Probabilistic model of wind power output considering temporal autocorrelation and cross-correlation with forecast output

Abstract

Wind power output possesses temporal autocorrelation and cross-correlation with the forecast output. However, these two types of correlations are quite different. Existing studies fail to model them simultaneously, which reduces the accuracy of the conditional probability distribution of wind power output and has a negative effect on the economic operation of wind power systems. Therefore, this paper proposes a method for modeling the conditional probability distribution of wind power output that can take both correlations into account. The copula function is used to separately model the dependency between wind power outputs and the uncertainty of actual/forecast output. Based on pair-copula theory, the modeling of the conditional probability distribution of wind power output with respect to the two correlations is broken down into the chain transfer process of the traditional two-dimensional copula function. The probability distribution function of wind power output can be obtained as long as enough historical data is available. No specific forecast techniques are required. Using the actual wind field output data in Belgium, the proposed method is compared with the methods that only consider one of the correlations to prove its effectiveness.