Application of WRF-Based Single-Point Data Artificial Intelligence Post-Processing Correction Method in Practical Short-Term Wind Speed and Power Forecasting

Analysis and application of forecasting models in wind power integration: A review of multi-step-ahead wind speed forecasting models

Abstract. Accurate prediction of wind speed is an important aspect of various tasks related to wind energy management such as wind turbine predictive control and wind power scheduling. The most typical characteristic of wind speed data is its persistent temporal variations. Most of the techniques reported in the literature for prediction of wind speed and power are based on statistical methods or probabilistic distribution of wind speed data. In this paper we demonstrate that deterministic forecasting methods can make accurate short-term predictions of wind speed using past data, at locations where the wind dynamics exhibit chaotic behaviour. The predictions are remarkably accurate up to 1 h with a normalised RMSE (root mean square error) of less than 0.02 and reasonably accurate up to 3 h with an error of less than 0.06. Repeated application of these methods at 234 different geographical locations for predicting wind speeds at 30-day intervals for 3 years reveals that the accuracy of prediction is more or less the same across all locations and time periods. Comparison of the results with f-ARIMA model predictions shows that the deterministic models with suitable parameters are capable of returning improved prediction accuracy and capturing the dynamical variations of the actual time series more faithfully. These methods are simple and computationally efficient and require only records of past data for making short-term wind speed forecasts within practically tolerable margin of errors.

The forecast of wind speed and the power produced from wind farms has been a challenge for a long time and continues to be so. This work introduces a method that we label as Wavelet Decomposition-Neural Networks (WDNN) that combines wavelet decomposition principles and deep learning. By merging the strengths of signal processing and machine learning, this approach aims to address the aforementioned challenge. Treating wind speed and power as signals, the wavelet decomposition part of the model transforms these inputs, as appropriate, into a set of features that the neural network part of the model can ingest to output accurate forecasts. WDNN is unconstrained by the shape, layout, or number of turbines in a wind farm. We test our WDNN methods using three large datasets, with multiple years of data and hundreds of turbines, and compare it against other state-of-the-art methods. It’s very short-term forecast, like 1-h ahead, can outperform some deep learning models by as much as 30%. This shows that wavelet decomposition and neural network are a potent combination for advancing the quality of short-term wind forecasting.

Turbine-specific short-term wind speed forecasting considering within-farm wind field dependencies and fluctuations

The growth of wind energy poses challenges to the integration of wind energy into the power grid. Within a wind farm, the conditions of local wind exhibit sizeable variations in very short term period and temporal wind speed patterns vary from turbine to turbine. Hence, short-term wind forecasting has been adopted to assist power system operations. In this work, we propose a wind plant-level short term wind speed and power forecasting methodology considering turbine contributions. The proposed model utilizes spatio-temporal dependencies and nonstationarity to accommodate the characteristics of wind farm data by using a novel regime-switching spatiotemporal generalized autoregressive conditional heteroscedasticity (RS-stGARCH) model. Case studies based on 2 years of data from a wind farm shows that the proposed RS-stGARCH method outperforms benchmark models by up to 21.10% for wind speed forecasting and up to 58.62% for the wind power forecasting.

This paper introduces Generalized Regression Neural Network (GRNN) for long term wind speed prediction of major wind power potential states in India. The performance of proposed GRNN model is evaluated using the publicly available online dataset of National Aeronautics and Space Administration (NASA). Data samples of 26 cities are used for training the generalized regression neural network and remaining 5 cities data samples are used for testing purpose. Air temperature, earth temperature, relative humidity, daily solar radiation, elevation, latitude, heating degree days, cooling degree days, frost days, longitude and atmospheric pressure are used as input variables. Mean square error between measured and forecasted wind speed using training data samples and testing data samples are found to be 0.000042279 and 0.1543. Here it is important to impart that the proposed GRNN model is trained and tested with data samples of different geographical locations in order to make it feasible for wind speed prediction of any other location. Wind power of prominent wind power potential states in India are predicted by a variable pitch and speed control wind turbine G80-2MW.

Wind power generated by wind has non-schedule nature due to stochastic nature of meteorological variable. Hence energy business and control of wind power generation requires prediction of wind speed (WS) from few seconds to different time steps in advance. To deal with prediction shortcomings, various WS prediction methods have been used. Predictive data mining offers variety of methods for WS predictions where artificial neural network (ANN) is one of the reliable and accurate methods. It is observed from the result of this study that ANN gives better accuracy in comparison conventional model. The accuracy of WS prediction models is found to be dependent on input parameters and architecture type algorithms utilized. So the selection of most relevant input parameters is important research area in WS predicton field. The objective of the paper is twofold: first extensive review of ANN for wind power and WS prediction is carried out. Discussion and analysis of feature selection using Relief Algorithm (RA) in WS prediction are considered for different Indian sites. RA identify atmospheric pressure, solar radiation and relative humidity are relevant input variables. Based on relevant input variables Cascade ANN model is developed and prediction accuracy is evaluated. It is found that root mean square error (RMSE) for comparison between predicted and measured WS for training and testing wind speed are found to be 1.44 m/s and 1.49 m/s respectively. The developed cascade ANN model can be used to predict wind speed for sites where there are not WS measuring instruments are installed in India.

ARMA based approaches for forecasting the tuple of wind speed and direction

Accurate prediction of wind turbine power is an important means to ensure the stable operation of wind turbines. The traditional wind speed-power curve only shows the relationship between wind turbine power and wind speed, without considering other influencing factors, such as rotor speed and pitch angle, and has certain limitations. Therefore, based on the supervisory control and data acquisition data of a wind farm, this paper proposes a hybrid multivariate prediction model that combines polynomial regression and random forest (RF) to predict wind turbine output power. First, to reduce the difficulty of expert analysis, the maximum information coefficient is used to analyze the correlation between data and select features. Wind speed, rotor speed, pitch angle, and wind direction are considered important and chosen for power prediction modeling. Then the relationship between multiple variables and output power is established through polynomial regression. Finally, the new polynomial features and output power are used to train the RF model to predict the output power. The experimental results show that the predicted wind speed and power diagram shows a band-like distribution, which matches closely with the real one; compared with multiple models such as a single RF model, a polynomial regression model, and a feedforward artificial neural network model, the polynomial-RF model has the highest prediction accuracy. The mean absolute percentage error (MAPE) of the polynomial-RF model prediction is 0.06, and the root mean square error (RMSE) is 114. Compared with the single RF model, the MAPE predicted by the polynomial-RF model is reduced by 14%, and the RMSE is reduced by 6%. It can be seen that the polynomial-RF model proposed in this paper has a good application prospect in the prediction of wind turbine output power.

Introduction: Wind energy is a kind of clean energy, and wind power generation is an important means of utilizing wind energy. However, wind power generation is highly volatile and prone to causing grid operation accidents, so it is necessary to predict wind power generation. The annual power generation of wind turbines is determined by the Turbine Power Curve (TPC), which, however, is easily affected by different meteorological conditions. The successful deployment of wind turbines requires accurate predictions of wind farm power before construction and near-real-time power predictions after construction to facilitate grid uptake. However, the existing research methods have not considered the quantitative influence of climate characteristics on the prediction accuracy of wind power, limiting the accuracy of wind power generation predictions. In order to solve this problem, this investigation applies a machine learning model that is based on decision trees to decrease the ambiguity associated with wind resource evaluations and to enhance the precision of wind energy forecasts. Method: The machine learning model, based on decision trees, was trained on four distinct classifications of vertical wind profiles to depict wind velocities necessitating complex computations across different rotor layer altitudes. Results: The findings indicated that the model after integration of rotor-equivalent wind speed and temperature lapse rate achieved a 21.48% enhancement in predictive accuracy for the dataset in question, surpassing traditional power curve techniques. Discussion: The model scrutinized the utility of incorporating parameters such as the wind speed at hub height, the wind speed equivalent to the rotor, and the rate of direct temperature decrease as variables for predicting power output. Climate feature data were also utilized to train the regression tree model, enabling the correlation of wind power with wind profile and climate features for predicting wind power based on physical relationships. Conclusion: This methodology has emerged as the optimal strategy for power forecasting across all categorized vertical wind profile types. Notably, the model incorporating the temperature lapse rate in the prediction exhibited higher accuracy than the other one, highlighting the significance of considering climatic characteristics in wind power prediction.

Wind power predictions from nowcasts to 4-hour forecasts: A learning approach with variable selection

Locally recurrent neural networks for long-term wind speed and power prediction

Wind speed forecasting using nonlinear-learning ensemble of deep learning time series prediction and extremal optimization

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

<p>The amount of wind farms and wind power production in Europe, both on- and off-shore, increased rapidly in the past years. To ensure grid stability, on-time (re)scheduling of maintenance tasks and mitigate fees in energy trading, accurate predictions of wind speed and wind power are needed. It has become particularly important to improve wind speed predictions in the short range of one to six hours as wind speed variability in this range has been found to pose the largest operational challenges. Furthermore, accurate predictions of extreme wind events are of high importance to wind farm operators as timely knowledge of these can both prevent damages and offer economic preparedness. We propose in this work a deep convolutional recurrent neural network (RNN) based regression model for the spatio-temporal prediction of extreme wind speed events over Europe in the short-to-medium range (12 hour lead-time in 1 hour intervals). This is achieved by training a multi-layered convolutional long short-term memory (ConvLSTM) network with imbalanced regression loss, to which end we investigate three different loss functions: the inversely weighted mean absolute error (W-MAE) loss, the inversely weighted mean squared error (W-MSE) loss and the squared error-relevance area (SERA) loss. </p><p>The results indicate superior performance of the SERA loss, showing significant improvements on high intensity extreme events. The W-MAE and W-MSE shows no improvements over the standard MSE loss and we thus discourage the usage of the inverse weighting method. We conclude that the SERA loss provides an effective way to adapt deep learning to the task of imbalanced spatio-temporal regression and its application to the forecasting of extreme wind events in the short-to-medium range. </p><p> This work was performed as a part of the MEDEA project, which is funded by the Austrian Climate Research Program to further research on renewable energy and meteorologically induced extreme events.</p>