Akademik Veri Yönetim Sistemi
Journal of Environmental Management, cilt.397, 2026 (SCI-Expanded, Scopus)
An essential part of green transformation and a low-carbon circular economy system is the recycling of organic waste through composting. Composting is a cost-effective and environmentally friendly alternative approach that safely converts organic waste into biofertilizers. Composting process efficiency and compost quality can be predicted by machine learning-based models using data from a limited number of experiments. In this study, the effects of composting market waste with hazelnut shells and hulls on compost maturity were modeled using a hybrid model. The hybrid model offers superior features not in existing modeling tools in the literature, such as the ability to simultaneously model linear and nonlinear relationships, providing a fuzzy approach to process uncertainty, and incorporating a deep learning strategy. During the composting, temperature, pH, C/N, moisture content, NH4+/NO3−, and germination index of the final composts were determined. The results showed that the compost with 12.5 % hazelnut shells reached the required maturity standard. The germination index of the final compost increased from 0.958 to 1.255 with the addition of hazelnut shells in all reactors. The hybrid model was compared with five benchmark methods and achieved improvements exceeding 70–80 % in some cases. The hybrid model produced valid and consistent predictions with proportional errors below 5 % for almost all process parameters and was unaffected by initial conditions. In the optimization with the Genetic Algorithm, the input parameters were reached 95 % and above desirability levels. As result, the model, proven accurate and robust, can provide process insights and serve as a reliable simulation tool.