Classification of solar variability using k-means method for the evaluation of solar photovoltaic systems performance

Mahmoud Farghaly, Sara Ragab and Gan, Chin Kim (2022) Classification of solar variability using k-means method for the evaluation of solar photovoltaic systems performance. International Journal of Renewable Energy Research, 12 (2). pp. 692-702. ISSN 1309-0127

Text 2022 - CLASSIFICATION OF SOLAR VARIABILITY - IJRER.PDF Download (2MB)

Abstract

This paper presents a classification of solar tilt irradiance using the k-means clustering method, and an evaluation of the impact of different solar variabilities on monocrystalline and thin-film photovoltaic (PV) systems. The variability index and clearness index were implemented to quantify five years of solar datasets to assist in clustering solar variabilities. The elbow method was used to validate the k-clustering for solar variabilities. Due to the compact solar datasets, the Silhouette Coefficient and Gap Statistic were utilized to validate the k-cluster numbers. The PV performance was evaluated using the generated power, energy, and performance ratio for solar datasets from 2015 and 2019. Equal number of samples was taken from each PV system to analyse the average calculated values. The results showed that the elbow method was inaccurate for clustering solar variabilities, although it showed a weak elbow at K2 that was inaccurate for grouping solar variabilities. However, the k-means validation methods detected K3, K4, and K5 as the best k-cluster numbers. Among them, K4 was compatible for separating four types of solar variabilities, namely, overcast, moderate, mixed (clear/mild), and high variability. Based on the average performance values of the monocrystalline and thin-film PV systems for 2015 compared to 2019, similar degradation values were detected, especially for the performance ratio (0.77) under overcast. The thin-film showed degraded generated power and energy under the moderate type. The degraded generated power and performance ratio for the monocrystalline were due to the high passing clouds under the mixed and high variability types.

Item Type:	Article
Uncontrolled Keywords:	Solar tilt irradiance, Variability index, Clearness index, K-means clustering methods, Monocrystalline, Thin-film
Divisions:	Faculty of Electrical Engineering
Depositing User:	Sabariah Ismail
Date Deposited:	02 Mar 2023 16:22
Last Modified:	02 Mar 2023 16:22
URI:	http://eprints.utem.edu.my/id/eprint/26235
Statistic Details:	View Download Statistic

Actions (login required)

View Item