Scopus İndeksli Yayınlar Koleksiyonu / Scopus Indexed Publications Collection
Permanent URI for this collectionhttps://hdl.handle.net/11147/7148
Browse
2 results
Search Results
Article Machine Learning Integrated Solvothermal Liquefaction of Lignocellulosic Biomass to Maximize Bio-Oil Yield(Elsevier Sci Ltd, 2025) Ocal, Bulutcem; Sildir, Hasan; Yuksel, AsliAccelerating consumption of limited fossil-based for economic growth and simultaneously mitigating greenhouse gas emissions create a dilemma that is waiting to be solved by researchers. In this context, solvothermal liquefaction of lignocellulosic biomass to produce bio-oil is a promising way to obtain green energy. However, maximizing bio-oil is challenging to optimize the operating parameters employing conventional techniques due to the complexity and non-linearity of the process. Lately, machine learning approaches have become powerful tools for addressing complex nonlinear problems by predicting process behavior and regulating operating parameters for optimization by learning from datasets. The current research demonstrates integrating experimental and a developed artificial neural network model to optimize solvothermal liquefaction of pinus brutia, based on temperature, water fraction, and biomass amount in maximizing bio-oil generation for the first time. The highest bio-oil yields were obtained at 31.40 %, 18.68 %, and 39.69 %, respectively, with 4 and 8 g biomass in the presence of water, ethanol, and water/ethanol mixture at 240 degrees C. Under the model conditions, the maximum biooil yield was experimentally verified at 46.20%, which was predicted at 48.8 %. Beyond providing accurate yield predictions, the approach highlights the potential of date-driven modeling to reduce experimental workload and cost while aiding parameter selection to improve efficiency. These outcomes emphasize the importance of machine learning integration into liquefaction process, providing remarkable results for future process design, optimization, and scalability. On the other hand, the study also includes characterization results (ultimate, proximate, FTIR, and GC-MS) of selected products and pinus brutia.Article Knowledge-Based Training of Learning Architectures Under Input Sensitivity Constraints for Improved Explainability(Pergamon-Elsevier Science Ltd, 2026) Sildir, Hasan; Erturk, Emrullah; Edizer, Deniz Tuna; Deliismail, Ozgun; Durna, Yusuf Muhammed; Hamit, BahtiyarThe traditional machine learning (ML) training problem is unconstrained and lacks an explicit formulation of the underlying driving phenomena. Such a formulation, based solely on experimental data, does not ensure the delivery of qualitative knowledge among variables due to many theoretical issues in the optimization task. This study further tightens Artificial Neural Networks (ANNs) training by including input sensitivities as additional constraints and applies to regression and classification tasks based on literature data. In theory, such sensitivity represents the change direction of the target variable per change in measurements from indicators. The resulting nonlinear optimization problem is solved th rough a rigorous solver and includes the sensitivity expressions through algorithmic differentiation. Compared to traditional methods, with an acceptable decrease in the prediction capability, the proposed model delivers more intuitive, explainable, and experimentally verifiable predictions under input variable variations, under robustness to overfitting, while serving robust identification tasks. A classification case study includes a patient-oriented clinical decision support system development based on the impact of cancer-indicating variables. A competitive test prediction accuracy is obtained compared to commonly used algorithms despite 10 % decrease in the training. The regression case is built upon the energy load estimation to account for prominent considerations to obtain desired sensitivity patterns and proposed methodology delivers significant accuracy drop compared to some formulations to address knowledge patterns. The approach delivers a compatible pattern with practitioner expertise and is compared to widely used machine learning algorithms, whose performances are evaluated through common statistics in addition to multi-variable response graphs.