Abstract:In recent years, the installed capacity of wind power has rapidly increased. And the wind power prediction is the foundation for ensuring large-scale wind power grid connection. The current short-term prediction methods of wind farm cluster (WFC) are difficult to sufficiently extract spatiotemporal features to achieve high-precision prediction. The article proposes a short-term power prediction method for WFC based on deep attention embedded graph clustering-TimesNet (DAEGC-TimesNet). Firstly, the directed power curves of WFC are proposed to analyze wind data. Then, a graph attention network is constructed based on geographic location and numerical weather prediction (NWP) information to guide clustering algorithms to achieve effective cluster partitioning. Finally, the input of model is constructed based on the feature information from various sub-clusters of WFC and the prediction result is obtained through the TimesNet. The method is applied to WFCs in Inner Mongolia, Jilin province and Yunnan province of China. The result shows that the RMSE reduces 0.0155 and the MAE reduces 0.0156 as well as the coefficient of determination and accuracy rate are highest comparing with comparative algorithms averagely based on above three WFCs. The simulation results are superior to the comparison algorithms, which makes greater contributions to ensure regional power supply.

What problem does this paper attempt to address?

### Problems Addressed by the Paper The paper aims to address the issue of high-precision short-term power prediction in Wind Farm Clusters (WFC). Specifically, current short-term prediction methods for WFCs struggle to fully extract spatiotemporal features, resulting in low prediction accuracy. Therefore, the paper proposes a short-term power prediction method for WFCs based on the fusion of multi-source spatiotemporal feature information—Deep Attention Embedded Graph Clustering-TimesNet (DAEGC-TimesNet), to improve prediction accuracy. ### Main Contributions 1. **Constructing a Graph Attention Network based on Geographic Location Information and Numerical Weather Prediction (NWP) Information**: - Using an AutoEncoder to effectively utilize the interaction between graph structure and content information, jointly optimizing the loss functions of both parts to learn embeddings and clustering. - The self-training module improves clustering performance by operating on embeddings, further mining hidden information between graph structure and content, and enhancing clustering accuracy. 2. **Proposing a Directional Power Curve**: - Analyzing the relationship between wind measurement data and wind direction, dividing wind speed and wind power data into 16 wind direction sectors, and plotting wind speed-power scatter plots. - Fitting the wind speed-power curve and identifying and removing outlier data points using the quartile method to improve fitting accuracy. 3. **Using the TimesNet Model to Explore the Multi-Periodicity of NWP Time Series**: - Further capturing periodic changes within different periods to enhance the model's expressive ability. - Extending one-dimensional time series data into a two-dimensional space for analysis, reducing the impact of mixing and overlapping multiple time series (such as rising, falling, and fluctuating), and improving prediction accuracy. ### Experimental Validation The paper applied this method to WFCs in Inner Mongolia, Jilin, and Yunnan, and the results showed: - Root Mean Square Error (RMSE) reduced by 0.0155. - Mean Absolute Error (MAE) reduced by 0.0156. - The coefficient of determination and accuracy were higher in the aforementioned three WFCs compared to the benchmark algorithms. These results indicate that the proposed DAEGC-TimesNet method has a higher contribution to ensuring regional power supply.

A short-term power prediction method for wind farm cluster based on the fusion of multi-source spatiotemporal feature information

Ultra-short-term wind farm cluster power prediction based on FC-GCN and trend-aware switching mechanism

Short-Term Power Forecasting of Wind Farm Cluster Based on Global Information Adaptive Perceptual Graph Convolution Network

Short term wind power prediction for regional wind farms based on spatial-temporal characteristic distribution

A Spatiotemporal Coupling Calculation-Based Short-Term Wind Farm Cluster Power Prediction Method

An Ultra‐Short‐Term Wind Power Prediction Method Based on Spatiotemporal Characteristics Fusion

M2GSNet: Multi-Modal Multi-Task Graph Spatiotemporal Network for Ultra-Short-Term Wind Farm Cluster Power Prediction

Prediction method of wind farm power generation capacity based on feature clustering and correlation analysis

Dynamic spatio-temporal correlation and hierarchical directed graph structure based ultra-short-term wind farm cluster power forecasting method

Short-term Power Prediction Based on Improved NWP Data Considering Wind Speed Distribution Inner Wind Farm

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

On the Use of Spatial-temporal Feature Graph Convolutional Network for Ultra-Short Term Prediction of Wind Power Generation

Wind Power Prediction for Wind Farm Clusters Based on the Multifeature Similarity Matching Method

Spatio-Temporal Graph Neural Network and Pattern Prediction Based Ultra-Short-Term Power Forecasting of Wind Farm Cluster

Short-Term Regional Wind Power Statistical Upscaling Forecasting Based on Feature Clustering

Short-term wind farm cluster power prediction based on dual feature extraction and quadratic decomposition aggregation

Ultra-short-term power forecast method for the wind farm based on feature selection and temporal convolution network

Wind Power Group Prediction Model Based on Multi-Task Learning

Short-term wind farm cluster power point-interval prediction based on graph spatio-temporal features and S-Stacking combined reconstruction

An Ultra-Short-Term Power Prediction Method for Wind Farms in Northwest China Based on Federated Learning

Ultra-short-term/short-term wind power continuous prediction based on fuzzy clustering analysis