Stochastic simulation framework for renewable power output: Integrating hybrid discrete-continuous distributions with vine copula function

Abstract

The stochastic simulation of wind and photovoltaic power output is an effective approach for supporting the efficient utilization of renewable energy. Existing methods are facing difficulty in simultaneously capturing the stochasticity, intermittency, and multi-order temporal dependencies in renewable energy power output. This study introduces a novel hybrid method that integrates discrete-continuous distribution function with high-dimensional vine copula function to simulate intermittent hourly renewable energy power output sequences. Applied to the Qingshui River hydro-wind-solar hybrid energy system, our methodology demonstrates superior performance over Deep Convolutional Generative Adversarial Networks (DCGAN). The conclusions are as follows: (1) The vine copula model based on the hybrid discrete-continuous distribution function (improved vine copula model) can accurately capture the intermittency of power output, and provides better simulation results in deviation and frequencies; (2) Evaluation metrics reveal significant improvements, with wind power simulation errors reduced from 2.93 % to 1.66 %, and photovoltaic power simulation errors decreasing from 19.56 % to 5.25 %. (3) The improved vine copula model preserves multi-order temporal dependencies of wind and photovoltaic power output within 10-h time horizons; (4) The improved vine copula model significantly reduces computational complexity. The study contributes to the development of refined stochastic simulation methods for renewable energy power output.