Predicting green hydrogen yield and gross energy quantity from water hyacinth images: Deep learning approach

Ghada Dahy, Mohamed Torky, Aboul Ella Hassanien,
Predicting green hydrogen yield and gross energy quantity from water hyacinth images: Deep learning approach,
International Journal of Hydrogen Energy,
Volume 189,
2025,
152027,
ISSN 0360-3199,
https://doi.org/10.1016/j.ijhydene.2025.152027.
(https://www.sciencedirect.com/science/article/pii/S036031992505030X)
Abstract: Green hydrogen production is fast becoming an important research issue in updating the classical operations of energy production by relying on sustainable and green energy sources. Because of the heavy cost of generating green hydrogen based on thermochemical and electrolysis, the water hyacinth plant can be utilized as a rich source and cost-effective solution for producing green hydrogen and sustainable energy quantities. This paper introduces a real research trial about the potential of using a proposed deep learning approach for predicting green hydrogen yield and gross energy quantity from given Water hyacinth images. Four deep learning models, MobileNet-CNN, Xception-CNN, VGG 16-CNN, and ResNet 50-CNN for recognizing water hyacinth images from a given dataset of water waste mixtures. Then, a proposed HSV color model-based image segmentation and mathematical model are proposed to estimate the total real weight (in kilograms) and square area (in m2) of water hyacinth to predict green hydrogen yield (in kilograms) and the generated gross energy (in kilowatt) from water hyacinth images (in pixels). The proposed approach is validated and verified using two merged benchmark datasets of water waste mixtures. The water hyacinth recognition results clarified the outperforming of ResNet 50-CNN in recognizing water hyacinth images from water waste mixtures compared to MobileNet-CNN, Xception-CNN, VGG 16-CNN, and other deep learning models in the literature, where, it achieved training and testing accuracy of 100 %, and 98 % respectively. another important and interesting finding clarified that 117 images of a water hyacinth plant can produce an AVG of total GH2 yield of 6.815 (Kg) and AVG of total Gross Energy quantity of about 228.984 (Kwh).
Keywords: Green hydrogen; Water hyacinth; Green energy; Deep learning; Image segmentation