Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

WaveHiTS: Wavelet-Enhanced Hierarchical Time Series Modeling for Wind Direction Nowcasting in Eastern Inner Mongolia

Hailong Shu, Weiwei Song, Yue Wang, Jiping Zhang

TL;DR

The paper tackles the challenging problem of wind direction nowcasting, where circular data and multi-step error accumulation hinder traditional models. It introduces WaveHiTS, which integrates a discrete wavelet transform with Neural Hierarchical Interpolation and couples wind direction to linear U–V components to handle circularity. Empirical results on Inner Mongolia data show WaveHiTS significantly outperforms recurrent, transformer, and hybrid baselines across six-step horizons, achieving the lowest RMSE and highest VCC/HR, with robust performance and minimal error propagation. The ablation study confirms contributions from wavelets, UV decomposition, and hierarchical structure, underscoring practical implications for improved yaw control and wind-energy grid integration.

Abstract

Wind direction forecasting plays a crucial role in optimizing wind energy production, but faces significant challenges due to the circular nature of directional data, error accumulation in multi-step forecasting, and complex meteorological interactions. This paper presents a novel model, WaveHiTS, which integrates wavelet transform with Neural Hierarchical Interpolation for Time Series to address these challenges. Our approach decomposes wind direction into U-V components, applies wavelet transform to capture multi-scale frequency patterns, and utilizes a hierarchical structure to model temporal dependencies at multiple scales, effectively mitigating error propagation. Experiments conducted on real-world meteorological data from Inner Mongolia, China demonstrate that WaveHiTS significantly outperforms deep learning models (RNN, LSTM, GRU), transformer-based approaches (TFT, Informer, iTransformer), and hybrid models (EMD-LSTM). The proposed model achieves RMSE values of approximately 19.2°-19.4° compared to 56°-64° for deep learning recurrent models, maintaining consistent accuracy across all forecasting steps up to 60 minutes ahead. Moreover, WaveHiTS demonstrates superior robustness with vector correlation coefficients (VCC) of 0.985-0.987 and hit rates of 88.5%-90.1%, substantially outperforming baseline models. Ablation studies confirm that each component-wavelet transform, hierarchical structure, and U-V decomposition-contributes meaningfully to overall performance. These improvements in wind direction nowcasting have significant implications for enhancing wind turbine yaw control efficiency and grid integration of wind energy.

WaveHiTS: Wavelet-Enhanced Hierarchical Time Series Modeling for Wind Direction Nowcasting in Eastern Inner Mongolia

TL;DR

The paper tackles the challenging problem of wind direction nowcasting, where circular data and multi-step error accumulation hinder traditional models. It introduces WaveHiTS, which integrates a discrete wavelet transform with Neural Hierarchical Interpolation and couples wind direction to linear U–V components to handle circularity. Empirical results on Inner Mongolia data show WaveHiTS significantly outperforms recurrent, transformer, and hybrid baselines across six-step horizons, achieving the lowest RMSE and highest VCC/HR, with robust performance and minimal error propagation. The ablation study confirms contributions from wavelets, UV decomposition, and hierarchical structure, underscoring practical implications for improved yaw control and wind-energy grid integration.

Abstract

Wind direction forecasting plays a crucial role in optimizing wind energy production, but faces significant challenges due to the circular nature of directional data, error accumulation in multi-step forecasting, and complex meteorological interactions. This paper presents a novel model, WaveHiTS, which integrates wavelet transform with Neural Hierarchical Interpolation for Time Series to address these challenges. Our approach decomposes wind direction into U-V components, applies wavelet transform to capture multi-scale frequency patterns, and utilizes a hierarchical structure to model temporal dependencies at multiple scales, effectively mitigating error propagation. Experiments conducted on real-world meteorological data from Inner Mongolia, China demonstrate that WaveHiTS significantly outperforms deep learning models (RNN, LSTM, GRU), transformer-based approaches (TFT, Informer, iTransformer), and hybrid models (EMD-LSTM). The proposed model achieves RMSE values of approximately 19.2°-19.4° compared to 56°-64° for deep learning recurrent models, maintaining consistent accuracy across all forecasting steps up to 60 minutes ahead. Moreover, WaveHiTS demonstrates superior robustness with vector correlation coefficients (VCC) of 0.985-0.987 and hit rates of 88.5%-90.1%, substantially outperforming baseline models. Ablation studies confirm that each component-wavelet transform, hierarchical structure, and U-V decomposition-contributes meaningfully to overall performance. These improvements in wind direction nowcasting have significant implications for enhancing wind turbine yaw control efficiency and grid integration of wind energy.

Paper Structure

This paper contains 33 sections, 10 equations, 5 figures, 8 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Traditional Statistical Methods
  4. Machine Learning Approaches
  5. Deep Learning Models
  6. Attention-Based Models and Transformers
  7. Hybrid and Ensemble Methods
  8. Wavelet Transform in Wind Forecasting
  9. Methodology
  10. Motivation for Model Design
  11. Proposed Methodology Overview
  12. Circular Data Handling
  13. Wavelet Transform for Frequency Analysis
  14. Neural Hierarchical Interpolation for Time Series (N-HiTS)
  15. Experimental Setup
  16. ...and 18 more sections

Figures (5)

  • Figure 1: Flowchart of the proposed approach combining wavelet and NHiTS
  • Figure 2: The wind speed, wind direction, U, and V components of some data are displayed
  • Figure 3: Comparison forecasted values with EMDLSTM and WaveHiTS.
  • Figure 4: Comparison forecasted values with EMDLSTM and WaveHiTS.
  • Figure 5: Comparison forecasted values with EMDLSTM and WaveHiTS.