Modeling Domestic Solar Generation Profiles Using Elbow and K-Means Clustering Techniques

Solar photovoltaic (PV) power generation started growing at the second-fastest rate of all renewable energy technologies in 2020. Researchers and power generation industries are progressively interested in discussing the behavior analysis of solar energy to daily weather patterns over time. Previous research has often proposed clustering techniques to evaluate solar PV production profiles that are mainly focused on large-scale PV facilities. Therefore, this study explores domestic solar energy clustering techniques to develop a cluster-based representation of solar generation behavior profiles by using solar data collected from single-phase home solar panels of a residence located at Setia Alam, Selangor, Malaysia. Elbow and K-Means clustering approaches were used to classify the entire data into three clusters based on similar meteorological parameters developed using the Python programming language. Data analysis was then performed to determine the correlation coefficient between PV power output and weather conditions. Due to the high ambient temperature and low humidity, it is observed that March 2022 has the potential to produce the largest percentage value of clustered data in high solar generation, with a value of 25.27% ranging from 2839 to 4867W. The results of this study offer a technique that users can apply to evaluate the overall behavior profile of existing residential solar PV systems.