Handling Multi-scale Data via Multi-target Learning for Wind Speed Forecasting

Ultra-short-term wind speed forecasting based on EMD-VAR model and spatial correlation

Short-term wind speed forecasting based on the Jaya-SVM model

Accurate wind speed forecasting is essential for the reliability and security of the power system, and optimal operation and management of wind integrated smart grids. However, it is still a challenging task due to the highly uncertain and volatile nature of wind speed. Accordingly, in this work, a novel deep learning-based model integrating the discrete wavelet packet transform (DWPT) and bidirectional long short-term memory (BLSTM) is developed to precisely capture deep temporal features and learn the time-varying relationship of wind speed time series. In the proposed method, by applying the DWPT, both approximations and details parts are decomposed by passing through the filters to choose the frequency band related to the features of the original signal more adaptively. The BLSTM networks are incorporated to deal with the uncertainties more effectively as they have bidirectional memory capability (feedforward and feedback loops) to investigate both previous and future hidden layers data. To simultaneously improve the forecasting performance and decrease the learning complexity, the reconstructed state space of historical wind data is employed to reflect the evolution laws of wind speed. Two case studies using real-world wind speed datasets gathered from Flatirons campus (M2) of National Renewable Energy Laboratory (NREL) located in Colorado, USA and weather station of Edmonton, Canada are implemented to demonstrate the effectiveness and superiority of the proposed hybrid method compared to the shallow architectures and state-of-the-art deep learning models in the recent literature.

ABSTRACTWind resources are becoming increasingly significant due to their clean and renewable characteristics, and the integration of wind power into existing electricity systems is imminent. To maintain a stable power supply system that takes into account the stochastic nature of wind speed, accurate wind speed forecasting is pivotal. However, no single model can be applied to all cases. Recent studies show that wind speed forecasting errors are approximately 25% to 40% in Chinese wind farms. Presently, hybrid wind speed forecasting models are widely used and have been verified to perform better than conventional single forecasting models, not only in short-term wind speed forecasting but also in long-term forecasting. In this paper, a hybrid forecasting model is developed, the Similar Coefficient Sum (SCS) and Hermite Interpolation are exploited to process the original wind speed data, and the SVM model whose parameters are tuned by an artificial intelligence model is built to make forecast. The results of case studies show that the MAPE value of the hybrid model varies from 22.96% to 28.87 %, and the MAE value varies from 0.47 m/s to 1.30 m/s. Generally, Sign test, Wilcoxon’s Signed-Rank test, and Morgan--Granger--Newbold test tell us that the proposed model is different from the compared models.

Accurate forecasting of wind speed plays a fundamental role in enabling reliable operation and planning for large‐scale integration of wind turbines. It is difficult to obtain the accurate wind speed forecasting (WSF) due to the intermittent and random nature of wind energy. In this paper, a multiperiod‐ahead WSF model based on the analysis of variance, stacked denoising autoencoder (SDAE), and ensemble learning is proposed. The analysis of variance classifies the training samples into different categories. The stacked denoising autoencoder as a deep learning architecture is later built for unsupervised feature learning in each category. The ensemble of extreme learning machine (ELM) is applied to fine‐tune the SDAE for multiperiod‐ahead wind speed forecasting. Experimental results are made to demonstrate that the proposed model has the best performance compared with the classic WSF methods including the single SDAE‐ELM, ELMAN, and adaptive neuron‐fuzzy inference system (ANFIS).

Multi-step-ahead wind speed forecasting based on a hybrid decomposition method and temporal convolutional networks

The Croatian Meteorological and Hydrological Service (DHMZ) has a long experience in post-processing wind speed and wind gust variables. One of the most used methods at DHMZ is the analog method. Although notable improvements in wind speed forecasts have been observed, one of the disadvantages of the analog method is its accuracy in forecasting high wind speed episodes, especially when using an analog ensemble for deterministic post-processing (HRAN). Since onshore parts of Croatia regularly experience violent winds several times a year, leading to disruptions in electricity production and traffic, accurate high wind speed forecasts hold significant importance.  This study investigates the impact of reducing the analog ensemble size on HRAN accuracy. Analysing wind speed forecasts from onshore locations in Croatia, we compare raw model output and HRAN with predictor weight optimisation. In the analysis, we use ALADIN-HR NWP, tuned for wind across Croatian territory, providing us with forecasts 72 hours ahead. Utilising measures for continuous forecast verification, the overall forecast quality is inspected, while the categorical approach is used for examining forecast performances for more extreme events. Generally, the highest enhancements are achieved with an analog ensemble size of 15 members. Since high winds are rare, compared to low and moderate winds, employing larger ensembles often results in overall error reduction. However, our study shows how fewer than 15 ensemble members can provide more favourable results for the highest wind categories, causing less pronounced underestimations. Thus, varying ensemble size might be an optimal way to address these issues. In the ongoing work, besides wind speed forecasts, the analog method is also used to improve visibility forecasts, focusing on low-visibility events often caused by fog.

Many renewable energy resources, including wind energy, are uncertain and often unavailable when needed, with high variability and dependency on atmospheric and climatic conditions. Variability and uncertainty of wind energy follow those wind speed and occur at multiple timescales; that is, from seconds to minutes and then to hours. They also require movement of other resources to ensure the balance between generation and load. This requires an accurate wind speed prediction. However, most existing wind speed forecasting models are based on data that do not take into account the difference between day and night, which can limit the accuracy of wind speed forecasting. Therefore, to improve the prediction accuracy, this study proposes three scenarios for wind speed prediction: first, the prediction model does not take into account the difference between day and night. The second model uses only data recorded during the day, and the third model deals with data measured during the night only. A comparative analysis and comprehensive evaluation will be given at the end of this work to verify the proposed hypothesis. The recorded climate data and measured wind speed at AUMET station will be used in this study.

With the large-scale development of wind energy, wind power forecasting plays a key role in power dispatching in the electric power grid, as well as in the operation and maintenance of wind farms. The most important technology for wind power forecasting is forecasting wind speed. The current mainstream methods for wind speed forecasting involve the combination of mesoscale numerical meteorological models with a post-processing system. Our work uses the WRF model to obtain the numerical weather forecast and the gradient boosting decision tree (GBDT) algorithm to improve the near-surface wind speed post-processing results of the numerical weather model. We calculate the feature importance of GBDT in order to find out which feature most affects the post-processing wind speed results. The results show that, after using about 300 features at different height and pressure layers, the GBDT algorithm can output more accurate wind speed forecasts than the original WRF results and other post-processing models like decision tree regression (DTR) and multi-layer perceptron regression (MLPR). Using GBDT, the root mean square error (RMSE) of wind speed can be reduced from 2.7–3.5 m/s in the original WRF result by 1–1.5 m/s, which is better than DTR and MLPR. While the index of agreement (IA) can be improved by 0.10–0.20, correlation coefficient be improved by 0.10–0.18, Nash–Sutcliffe efficiency coefficient (NSE) be improved by −0.06–0.6. It also can be found that the feature which most affects the GBDT results is the near-surface wind speed. Other variables, such as forecast month, forecast time, and temperature, also affect the GBDT results.

Accurate and consistent wind speed forecasting is vital for efficient energy management and the market economy. Wind speed is non-linear, non-stationary, and irregular, so it is very difficult to forecast. There are many forecasting methods currently in use; however, selecting and developing the most appropriate method for a particular region in wind speed forecasting is still a hot topic. This study presents a new and unique neural network-based ensemble system for forecasting wind speed, which is very difficult to predict but is directly related to the power generated by wind farms for individual and different sites. With the developed ensemble model, average mean absolute error, mean absolute percentage error and root mean square error values are obtained as 0.1269, 3.074%, 0.1596 respectively. Test results demonstrate significant contributions of the proposed system compared to existing statistical, heuristic and ensemble models, indicating that the developed model is a promising alternative for wind speed forecasting models. The obtained results show that this system is an effective and useful intelligent tool that can be used by various companies and government facilities that invest and operate in intelligent wind energy technologies.

Multi-step ahead wind speed forecasting using an improved wavelet neural network combining variational mode decomposition and phase space reconstruction

Short-term wind speed forecasting bias correction in the Hangzhou area of China based on a machine learning model

A novel ensemble system for short-term wind speed forecasting based on hybrid decomposition approach and artificial intelligence models optimized by self-attention mechanism

As the integration of wind power into electrical energy network increasing, accurate forecast of wind speed becomes highly important in the case of large-scale wind power connected into the grid. In order to improve the accuracy of wind speed forecast and the generalization ability of the model, Extreme Gradient Boosting (XGBoost) as an improvement from gradient boosting decision tree (GBDT) is trained and deployed in the cheaper central processing unit (CPU) devices instead of graphics processing unit (GPU) devices, thus, a wind speed forecast model based on Extreme Gradient Boosting is proposed in this paper. Firstly, the historical data is taken as a part of the input vectors for the model. Moreover, considering the monthly change of wind speed characteristics, the dataset of wind power is divided into four parts by month so that the models are constructed in different complexity by month. Finally, compared with back propagation neural networks (BPNN) and linear regression (LR) models, the experimental results show that the improved XGBoost model can promote the forecast accuracy effectively.

Wind energy has gained significant attention as a clean, renewable resource due to fossil fuels' environmental impact. Accurate wind speed forecasting is essential to address variability and intermittency challenges. Current forecasting difficulties arise from wind speed's high susceptibility to meteorological conditions. This study proposes a GA-based ensemble framework that combines forecasting models using genetic algorithms. We systematically compared 14 models: linear models (AR, ARMA), advanced neural networks (MLP, RBF), hybrid models, and ensembles. Models were evaluated using minute-by-minute data from five major Brazilian cities: Brasília, Florianópolis, Petrolina, Natal, and São Luís. Key findings include: I) Superior Performance: The proposed framework achieved MSE values from 0.0802 to 0.9020 and MAE values from 0.1970 to 0.6140 across all datasets; II) Robust Prediction: values ranged from 0.7139 to 0.8723, demonstrating strong predictive capability; III) Statistical Validation: Friedman test ( ) confirmed significant differences with perfect rank stability across all locations; IV) High Scalability: Runtimes ranged from 58,077.3 to 77,815.7 s, determined by the base model combination; and V) Computational Efficiency: One-step-ahead forecasting requires only 0.0003 s for weighting and combination.