Efficient Net Driven Smart Detection of Dust Accumulation on Solar Panels

The accumulation of dust on PV panels results in reduced energy efficiency and output power from the panels. This work proposes an effective dust detection system using the EfficientNet-B0 convolutional neural network (CNN) model. Channel normalization and data cleaninghave been incorporated to enhance accuracy. The proposed system categorizes panels into clean and dusty categories. The EfficientNet B0 architecture ensures high accuracy (88%) with minimal computational complexity. This method enables hands-on maintenance and reduces the need for manual inspection. The system also ensures enhanced solar energy production.