Short-term Wind Speed Forecasting Research Articles

Prediction interval forecasting of wind speed and wind power using modes decomposition based low rank multi-kernel ridge regression

Abstract In this paper a new hybrid method combining variational mode decomposition (VMD) and low rank Multi-kernel ridge regression (MKRR) is presented for direct and effective construction of prediction intervals (PIs) for short-term forecasting of wind speed and wind power. The original time series signals are decomposed using VMD approach to prevent the mutual effects among the different modes. The proposed VMD-MKRR method is used to construct the PIs with different confidence levels of 95%, 90% and 85% for wind speed and wind power of two wind farms which are located in the state of Wyoming, USA for time intervals of 10 min, 30 min and 1 h and in the state of California for time interval of 1 h respectively. Comparison with empirical mode decomposition (EMD) based low rank kernel ridge regression is also presented in the paper to validate the superiority of the VMD based wind speed and wind power model. Further to enhance the proposed model performance their parameters are optimized using Mutated Firefly Algorithm with Global optima concept (MFAGO).

A Hybrid Model Based on A Modified Optimization Algorithm and An Artificial Intelligence Algorithm for Short-Term Wind Speed Multi-Step Ahead Forecasting

In the last few years, researchers have paid increasing attention to improving the accuracy of wind speed forecasting because of its vital impact on power dispatching and grid security. However, it is difficult to achieve a good forecasting performance due to the randomness and intermittency characteristics of wind speed time series. Current forecasting models based on neural network theory could adapt to various types of time series data; however, these models ignore the importance of data pre-processing and model parameter optimization, which leads to poor forecasting accuracy. In this paper, a new hybrid model is developed for short-term multi-step wind speed forecasting, which includes four modules: (1) the data pre-processing module; (2) the optimization module; (3) the hybrid nonlinear forecasting module and (4) the evaluation module. In order to estimate the forecasting ability of the proposed hybrid model, 10 min wind speed data were applied in this paper as a case study. The experimental results in six real forecasting cases indicate that the proposed hybrid model can provide not only accurate but also stable performance in terms of multi-step wind speed forecasting can be considered an effective tool in planning and dispatching for smart grids.

Research and application of an innovative combined model based on a modified optimization algorithm for wind speed forecasting

Wind energy is emerging as a promising substitute for conventional energy and plays a pivotal role in the power industry. For wind speed forecasting, many challenges have exposed due to its fluctuation and intermittence. To address these difficulties, different models have been adopted to various wind speed time series in previous studies. However, few methodologies have focused on the importance of model parameter optimization or data pre-processing, resulting in undesirable forecasting performance. In this study, an innovative combined model that combines data pre-processing, modified optimization algorithms, three neural networks and an effective deciding weight method is proposed for short-term wind speed forecasting. To improve the forecasting capacity of the combined model, a modified optimization algorithm is proposed and employed to determine the parameters of the single models. Furthermore, a deciding weight method based on multivariate statistical estimation is applied for weight optimization. Additionally, ten-minute wind speed data from a wind farm in Penglai, China, are selected for multi-step ahead forecasting. The results obtained confirmed an adequate approximation of the actual wind speed series and a significant improvement of the forecasting accuracy of the proposed model.

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

As an essential issue in wind energy industry, wind speed forecasting plays a vital role in optimal scheduling and control of wind energy generation and conversion. In this paper, a novel method called EnsemLSTM is proposed by using nonlinear-learning ensemble of deep learning time series prediction based on LSTMs (Long Short Term Memory neural networks), SVRM (support vector regression machine) and EO (extremal optimization algorithm). First, in order to avert the drawback of weak generalization capability and robustness of a single deep learning approach when facing diversiform data, a cluster of LSTMs with diverse hidden layers and neurons are employed to explore and exploit the implicit information of wind speed time series. Then predictions of LSTMs are aggregated into a nonlinear-learning regression top-layer composed of SVRM and the EO is introduced to optimize the parameters of the top-layer. Lastly, the final ensemble prediction for wind speed is given by the fine-turning top-layer. The proposed EnsemLSTM is applied on two case studies data collected from a wind farm in Inner Mongolia, China, to perform ten-minute ahead utmost short term wind speed forecasting and one-hour ahead short term wind speed forecasting. Statistical tests of experimental results compared with other popular prediction models demonstrated the proposed EnsemLSTM can achieve a better forecasting performance.

Short-term wind speed prediction using an extreme learning machine model with error correction

Wind speed forecasting is an important technology in the wind power field; however, because of their chaotic nature, predicting wind speeds accurately is difficult. Aims at this challenge, a new hybrid model is proposed for short-term wind speed forecasting, where the short-term forecasting period is ten minutes. The model combines extreme learning machine with improved complementary ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) and autoregressive integrated moving average (ARIMA). The extreme learning machine model is employed to obtain short-term wind speed predictions, while the autoregressive model is used to determine the best input variables. An ensemble method is used to improve the robustness of the extreme learning machine. To improve the prediction accuracy, the ICEEMDAN-ARIMA method is developed to postprocess the errors; this method can also be used to preprocess original wind speed. Additionally, this paper reports the results of a comparative study on preprocessing and postprocessing time series data. Three experimental results show that: (1) the error correction is effective in decreasing the prediction error, and the proposed models with error correction are suitable for short-term wind speed forecasting; (2) the ICEEMDAN method is more powerful than other variants of empirical mode decomposition in performing non-stationary decomposition, and the ICEEMDAN-ARIMA method achieves satisfactory performance both for preprocessing and postprocessing; and (3) for prediction, the preprocessing of time series is more effective than its postprocessing.

Data mining-assisted short-term wind speed forecasting by wavelet packet decomposition and Elman neural network

On the basis of data-mining technology, a hybrid method of short-term wind speed forecast is proposed by the wavelet packet decomposition, density-based spatial clustering of applications with noise, and the Elman neural network (WPD-DBSCAN-ENN). First, the WPD is applied to decompose a raw wind speed series into several sub-series. The gradient boosted regression trees (GBRT) algorithm is then applied to determine the structure of the ENNs for each sub-wind series. Next, the training dataset is clustered by the DBSCAN to select the representative data for the ENNs. A key parameter in the DBSCAN is chosen through a new method. Finally, the wind speed forecast is conducted by the ENNs. Case studies are adopted to validate the accuracy of the hybrid methods. The results are compared with those obtained using the WPD-ENN hybrid method and a single ENN via four general error criteria. The performance of the WPD-DBSCAN-ENN hybrid method outperformed those of the other methods indicated above.

Spatio-temporal asymmetry of local wind fields and its impact on short-term wind forecasting.

The massive amounts of spatio-temporal data collected in today's wind farms have created a necessity for accurate spatio-temporal models. Despite the growing recognition for non-separable spatio-temporal models, a significant reliance on separable, symmetric models is still the norm in today's renewable industry. We discover that the broad use of separable models is due to the handling of wind data in a setting that does not reveal their fine-scale spatio-temporal structure. The contribution of this research is two-fold. First, we devise a special pair of spatio-temporal "lens" that allows us to see the fine-scale spatio-temporal variations and interactions, and subsequently, we conclude that local wind fields exhibit strong signs of non-separability and asymmetry. Using one year of turbine-specific wind measurements, we show that asymmetry can in fact be detected in more than 93% of the time. Second, making use of the spatio-temporal lens, we propose an enhanced procedure for short-term wind speed forecast. Substantial improvements in forecast accuracy in both wind speed and wind power were observed. When combined with certain intelligent methods such as support vector machine, additional improvements are possible.

Wind Power Forecasting

Abstract Accurate short-term wind power forecast is very important for reliable and efficient operation of power systems with high wind power penetration. There are many conventional and artificial intelligence methods that have been developed to achieve accurate wind power forecasting. Time-series based algorithms are known to be simple, robust, and have been used in the past for forecasting with some level of success. Recently some researchers have advocated for artificial-intelligence based methods such as Artificial Neural Networks (ANNs), Fuzzy Logic, etc., for forecasting because of their flexibility. This paper presents a comparison of conventional and two artificial intelligence methods for wind power forecasting. The conventional method discussed in this paper is the Autoregressive Moving Average (ARMA) which is one of the most robust and simple time-series methods. The artificial intelligence methods are Artificial Neural Networks (ANNs) and Adaptive Neuro-fuzzy Inference Systems (ANFIS). Simulation results for very-short-term and short-term forecasting show that ANNs and ANFIS are suitable for the very-short-term (10 minutes ahead) wind speed and power forecasting, and the ARMA is suitable for the short-term (1 hour ahead) wind speed and power forecasting.

Wavelet Neural Network Based Multiobjective Interval Prediction for Short-Term Wind Speed

As a source of clean and pollution-free renewable energy, wind power has attracted much attention and has been increasingly integrated into the existing power system. However, the uncertain and volatile wind speed makes the utilization of wind power a challenging task. Hence, it is essential to design an accurate forecast method to deal with the uncertainty caused by wind speed. This paper proposes a multiobjective interval prediction method based on wavelet neural network (WNN) for short-term wind speed forecast. This method can generate a set of Pareto optimal solutions which represents a set of prediction models that can directly construct the prediction intervals. An advanced multiobjective evolutionary algorithm, preference inspired co-evolutionary algorithm using goal vectors, is investigated to train the WNN model. Two case studies are carried out with real wind speed data of Victoria and Edmonton in Canada to justify the effectiveness of the proposed method. The numerical results also show the superiority of the proposed forecast approach compared with some benchmark methods.

Short-term wind speed forecasting based on fast ensemble empirical mode decomposition, phase space reconstruction, sample entropy and improved back-propagation neural network

With the sharp consumption of fossil energy and deterioration of ecological environment, wind power, as a clean and renewable energy resource, gains more and more attention from all over the world. However, due to the intermittency and stochastic nature of wind speed, an accurate wind power/speed forecasting is not only crucial for availably dispatching the wind power resource but also has a direct relationship with the State Grid operation safely and steadily. In this paper, an innovative hybrid wind speed forecasting model, including fast ensemble empirical mode decomposition, sample entropy, phase space reconstruction and back-propagation neural network with two hidden layers, is proposed to enhance the accuracy of wind speed prediction. The data is firstly preprocessed by fast ensemble empirical mode decomposition and sample entropy. Subsequently, the prediction model called improved back-propagation neural network is built to forecast the sub-series, whose inputs and outputs are obtained in accordance to phase space reconstruction. To verify the effectiveness and advancement of the proposed model, the paper chooses the real data from two wind farms located in different site in China for experiments. Comparing with the benchmark models, the proposed model shows a better performance in short-term wind speed forecasting.

Rough Deep Neural Architecture for Short-Term Wind Speed Forecasting

Accurate wind speed forecasting is a fundamental requirement for large-scale integration of wind power generation. However, the intermittent and stochastic nature of wind speed makes this task challenging. Artificial neural networks (ANNs) are widely used in this area; however, they may fail to provide the accuracy that may be required. This is due to applying shallow architectures with error-prone hand-engineered features. This paper proposes a deep neural network (DNN) architecture with stacked autoencoder (SAE) and stacked denoising autoencoder (SDAE) for ultrashort-term and short-term wind speed forecasting. Autoencoders (AEs) are applied for unsupervised feature learning from the unlabeled wind data and a supervised regression layer is applied at the top of the AEs for wind speed forecasting. Several uncertain factors exist in the wind data that degrade the accuracy of current methodologies. In order to improve the accuracy, rough neural networks are incorporated in the proposed deep learning models to develop novel rough extensions of SAE and SDAE that are robust to wind uncertainties. Experimental results show that the proposed rough DNN models outperform classic DNNs and previous models that apply shallow architectures in the view of lower RMSE and mean absolute error measurements.

GSI-Based, Continuously Cycled, Dual-Resolution Hybrid Ensemble–Variational Data Assimilation System for HWRF: System Description and Experiments with Edouard (2014)

A Gridpoint Statistical Interpolation analysis system (GSI)-based, continuously cycled, dual-resolution hybrid ensemble Kalman filter–variational (EnKF-Var) data assimilation (DA) system is developed for the Hurricane Weather Research and Forecasting (HWRF) Model. In this system, a directed moving nest strategy is developed to solve the issue of nonoverlapped domains for cycled ensemble DA. Additionally, both dual-resolution and four-dimensional ensemble–variational (4DEnVar) capabilities are implemented. Vortex modification (VM) and relocation (VR) are used in addition to hybrid DA. Several scientific questions are addressed using the new system for Hurricane Edouard (2014). It is found that dual-resolution hybrid DA improves the analyzed storm structure and short-term maximum wind speed (Vmax) and minimum sea level pressure (MSLP) forecasts compared to coarser, single-resolution hybrid DA, but track and radius of maximum wind (RMW) forecasts do not improve. Additionally, applying VR and VM on the control background before DA improves the analyzed storm, overall track, RMW, MSLP, and Vmax forecasts. Further applying VR on the ensemble background improves the analyzed storm and forecast biases for MSLP and Vmax. Also, using 4DEnVar to assimilate tail Doppler radar (TDR) data improves the analyzed storm and short-term MSLP and Vmax forecasts compared to three-dimensional ensemble–variational (3DEnVar) although 4DEnVar slightly degrades the track forecast. Finally, the new system improves overall RMW, MSLP, and Vmax forecasts upon the operational HWRF, while no improvement on track is found. The intensity forecast improvement during the intensifying period is due to the better analyzed structures for an intensifying storm.

Composite quantile regression extreme learning machine with feature selection for short-term wind speed forecasting: A new approach

With the growing wind penetration of wind resources into power generation worldwide, accurate and comprehensive wind speed forecasting (WSF) is becoming increasingly significant to ensure continuously economical and reliable power system operations. In this paper, a novel WSF framework based on composite quantile regression outlier-robust extreme learning machine (CQR-ORELM) with feature selection and parameter optimization using a hybrid population-based algorithm is developed. The CQR-ORELM model offers flexibility and efficiency to explore potential nonlinear characteristic among wind speed variables and improve model robustness and predictive capability. A hybrid algorithm with the combination of particle swarm optimization and gravitational search algorithm (PSOGSA) is utilized to fine-tune the optimal value of weights and bias in the ORELM network structure, while the binary version of PSOGSA (BPSOGSA) is exploited asa feature selection method to construct the most relevant input feature matrix for the model. Aimed to alleviate the influence of uncertainty in wind speed time series, a time adaptive filter based empirical mode decomposition (TVF-EMD) approach is used to decompose the original wind speed series into several intrinsic mode functions (IMFs). Each IMF is well-developed using the proposed method and the final forecasting results are obtained through aggregate calculation. This proposed wind forecasting methodology is compared to several benchmarks. The effectiveness of the proposed forecast strategy to predict wind speed is evaluated by the trials of 10min ahead wind speed forecasting at two locations of the National Renewable Energy Laboratory (NREL). Comparative results confirm that the developed CQR-ORELM with feature selection procedure can describe the complete conditional distribution information hidden in variables and provide satisfactory wind speed information.

A hybrid approach based on the Gaussian process with t-observation model for short-term wind speed forecasts

Accurate wind speed prediction is a significant factor in improving and optimizing wind power production. Particularly, reliable short-term wind speed forecasting contributes to the real-time optimization of wind farm operation. However, this short-term forecasting task remains challenging due to the strong stochastic nature and dynamic uncertainty of wind speed. This paper proposes a hybrid model that consists of the Empirical Wavelet Transform (EWT), Expectation Propagation (EP) algorithm and Gaussian process regression with the Student-t Observation Model (GPR-t) for short-term wind speed forecasting. The proposed approach firstly extracts meaningful information from a short-term wind speed series and subsequently models the inherent uncertainty and the dynamic features of the wind speed time-series. Additionally, the wind speed series presents a time-varying characteristic. Thus, this study adopts a moving window approach in the prediction processes, thereby permitting the proposed model to respond quickly to the dynamic characteristic of wind speed. To examine the forecasting performance of the suggested hybrid model, the validation of the proposed model is performed against several other existing models with half-hour and hourly wind speed data obtained from a windmill farm located in northwestern China. The computational results demonstrate that the proposed hybrid approach generates satisfactory point predictive accuracy and interval forecasting performance.

Short-Term Wind Speed Forecasting Using Regularization Extreme Learning Machine

High efficient and accurate wind speed prediction is the basis of the wind farm power prediction. So it is helpful for the control of the wind power and has great importance to the parallel operation of the wind farms. Wind speed time series has strong nonlinear and volatility. Besides, it is very difficult to accurately predict. A new method of short-term wind speed forecasting is proposed based on regularized extreme learning machine (regularized extreme learning machine, RELM). First of all, the autocorrelation function (ACF) is used to analyze the correlation of wind speed time series. After that the number embedded in the time dimension is gotten. And the forecast network parameters such as inputs, output and so on are determined. That is to say the RELM model is set up. Then, using the training set trains the network parameters to get the RELM prediction model trained. Finally, prediction results are obtained with the test set data. And the wind speed data from the American wind energy technology center is carried out on the experiment. It shows that the new method has better prediction precision compared with the standard ELM and the traditional neural network.

Research and application of a hybrid wavelet neural network model with the improved cuckoo search algorithm for electrical power system forecasting

Electricity forecasting plays an important role in the operation of electrical power systems. Many models have been developed to obtain accurate forecasting results, but most of them focus more on a single forecasting indicator, such as short-term load forecasting (STLF), short-term wind speed forecasting (STWSF) or short-term electricity price forecasting (STEPF). In this paper a new hybrid model based on the singular spectrum analysis (SSA) and modified wavelet neural network (WNN) is proposed for all the short-term load forecasting, short-term wind speed forecasting and short-term electricity price forecasting. In this model, a new improved cuckoo search (CS) algorithm is proposed to optimize the initial weights and the parameters of dilation and translation in WNN. Case studies of half-hourly electrical load data, 10-min-ahead wind speed data and half-hourly electricity price data are applied as illustrative examples to evaluate the proposed hybrid model, respectively. Experiments show that the hybrid model resulted in 46.4235%, 31.6268% and 25.8776% reduction in the mean absolute percentage error compared to the comparison models in short-term load forecasting, short-term wind speed forecasting and short-term electricity price forecasting, respectively.

A compound structure of ELM based on feature selection and parameter optimization using hybrid backtracking search algorithm for wind speed forecasting

Reliable wind speed forecasting is essential for wind power integration in wind power generation system. The purpose of paper is to develop a novel hybrid model for short-term wind speed forecasting and demonstrates its efficiency. In the proposed model, a compound structure of extreme learning machine (ELM) based on feature selection and parameter optimization using hybrid backtracking search algorithm (HBSA) is employed as the predictor. The real-valued BSA (RBSA) is exploited to search for the optimal combination of weights and bias of ELM while the binary-valued BSA (BBSA) is exploited as a feature selection method applying on the candidate inputs predefined by partial autocorrelation function (PACF) values to reconstruct the input-matrix. Due to the volatility and randomness of wind speed signal, an optimized variational mode decomposition (OVMD) is employed to eliminate the redundant noises. The parameters of the proposed OVMD are determined according to the center frequencies of the decomposed modes and the residual evaluation index (REI). The wind speed signal is decomposed into a few modes via OVMD. The aggregation of the forecasting results of these modes constructs the final forecasting result of the proposed model. The proposed hybrid model has been applied on the mean half-hour wind speed observation data from two wind farms in Inner Mongolia, China and 10-min wind speed data from the Sotavento Galicia wind farm are studied as an additional case. Parallel experiments have been designed to compare with the proposed model. Results obtained from this study indicate that the proposed hybrid model can provide an effective modeling approach to capture the nonlinear characteristics of wind speed signal and thus providing more satisfactory forecasting results.

A Hybrid Method for Short-Term Wind Speed Forecasting

The accuracy of short-term wind speed prediction is very important for wind power generation. In this paper, a hybrid method combining ensemble empirical mode decomposition (EEMD), adaptive neural network based fuzzy inference system (ANFIS) and seasonal auto-regression integrated moving average (SARIMA) is presented for short-term wind speed forecasting. The original wind speed series is decomposed into both periodic and nonlinear series. Then, the ANFIS model is used to catch the nonlinear series and the SARIMA model is applied for the periodic series. Numerical testing results based on two wind sites in South Dakota show the efficiency of this hybrid method.

WRF Hub-Height Wind Forecast Sensitivity to PBL Scheme, Grid Length, and Initial Condition Choice in Complex Terrain

Abstract This study evaluates the sensitivity of wind turbine hub-height wind speed forecasts to the planetary boundary layer (PBL) scheme, grid length, and initial condition selection in the Weather Research and Forecasting (WRF) Model over complex terrain. Eight PBL schemes available for the WRF-ARW dynamical core were tested with initial conditions sources from the North American Mesoscale (NAM) model and Global Forecast System (GFS) to produce short-term wind speed forecasts. The largest improvements in forecast accuracy primarily depended on the grid length or PBL scheme choice, although the most important factor varied by location, season, time of day, and bias-correction application. Aggregated over all locations, the Asymmetric Convective Model, version 2 (ACM2) PBL scheme provided the best forecast accuracy, particularly for the 12-km grid length. Other PBL schemes and grid lengths, however, did perform better than the ACM2 scheme for individual seasons or locations.

Short-term wind speed forecasting by spectral analysis from long-term observations with missing values

In this paper, we propose a novel wind speed forecasting framework. The performance of the proposed framework is assessed on the wind speed measurements collected from the five meteorological stations in the Marmara region of Turkey. The experimental results show that trimming of the diurnal, the weekly, the monthly, and the annual patterns in the measurements significantly enhances the estimation accuracy. The proposed framework builds on data de-trending, covariance-factorization via a recently developed subspace method, and one-step-ahead and/or multi-step-ahead Kalman filter prediction ideas. The data sets do not have to be complete. In fact, as in sensor failures, intermittently or sequentially missing measurements are permitted. The numerical experiments on the real data sets show that the wind speed forecasts, in particular the multi-step-ahead forecasts, outperform the benchmark values computed with the persistence forecasting models by a clear difference.