Data Analysis and Modeling of a Solar Photovoltaic System

This project aims to conduct a study of modeling the power conversion process of a solar photovoltaic system with data-driven approaches. A solar photovoltaic (PV) system located on the rooftop of an institutional building in the Coloane island of Macau Special Administrative Region is considered as the research target. An experiment of operating the solar PV system will be designed and the experimental data will be collected for the data analysis. Techniques of design of experiments (DOE) will be considered to generate a sample of parameter settings for operating the solar PV system in the experiment. Existing data mining algorithms will be employed to extract the solar power model from the collected dataset. Besides the existing algorithms, novel data-driven approaches, such as ensemble learning, will be investigated to develop the solar power model. The multivariate linear regression and the autoregressive with exogenous model will be considered as baseline models in this research. A comparative analysis of the proposed approaches and baseline models will be performed. Statistical metrics will be introduced to assess the effectiveness of data-driven approaches in modeling the solar PV system. The most suitable algorithm for modeling the solar PV system will be recommended.