A NOVEL PROBABILISTIC GENERATION MODEL FOR GRID CONNECTED PV BASED DISTRIBUTED GENERATION

Abstract

The feasibility of renewable energy resources such as solar energy is marked by uncertainties that make it an unpredictable mean of power generation. To guarantee an uninterrupted power supply, solar irradiance modelling can be taken as a useful step towards meeting the operational challenges of electric power grid. This paper proposes a dynamic Probabilistic generation model to estimate and generate the time-coupled solar irradiance patterns. Initially, clustering of yearly solar irradiance measurements is performed to obtain a meaningful grouping of similar days. One-hour time step is considered to construct a time-coupled probabilistic model of solar irradiance data based on a Beta distribution. The parameters of beta distribution are found by considering the variations of irradiance patterns at two successive time steps. The probabilistic model is then used to generate number of aggregate solar irradiance generation scenarios. The effectiveness of proposed scenario generation approach is evaluated through Average Mean Absolute Percentage Error (AMAPE) and comparison with the probabilistic model already available in the literature.