Key points are not available for this paper at this time.
The identification and categorization of soiling using artificial intelligence is essential for improving solar photovoltaic (PV) system efficiency. It is possible to use focused and effective cleaning techniques on solar panels by classifying and comprehending the particular types of pollutants. This research proposes a modified convolutional neural network (CNN) approach to efficiently classify soiling on solar PV panels. In contrast to the conventional method, the suggested CNN model is created using simple images rather than images plus time series data. As a result, processing speed is increased. Additionally, the suggested model achieved 96.91 percent accuracy. This success establishes our suggested methodology as a ground-breaking development in soiling type detection, providing improved accuracy and a unique strategy in contrast to current approaches. The trained and tested model is deployed on the web using the Flask. Flask is a powerful and flexible microweb framework for Python, Finally, a comparison is made between the existing method and the proposed method.
Chaki et al. (Tue,) studied this question.