Jie Yan

A Hybrid Forecasting Method for Wind Power Ramp Based on Orthogonal Test and Support Vector Machine (OT-SVM)

In an electric power system with a high penetration of wind power, incoming power ramps pose a serious threat to the power system. To adopt suitable response strategies for wind power ramps, it is important to predict them accurately and in a timely manner. Since power ramps are caused by various factors, their occurrence has irregular characteristics and vary by location, bringing great difficulty in forecasting. To solve this problem, a hybrid forecasting model termed as orthogonal test and support vector machine (OT-SVM) was developed in this paper, which combines an orthogonal test (OT) with a support vector machine (SVM). A novel factor analysis method was established based on the theory of the OT, and applied to determine the optimal inputs of a SVM. The effectiveness of OT-SVM was tested with three wind farms in China, while comparing the results with other related methods. The results show that the proposed OT-SVM has the highest accuracy covering different input numbers and time resolutions. In addition, a novel evaluation index mean accuracy index was proposed, considering both the missed ramps and false ramps, which can be used as a supplementary index for critical success index.

Clustering methods of wind turbines and its application in short-term wind power forecasts

Commonly used wind power forecasts methods choose only one representative wind turbine to forecast the output power of the entire wind farm; however, this approach may reduce the forecasting accuracy. If each wind turbine in a wind farm is forecasted individually, this considerably increases the computational cost, especially for a large wind farm. In this work, a compromise approach is developed where the turbines in the wind farm are clustered and a forecast made for each cluster. Three clustering methods are evaluated: K-means; a self-organizing map (SOM); and spectral clustering (SC). At first, wind turbines in a wind farm are clustered into several groups by identifying similar characteristics of wind speed and output power. Sihouette coefficient and Hopkins statistics indices are adopted to determine the optimal cluster number which is an important parameter in cluster analysis. Next, forecasting models of the selected representative wind turbines for each cluster based on correlation analysis are established separately. A comparative study of the forecast effect is carried to determine the most effective clustering method. Results show that the short-term wind power forecasting on the basis of SOM and SC clustering are effective to forecast the output power of the entire wind farm with better accuracy, respectively, 1.67% and 1.43% than the forecasts using a single wind speed or power to represent the wind farm. Both Hopkins statistics and Sihouette coefficient are effective in choosing the optimal number of clusters. In addition, SOM with its higher forecast accuracy and SC with more efficient calculation when applied into wind power forecasts can provide guidance for the operating and dispatching of wind power. The emphasis of the paper is on the clustering methods and its effect applied in wind power forecasts but not the forecasting algorithms.

Novel Cost Model for Balancing Wind Power Forecasting Uncertainty

The intermittency of wind generation creates nonlinear uncertainties in wind power forecasting (WPF). Thus, additional operating costs can be incurred for balancing these forecasting deviations. Normally, large wind power penetration requires accurate quantification of the uncertainty-induced costs. This paper defines this type of costs as wind power uncertainty incremental cost (WPUIC) and wind power uncertainty dispatch cost (WPUDC), and it then formulates a general methodology for deriving them based on probabilistic forecasting of wind power. WPUIC quantifies the incremental cost induced from balancing the uncertainties of wind power generation. WPUDC is a balancing cost function with a quadratic form considering diverse external conditions. Besides, the risk probability (RP) of not meeting the scheduled obligation is also modelled. Above models are established based on a newly developed probabilistic forecasting model, varying variance relevance vector machine (VVRVM). Demonstration results show that the VVRVM and RP provide accurate representation of WPF uncertainties and corresponding risk, and thus they can better support and validate the modelling of WPUDC and WPUIC. The proposed cost models have the potential to easily extend traditional dispatches to a new low-carbon system with a high penetration of renewables.

Adaptabilities of three mainstream short-term wind power forecasting methods

Variability and intermittency of wind is the main challenge for making a reliable wind power forecasting (WPF). Meteorological and topological complexities make it even harder to fit any forecasting algorithm to one particular case. This paper presents the comparison of three short term WPF models based on three wind farms in China with different terrains and climates. The sensitivity effects of training samples on forecasting performance are investigated in terms of sample size, sample quality, and sample time scale. Then, their adaptabilities and modeling efficiency are also discussed under different seasonal and topographic conditions. Results show that (1) radial basis function (RBF) and support vector machine (SVM) generally have higher prediction accuracy than that of genetic algorithm back propagation (GA-BP), but different models show advantages in different seasons and terrains. (2) WPF taking a month as the training time interval can increase the accuracy of short-term WPF. (3) The change of sample number for the GA-BP and RBF is less sensitive than that of the SVM. (4) GA-BP forecasting accuracy is equally sensitive to all size of training samples. RBF and SVM have different sensibility to different size of training samples. This study can quantitatively provide reference for choosing the appropriate WPF model and further optimization for specific engineering cases, based on better understanding of algorithm theory and its adaptability. In this way, WPF users can select the suitable algorithm for different terrains and climates to achieve reliable prediction for market clearing, efficient pricing, dispatching, etc.

Uncertainty Analysis of Wind Power Prediction Based on Quantile Regression

Short-term wind power prediction is an effective way to mitigate the impact of large-scale wind power variability incurring to the electric power system. Given the fluctuation of wind energy is random; uncertainty analysis of wind power prediction is very important for engineering application. A risk assessment index of wind power prediction named PaR (Predict at Risk) was proposed based on quantile regression. And an uncertainty analysis model for wind power prediction was established to provide a possible fluctuation range of predicted wind power at any confidence level. Operation data and predicted power from a wind farm in north China are used as a test case to validate proposed model. The results show that the model can tolerate a wide range of different conditions without hypothesis of the error distribution of wind power prediction and is an effective, practical way to provide uncertainty information.

Multi-step-ahead Method for Wind Speed Prediction Correction Based on Numerical Weather Prediction and Historical Measurement Data

Increasing the accuracy of wind speed prediction lays solid foundation to the reliability of wind power forecasting. Most traditional correction methods for wind speed prediction establish the mapping relationship between wind speed of the numerical weather prediction (NWP) and the historical measurement data (HMD) at the corresponding time slot, which is free of time-dependent impacts of wind speed time series. In this paper, a multi-step-ahead wind speed prediction correction method is proposed with consideration of the passing effects from wind speed at the previous time slot. To this end, the proposed method employs both NWP and HMD as model inputs and the training labels. First, the probabilistic analysis of the NWP deviation for different wind speed bins is calculated to illustrate the inadequacy of the traditional time-independent mapping strategy. Then, support vector machine (SVM) is utilized as example to implement the proposed mapping strategy and to establish the correction model for all the wind speed bins. One Chinese wind farm in northern part of China is taken as example to validate the proposed method. Three benchmark methods of wind speed prediction are used to compare the performance. The results show that the proposed model has the best performance under different time horizons.