Ensemble of Spatial-Temporal Attention Network and Multi-Layer Perceptron

A Graph Attention Network with Spatio-Temporal Wind Propagation Graph for Wind Power Ramp Events Prediction

Short-term multi-step wind power forecasting based on spatio-temporal correlations and transformer neural networks

A Spatiotemporal Deep Neural Network for Fine-Grained Multi-Horizon Wind Prediction