Biodiesel Yield Prediction from Sunflower Oil Using Artificial Intelligence: Towards Sustainable, and Renewable Energy Sources

The world has become increasingly concerned about the sustainability of petroleum-based fuel due to its widespread use across a range of industries, the loss of fossil fuel resources, and the uncertainty surrounding crude oil market prices. These environmental issues are resulting from the growing hazardous pollutant emissions and greenhouse gases. Consequently, it is crucial to explore clean energy alternatives such as biodiesel. An eco-friendly alternative to diesel fuel is biodiesel which is produced using renewable resources, such as animal fats and vegetable oils. It can be used in diesel-powered vehicles found in automobiles, buses, lorries, construction machinery, boats, generators, and oil-fired residential heaters. Sunflower, a leading oilseed crop cultivated worldwide, has gained significant attention in recent decades as a potential source for producing biodiesel. This paper proposes an artificial intelligence (AI) framework to enhance the production process of biodiesel from sunflower oil. The framework consists of two phases: in the first phase, a Deep Learning (DL) model called VGG16 is employed to classify diseases affecting sunflower leaves, specifically Downy mildew, Gray Mold, and Leaf scars. This classification is based on a dataset of 467 images of sunflowers. The second phase involves using healthy sunflowers to train a Machine Learning (ML) model called Gradient Boosting Regression (GBR). This model predicts the biodiesel yield based on various parameters involved in manufacturing of biodiesel, such as reaction temperature, reaction time, and catalyst weight. Experimental results demonstrate that the proposed models outperform other approaches in the literature. VGG16 achieves a 99.6% accuracy in classifying sunflower leaf diseases, while GBR achieves a high coefficient of determination (R2 = 0.994) for predicting biodiesel yield.