首页|Enhancing solar panel performance: A machine learning approach to dust detection and automated water sprinkle-based cleaning strategy
Enhancing solar panel performance: A machine learning approach to dust detection and automated water sprinkle-based cleaning strategy
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
Elsevier
The efficiency of photovoltaic (PV) modules significantly reduces due to accumulation of dust. To minimize the dust effect on PV in a cost-effective manner, optimal cleaning interval need to be decided. To accomplish this objective, machine learning(ML) models can be utilized to detect dust level on PV beyond predefined threshold which would then aid in deciding whether or not to clean the panels without on-site human intervention. With this goal, this study analyzes the detrimental impact of dust on PV systems in Bangladesh and proposes a novel ML classification based dust detection method followed by development of a cleaning system. Several ML classifiers have been implemented and their performance are evaluated, with the best performing model Artificial Neural Network (ANN) achieving the highest accuracy of 98.11%. Upon dust detection by the ML model, the water sprinkler cleaning system gets wirelessly activated by the user, which effectively removes dust by spraying pressurized water onto the panel. The proposed cleaning system restores dusty PV module efficiency to match to that of the clean module (14.87%). Moreover, an economic study has been done by quantifying the decrease inefficiency as a financial loss to assess the viability of the cleaning system. The result shows that the proposed cleaning system is economically viable for PV systems having capacities above 2.89 kWp.
NETL
NSTL
Elsevier
