Estimation of the probability density function of renewable power production using a hybrid method of minimum frequency and maximum entropy

Abstract

Accurately estimating the probability distribution of renewable power production is a fundamental and challenging task in the probabilistic analysis of power systems with a high penetration of renewable energy. In this study, a novel hybrid method of minimum frequency and maximum entropy (MFME) is proposed for accurately and rapidly estimating the probability density function (PDF) of renewable power production. Based on the maximum entropy (ME) principle, a probability distribution optimization model is built to obtain a PDF estimator with the maximum distribution entropy. For convenience in solving the model, the probability density estimates of actual samples calculated by the minimum frequency (MF) method are introduced as a supplement to the moment constraints of the ME optimization model. The results indicate that the MFME has a higher accuracy compared with the conventional parameter distribution estimation(CPDE) and Gaussian kernel density estimation (GKDE), and its advantages of no boundary effects and a fast sampling speed for a large original sample size are more suitable for the PDF estimation of renewable power production.