图像处理
直方图
云量
环境科学
计算机科学
光伏系统
遥感
图像(数学)
人工智能
云计算
工程类
地理
操作系统
电气工程
作者
Wai Kean Yap,Roy Galet,Kheng Cher Yeo
摘要
Maximizing the efficiency of solar photovoltaic (PV) systems in the unique Northern Australia climate (dry and wet seasons annually) is critical in particular for rooftop installations. Optimum system efficiencies are usually achieved during the dry season, due to consistent sunshine hours, minimal cloud cover and no rainfall. However, the rate of dusting and soiling occurring on the panel surfaces are also high during this period, reducing the system’s efficiency. One of the solutions to maintain the optimum efficiency is to clean the panels regularly. However, this was rarely done, especially for rooftop systems due to access problems, and is an added cost. In addition, the cleaning frequency differs from locations and technology types. This paper presents a non-evasive methodology in quantifying the amount of dust and soiling on solar PVs by investigating five different image-processing techniques. This study looks at analyzing color histograms and statistical properties of the captured PV images. An image-processing Toolbox were developed in this study by adopting the following techniques: binarization, histogram model, statistical model, image matching and texture matching. Two image tests were presented: controlled image and actual image tests with average errors of 12.38% and 10.8% were achieved respectively. Results showed that the binarization algorithm exhibited the fastest and the most accurate reading on the controlled image test and the image matching algorithm exhibited the highest accuracy on the actual image test. The methods of analyzing PV panel dusting and soiling were proven to be accurate, low-cost, easy to implement and critically, provides the end-user the necessary information in maintaining their PV system efficiency over the wet and dry seasons of Northern Australia.
科研通智能强力驱动
Strongly Powered by AbleSci AI