Optimal artificial neural network for data driven methane steam reforming model using bfgs quasi- newton / Muhammad Hazman Kamil Bin Munauwir

Faculty of Chemical and Process Engineering Technology

Final Year Report (Bachelor of Chemical Engineering Technology) -- Universiti Malaysia Pahang Al-Sultan Abdullah - 2025

Include bibliographical reference

Methane steam reforming (MSR) is a commonly utilized method of hydrogen synthesis, contributing significantly to the global supply. The Broyden-Fletcher-Goldfarb-Shanno (BFGS) Quasi-Newton method is an iterative technique for solving nonlinear optimization problems. This work investigates the use of MATLAB tools to apply the BFGS Quasi-Newton approach for data-driven process design in the context of methane steam reforming. The BFGS algorithm is a second-order optimization method that uses gradients to approximate the Hessian matrix's inverse. Despite being more computationally demanding and requiring more storage than conjugate gradient approaches, it converges quickly and is suitable for large datasets. The study uses the BFGS Quasi-Newton method to optimize methane steam reforming, focusing on minimizing performance along the search direction and determining the minimum point using MATLAB for optimization problems. The BFGS Quasi-Newton method, when combined with MATLAB tools, provides a robust and efficient method for data-driven process design in methane steam reforming, potentially enhancing hydrogen production efficiency.