Data‐driven wind speed forecasting using deep feature extraction and LSTM

Wu, Yuxi; Wu, Qingbiao; Zhu, Jiaqi

doi:10.1049/iet-rpg.2018.5917

Cited by 86 publications

(40 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The prediction method based on statistical theory has been widely used recently. Common statistical models include grey model (GM) [18], [19], autoregressive integrated moving average (ARIMA) [20], [21], support vector regression (SVR) [22], [23], multiple linear regression (MLR) [24], long-term and short-term memory network (LSTM) [9], [25], [26], artificial neural network(ANN) [27], machine learning algorithm(ML) [22], deep learning algorithm(DL) [29] and other hybrid models [30]. More and more researchers use machine learning methods to extract internal patterns from data.…”

Section: Introductionmentioning

confidence: 99%

Hybrid Prediction Method for Wind Speed Combining Ensemble Empirical Mode Decomposition and Bayesian Ridge Regression

Yang

2020

IEEE Access

View full text Add to dashboard Cite

In recent years, with the rapid development of wind power generation, some problems are gradually highlighted. At present, one of the essential methods to solve these problems is to predict wind speed. In this paper, a hybrid BRR-EEMD method is proposed for short-term wind speed prediction based on the Bayesian ridge regression prediction method and ensemble empirical mode decomposition. We use ensemble empirical mode decomposition of the hybrid method to decompose complex time series of wind speed into several relatively milder, more regular, and stable subsequences. Then each subsequence is carried out by using the Bayesian ridge regression method. The value of each subsequence is predicted by it. Finally, the value of multiple subsequences is fused to form the prediction results of the original complex time series of wind speed. In order to verify the proposed method comprehensively, this paper selects two data to test. According to the results, predicted values have shown higher accuracy compared with the various prediction methods. Therefore, the hybrid BRR-EEMD method is accurate and effective in predicting wind speed, which has practical significance and potential value. INDEX TERMS Ensemble empirical mode decomposition, short-term predicting, Bayesian ridge regression, wind speed, time series.

show abstract

Section: Introductionmentioning

confidence: 99%

Hybrid Prediction Method for Wind Speed Combining Ensemble Empirical Mode Decomposition and Bayesian Ridge Regression

Yang

2020

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Furthermore, deep learning provides a systematic solution to handle big data problems, since deep learning models can take advantage of many samples to learn the hierarchical representations by combining the low-level input more effectively for big data with the characteristics of the large variety and large veracity. In a few recent publications, the results of deep learning have shown efficiency and greater accuracy in wind speed and wind power forecast problems [15][16][17][18]. Unfortunately, all these studies focused on the overall wind speed or wind power of a wind farm.…”

Section: Introductionmentioning

confidence: 99%

Deep learning approach for wind speed forecasts at turbine locations in a wind farm

Kou

Wang

Liang

et al. 2020

IET Renewable Power Generation

View full text Add to dashboard Cite

In a wind farm, individual turbines disturb the wind field by generating wakes, so wind speeds at various turbine locations are different. From the perspective of wind farm control, there is an interest in dynamic optimization of the power reference for each individual wind turbine, and the wind speed forecast at each turbine location is hence required. This paper develops a joint model of convolutional neural network (CNN) and the gated recurrent units (GRU) to forecast the wind speed at turbine locations. This model employs a two-layer architecture. At the lower-layer, the spatial features are automatically extracted by CNN. The extracted spatial features describe the spatial correlations among multiple wind turbines. At the upperlayer, GRU learns the temporal correlations across the extracted spatial features. This joint model is trained in an integrated manner. A salient characteristic of this model is that it extracts high-level spatial-temporal features from wind data. These automatically learnt features capture the spatial-temporal wind dynamics and interactions in a wind farm, thus being informative and appropriate for the forecasting at specific turbine locations. The simulation on actual data demonstrates the effectiveness of the presented model.

show abstract

“…With the rapid development of wind power generation technology, the installed capacity and unit capacity of wind turbines are increasing constantly. Improving the service life of wind turbines and the wind energy utilisation is a research focus [1]. Studies have shown that accurate wind direction prediction can effectively improve the performance of wind turbine yaw system [2].…”

Section: Introductionmentioning

confidence: 99%

Screening methodology of correlated wind turbines for wind direction prediction based on yawing manoeuvre data

Shen

2020

IET Renewable Power Generation

View full text Add to dashboard Cite

Accurate prediction of wind direction can improve wind energy utilisation and extend the life of the wind turbine yaw system effectively. The ‘measurement–sharing–correlation–prediction–verification’ (MSCPV) wind parameter sensing technology based on the turbine network provides a new perspective for wind direction prediction, in which the screening of correlated wind turbines is a key step. In this study, a calculation method of wind direction spatial correlation is proposed, and correlated turbines are selected according to the correlation strength. The study first analyses the limitations of traditional wind direction correlation based on the principle of MSCPV and proposes the concept of correlated turbine screening based on yawing correlation. Then, the modelling and calculation of yawing correlation are analysed in detail. The prediction based on correlated wind turbines selected by different methods is evaluated. Case study results show that compared to wind direction correlation and wind turbines' distance, the yawing correlation has a better effect on the screening of the spatially correlated turbines. The prediction accuracy based on different correlated turbines is positively correlated with the degree of yawing correlation, indicating the proposed screening method is effective.

show abstract

Data‐driven wind speed forecasting using deep feature extraction and LSTM

Cited by 86 publications

References 29 publications

Hybrid Prediction Method for Wind Speed Combining Ensemble Empirical Mode Decomposition and Bayesian Ridge Regression

Hybrid Prediction Method for Wind Speed Combining Ensemble Empirical Mode Decomposition and Bayesian Ridge Regression

Deep learning approach for wind speed forecasts at turbine locations in a wind farm

Screening methodology of correlated wind turbines for wind direction prediction based on yawing manoeuvre data

Contact Info

Product

Resources

About