Wind Power Forecasting Using Attention-Based Recurrent Neural Networks: A Comparative Study

With the rising demand for environmental protection, wind energy has made considerable progress as a clean energy source. Due to the uncertainty of wind power generation, accurate forecasting of wind power is conducive to achieving stable grid-connected control of wind turbines. This paper proposes an ultra-short-term wind power forecasting model composed of variational modal decomposition (VMD) and long short-term memory (LSTM) networks. Obtain the recorded historical wind speed and historical power from the wind field data acquisition and monitoring control system (SCADA), and use VMD to decompose and reconstruct the historical wind speed of the wind field to form 5 training data sets. Using historical power, historical wind speed, and predicted wind speed as input, wind power forecasts for the next 4 hours (15 minutes apart) are made. In order to evaluate the prediction ability of the proposed model, two algorithms, LSTM and XGBoost, are used for prediction. Taking root mean square error (RMSE) and mean absolute error (MAE) as evaluation indicators, the results show that the prediction accuracy of wind power is significantly improved after VMD processing, and the performance of LSTM is better than XGBoost in the two evaluation indicators.

Due to the uncertainty and volatility of wind energy resources, its large-scale consumption in power gird needs to be based on accurate prediction of output. This puts high demands on the integrity and accuracy of historical wind power data. However, in many wind farms, data loss due to equipment failure or human factors is common, which has a negative impact on wind power forecasting. In this paper, a Long Short-Term Memory (LSTM) strategy is incorporated in the recurrent neural network (RNN) to set up a prediction model and fill the wind power missing data, which behaves better than the traditional RNN methods. The case of this paper uses the historical wind power data of Liaoning Province, which obtains the ideal results, proving the validity of the proposed model and method.

Very short-term forecasting of wind power generation using hybrid deep learning model

In this work, an integrated wind power forecasting method has been carried out to analyze the spatial dynamics of wind power in a micro-scale wind farm, located in a gully region of the loess plateau, by integrating computational fluid mechanics (CFD) model and Kalman filter method. The CFD model was used to simulate the wind-speed distribution characteristics for obtaining six different wind zones. Then, we corrected the wind speed and wind power in each wind zone at periods I and II, respectively, using Kalman filter and theoretical wind power curve, with an hourly forecasting product of BJ-RUC system during in June 2014. Furthermore, we assessed the effects of correcting results and actual outputs in wind speed and wind power separately, for acquiring the uncertainties' results of the integrated model. The capacity factor and curtailment rate of wind power generation were also assessed to reflect the potential economic benefit under current conditions in the wind farm. The result showed that the integrated model used in the study could improve the wind power forecast accuracy to 97.55% and 80.79% at periods I and II, respectively. The integrated method in the case study can be used to optimize the forecast on wind speed and wind power in similar wind farms, which is helpful for the local managers to make decisions on sustainable wind power planning.

Online distributed learning in wind power forecasting

The electrical energy demand is growing every day. Fossil fuel-based electrical power generation pollutes the environment. So, to fulfil the electrical energy demand, clean renewable energy sources like solar and wind are emerging as viable alternatives. But the amount of power generated from solar and wind energy sources is varying in magnitude and is intermittent in nature. So, predicting solar and wind power generation is indispensable for efficient planning and controlling of a renewable energy system. This paper aims to identify a machine learning model suitable for time-series prediction of wind power and also surveys the predominantly existing machine learning followed by the deep learning approaches. The main objective of this study is to solve a real-life problem in the renewable energy sector by accurately estimating the amount of power generation production per hour by applying machine learning techniques using historical wind power energy production data. Hence, historical wind data set with input parameters like wind speed, wind direction air temperature, pressure, and output parameters of wind power generation is considered for training and testing using mean absolute error for evaluating the prediction accuracy. Layers from simple to complex deep layers are trained and tested for prediction accuracy, i.e. baseline, linear, dense, multidense, convolution neural networks, and long short-term memory. Investigating further, LSTM and Residual LSTM prove to have higher mean absolute prediction accuracy of 0.0987 and 0.0958. This eventually enlightens the wind power forecasting models and new paths for further research applications.

Objective: The estimation accuracy of wind power is an important subject of concern for reliable grid operations and taking part in open access. So, with an objective to improve the wind power forecasting accuracy. Methods: This article presents Wavelet Transform (WT) based General Regression Neural Network (GRNN) with statistical time series input selection technique. Results: The results of the proposed model are compared with four different models namely naïve benchmark model, feed forward neural networks, recurrent neural networks and GRNN on the basis of Mean Absolute Error (MAE) and Mean Absolute Percentage Error (MAPE) performance metric. Conclusion: The historical data used by the presented models has been collected from the Ontario Electricity Market for the year 2011 to 2015 and tested for a long time period of more than two years (28 months) from November 2012 to February 2015 with one month estimation moving window.

The fluctuation and uncertainty of wind power generation bring severe challenges to secure and economic operation of power systems. Because wind power forecasting error is unavoidable, probabilistic forecasting becomes critical to accurately quantifying the uncertainty involved in traditional point forecasts of wind power and to providing meaningful information to conduct risk management in power system operation. This paper proposes a novel direct quantile regression approach to efficiently generate nonparametric probabilistic forecasting of wind power generation combining extreme learning machine and quantile regression. Quantiles with different proportions can be directly produced via an innovatively formulated linear programming optimization model, without dependency on point forecasts. Multistep probabilistic forecasting of 10-min wind power is newly carried out based on real wind farm data from Bornholm Island in Denmark. The superiority of the proposed approach is verified through comparisons with other well-established benchmarks. The proposed approach forms a new artificial neural network-based nonparametric forecasting framework for wind power with high efficiency, reliability, and flexibility, which can be beneficial to various decision-making activities in power systems.

Evaluating the impact of Numerical Weather Prediction variables on wind power forecasting: A case study of the Alpha Ventus offshore wind farm

The accuracy of long-term wind power forecasting (WPF) is crucial for the efficient operation of grid systems. However, wind power generation is highly stochastic and intermittent due to the influence of weather, which makes long-term WPF less effective. Numerical weather prediction (NWP) data contains valuable weather forecast information, which can mitigate the negative effects of stochastic weather fluctuations on WPF. However, the accuracy of NWP data decreases over time, and multiple NWP data can have redundancy and errors that make it challenging to extract valid information. Reducing the variety and errors in NWP data and using more effective information extraction methods are essential for improving long-term WPF performance. In this article, we propose a novel long-term WPF hybrid model that corrects NWP wind speed and uses multi-scale deep learning regression prediction to exclude excessive NWP data. We use only the corrected NWP wind speed data to establish a nonlinear mapping relationship with actual power data. The validation case study shows that our proposed model reduces the mean squared error (MSE) and mean absolute error (MAE) by 65.0% and 43.8%, respectively, compared to the current state-of-the-art time series forecasting model in a seven-day forecasting scenario.

A CNN encoder decoder LSTM model for sustainable wind power predictive analytics

Wind power forecast plays an important role in wind farm operation and grid scheduling decisions. Prediction of power output from wind vector based on wind farm equivalent power curve is the final step of forecast, which has a lot of interaction with the grid dispatching, and it has a particularly prominent influence on the wind farm power output so as to cut down prediction accuracy. Focusing on the step of "wind-power transformation", a correction method of day-ahead wind power forecast including multiple factors such as wind-abandon coefficient is proposed in this paper. Firstly, it is proved that the wind power prediction error is affected by a variety of factors, such as the amplitude of predicted power, the volatility of forecast as well as the stability of recent wind power output. And a model is built to quantify them respectively. Then the wind-abandon coefficient is introduced to modeling the effect of wind power curtailment activities of grid dispatching to forecast accuracy. Finally, the evaluation model of wind power forecasting error is built by using the multiple linear regression method and corrects the predictive value. Case study basing on the actual operation data of a wind farm in northern China has shown that the proposed method is effective to improve the accuracy level of wind power forecast.

In accordance with the topographic features and other relevant factors, data affecting wind power generation in engineering practice may be difficult to obtain, and three short-term wind power forecasting methods are proposed based on incomplete data. Most wind power forecasting studies are based on wind speed data, but wind power forecasting needs more influential factors in addition to wind speed data, and it is difficult to obtain these data in actual production and life. Therefore, this paper proposes the wind power forecasting under the condition of incomplete data. First, this paper makes theoretical analysis on multi-state space Markov chain wind power forecasting, backpropagation (BP) neural network wind power forecasting, and genetic algorithm (GA)-BP neural network wind power forecasting, and the corresponding wind power forecasting models are constructed. Second, with the actually measured data of a wind farm in the Heilongjiang province as the research object, the historical data are preprocessed first and then imported into three wind power forecasting models for simulation, and the curves of relative error and absolute error of total wind power forecasting in this area are obtained. Finally, the simulation results of three methods based on incomplete data are analyzed and evaluated. The results show that under the condition of incomplete data, the maximum error of the GA-BP neural network wind power forecasting model improved by the genetic algorithm is reduced from 6.8% to 1.6%, and the forecasting accuracy is greatly improved.

Accurate information regarding the uncertainty of short-term forecast for aggregate wind power is a key to efficient and cost effective integration of wind farms into power systems. This paper presents a methodology for producing wind power forecast scenarios. Using historical wind power time series data and the Kernel Density Estimator (KDE), probabilistic wind power forecast scenarios were generated according to a rolling process. The improvement achieved in the accuracy of forecasts through frequent updating of the forecasts taking into account the latest realized wind power was quantified. The forecast scenarios produced by the proposed method were used as inputs to a unit commitment and optimal dispatch model in order to investigate how the uncertainty in wind forecast affect the operation of power system and in particular gas-fired generators.

A new intelligent method based on combination of VMD and ELM for short term wind power forecasting