Wind power curve modeling is a challenging task due to the existence of inconsistent data, in which the recorded wind power is far away from the theoretical wind power at a given wind speed. In this case, confronted with these samples, the estimated errors of wind power will become large. Thus, the estimated errors will present two properties: heteroscedasticity and error distribution with a long tail. In this paper, according to the above error characteristics, heteroscedastic spline regression model (HSRM) and robust spline regression model (RSRM) are proposed to obtain more accurate power curves even in the presence of the inconsistent samples. The results of power curve modeling on the real-world data show the effectiveness of HSRM and RSRM in different seasons. As HSRM and RSRM are optimized by variational Bayesian, except the deterministic power curves, probabilistic power curves, which can be used to detect the inconsistent samples, can also be obtained. Additionally, with the data processed by replacing the wind power in the detected inconsistent samples with the wind power on the estimated power curve, the forecasting results show that more accurate wind power forecasts can be obtained using the above data processing method.
20 April 2020 in IEEE Transactions on Sustainable Energy
An accurate wind turbine power curve (WTPC) plays a vital role in wind power forecasting and wind turbine condition monitoring. There are two major shortcomings of current WTPC models that prevent more accurate WTPC estimation, limited nonlinear fitting ability and the lack of in-depth understanding of the complex characteristics of WTPC. This paper proposes two novel regression models to overcome these two disadvantages simultaneously. First, they make use of multiple spline regression models (MSRM) with different basis functions and different numbers of knots to describe the complex nonlinear relationship between wind speed and wind power. Moreover, sparse prior distributions help avoid the adverse effects of redundant mapping features and useless basis functions on the model performance. Second, they embed the heteroscedasticity of WTPC modeling into MSRM based on Gaussian and Student's t-distributions, respectively. Finally, two sparse heteroscedastic MSRM with Gaussian and Student's t-distributions will be constructed and named as SHMSRM-G and SHMSRM-T, respectively. We compare the proposed models with fifteen benchmark models, and find that they can generate more accurate WTPCs than the others in different seasons and different wind farms. Thus, it is important to consider the complex nonlinear fitting ability and heteroscedasticity together in constructing accurate WTPC models.
Wind power curves play important roles in developing and utilizing wind energy, including wind power forecasting, wind turbine condition monitoring, wind energy potential estimation and wind turbine selection. In this paper, we show that the error distribution in wind power curve modeling is asymmetric and complex, rather than a Gaussian distribution. Considering this error characteristic, we develop two novel asymmetric spline regression models, MoAG-ASR and MoAEP-ASR, in which the error terms are assumed to obey mixture of asymmetric Gaussian distributions (MoAG) and mixture of asymmetric exponential power distribution (MoAEP), respectively. We compare the proposed models with sixteen popular wind power curve fitting models, and find that the proposed models rank higher than the others. Moreover, besides the asymmetry, two statistics (skewness and kurtosis) also show that the error distribution in wind power curve modeling is thick-tailed and leptokurtic. However, MoAG and MoAEP can precisely fit such complex error distribution. Thus, MoAG-ASR and MoAEP-ASR are effective tools for wind power curve modeling.
16 May 2023 in IEEE Transactions on Sustainable Energy
Accurate wind power curves (WPCs) are crucial for wind energy development and utilization, e.g., wind power forecasting and wind turbine condition monitoring. In the era of big data, large-scale datasets make the training of power curve models inefficient, especially for kernel-based models. Furthermore, most models do not take into account the error characteristics of WPC modeling. In this study, a large-scale generalized kernel-based regression model is proposed to solve the above problem. First, a generalized loss function, which can model both symmetric and asymmetric error distributions, is designed for model training. Then, the Nyström technique is employed to get the approximate kernel matrix, based on which an eigenvalue-based kernel regression framework is constructed. Next, a large-scale generalized kernel-based regression model is developed with model parameters tuned using the alternating direction method of multipliers. Before WPC modeling, a three-step data processing method based on isolation forest is designed to process missing data, irrational data, and outliers in the collected data. The WPC modeling results on four large-scale wind datasets demonstrate that the proposed model generates accurate WPCs with high efficiency. Furthermore, the effect of turbulence intensity on WPC modeling and the effectiveness of LSGKRM with multivariate inputs are also verified.
23 December 2016 in International Journal of Green Energy
Wind 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.