A Novel Stochastic Dynamic Modeling for PV Systems Considering Dust and Cleaning Effects

Abstract

Stochastic photovoltaic (PV) modeling is essential for the long-term planning of renewable power generation. One of the most prevalent problems that PV systems face is the accumulation of dust on the PV panel surface that negatively impacts the output power. Wind speed along with other weather variables including relative humidity, temperature, and precipitation are some of the major factors that contribute to dust accumulation. Unlike the available models in the literature, this thesis presents a novel dynamic model of the PV output power profile considering the effect of dust accumulation using a Markov chain model. The proposed model is composed of three stages and it incorporates the seasonal variations in the weather conditions as well as the desired cleaning frequency, which affects the overall energy yield of the PV system. The first stage is the data acquisition and processing stage where the raw data is discretized and categorized. The second stage utilizes the outcome of the first stage in a Markovian Chain model, which is the core of the overall model. The third and final stage is the cumulative distribution function generation, which is generated using the probability mass function output of the Markov Chain simulation. The outcome of the model can be described as virtual scenarios, which can help the investors to decide on the optimal size of the PV system and the optimal cleaning frequency for each season subject to some constraints. The model outcome shows an error of less than 5% when compared to actual data collected from the field without cleaning. Various case studies are presented to show the effectiveness of the proposed model and its benefits.