Abstract
Understanding the nature and characteristics of wind power variability is critical to improving power system operations. Using a relatively large volume of measured wind power time series, the conditional range metric (CRM) has successfully been employed to quantify intra-hour wind power variability. The objective of this paper is to extend the application of the CRMbased method to sparsely sampled wind power time series. To improve the accuracy and adaptability of the CRM, a gamma distribution is proposed to model the power-deviation series without introducing the joint probability density matrix. The shape and inverse scale parameters α and β of the gamma distribution are calculated by the maximum-likelihood estimator (MLE) at each production level. Bayesian estimators (BEs) for α and β are then developed as a posterior joint distribution to calculate their expectations by taking the mean value of the paired sample points via a modified rejection sampling strategy. The match between the two estimation results indicates a valid assumption of the likelihood distribution and exact estimated parameters. Applying data from the National Renewable Energy Laboratory and their down-sampled subsets, the efficacy of the proposed CRM with the gamma distribution approach is compared to that of the existing CRM with a quantile model.

This publication has 10 references indexed in Scilit: