Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Novel Denoising Technique and Deep Learning Based Hybrid Wind Speed Forecasting Model for Variable Terrain Conditions

Sourav Malakar, Saptarsi Goswami, Amlan Chakrabarti, Bhaswati Ganguli

TL;DR

This work tackles the challenge of short-term wind speed forecasting in complex terrain by integrating Ensemble Empirical Mode Decomposition (EEMD) with a PACF-driven IMF reduction, SampEn-based complexity assessment, and a novel bidirectional feature-LSTM approach. The model adaptively assigns simple IMFs to unidirectional forecasting and complex IMFs to a bidirectional LSTM, resulting in reduced training time and improved accuracy across plain and complex terrains, seasons, and forecasting horizons. Empirical results show substantial gains over persistence and various DL-based baselines, along with lower inter-terrain variance (as low as $1.75\%$ in some metrics) and strong multi-step performance. The findings highlight the value of adaptive, complexity-aware decomposition and bidirectional temporal modeling for wind speed prediction in heterogeneous landscapes, with implications for wind energy operations and grid integration.

Abstract

Wind flow can be highly unpredictable and can suffer substantial fluctuations in speed and direction due to the shape and height of hills, mountains, and valleys, making accurate wind speed (WS) forecasting essential in complex terrain. This paper presents a novel and adaptive model for short-term forecasting of WS. The paper's key contributions are as follows: (a) The Partial Auto Correlation Function (PACF) is utilised to minimise the dimension of the set of Intrinsic Mode Functions (IMF), hence reducing training time; (b) The sample entropy (SampEn) was used to calculate the complexity of the reduced set of IMFs. The proposed technique is adaptive since a specific Deep Learning (DL) model-feature combination was chosen based on complexity; (c) A novel bidirectional feature-LSTM framework for complicated IMFs has been suggested, resulting in improved forecasting accuracy; (d) The proposed model shows superior forecasting performance compared to the persistence, hybrid, Ensemble empirical mode decomposition (EEMD), and Variational Mode Decomposition (VMD)-based deep learning models. It has achieved the lowest variance in terms of forecasting accuracy between simple and complex terrain conditions 0.70%. Dimension reduction of IMF's and complexity-based model-feature selection helps reduce the training time by 68.77% and improve forecasting quality by 58.58% on average.

A Novel Denoising Technique and Deep Learning Based Hybrid Wind Speed Forecasting Model for Variable Terrain Conditions

TL;DR

This work tackles the challenge of short-term wind speed forecasting in complex terrain by integrating Ensemble Empirical Mode Decomposition (EEMD) with a PACF-driven IMF reduction, SampEn-based complexity assessment, and a novel bidirectional feature-LSTM approach. The model adaptively assigns simple IMFs to unidirectional forecasting and complex IMFs to a bidirectional LSTM, resulting in reduced training time and improved accuracy across plain and complex terrains, seasons, and forecasting horizons. Empirical results show substantial gains over persistence and various DL-based baselines, along with lower inter-terrain variance (as low as in some metrics) and strong multi-step performance. The findings highlight the value of adaptive, complexity-aware decomposition and bidirectional temporal modeling for wind speed prediction in heterogeneous landscapes, with implications for wind energy operations and grid integration.

Abstract

Wind flow can be highly unpredictable and can suffer substantial fluctuations in speed and direction due to the shape and height of hills, mountains, and valleys, making accurate wind speed (WS) forecasting essential in complex terrain. This paper presents a novel and adaptive model for short-term forecasting of WS. The paper's key contributions are as follows: (a) The Partial Auto Correlation Function (PACF) is utilised to minimise the dimension of the set of Intrinsic Mode Functions (IMF), hence reducing training time; (b) The sample entropy (SampEn) was used to calculate the complexity of the reduced set of IMFs. The proposed technique is adaptive since a specific Deep Learning (DL) model-feature combination was chosen based on complexity; (c) A novel bidirectional feature-LSTM framework for complicated IMFs has been suggested, resulting in improved forecasting accuracy; (d) The proposed model shows superior forecasting performance compared to the persistence, hybrid, Ensemble empirical mode decomposition (EEMD), and Variational Mode Decomposition (VMD)-based deep learning models. It has achieved the lowest variance in terms of forecasting accuracy between simple and complex terrain conditions 0.70%. Dimension reduction of IMF's and complexity-based model-feature selection helps reduce the training time by 68.77% and improve forecasting quality by 58.58% on average.
Paper Structure (17 sections, 5 equations, 7 figures, 7 tables)

This paper contains 17 sections, 5 equations, 7 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Background
  3. Sample entropy (SampEn)
  4. Ensemble Empirical Mode Decomposition (EEMD)
  5. Long Short-Term Memory (LSTM)
  6. Materials and methods
  7. Motivating dataset
  8. EEMD decomposition, IMF reduction and classification
  9. Data preprocessing for deep learning
  10. Proposed forecasting framework
  11. Evaluation of forecasting model
  12. Results and discussion
  13. Comparison with benchmark models
  14. Forecasting performance on multi-steps ahead WS prediction
  15. Season specific forecasting performance
  16. ...and 2 more sections

Figures (7)

  • Figure 1: EEMD decomposition and reduction for Gadhpati
  • Figure 2: Complex and simple IMF identification
  • Figure 3: The sliding window approach
  • Figure 6: The proposed model architecture
  • Figure 7: Deviation in nRMSE between simple and complex terrain
  • ...and 2 more figures