Abstarct: Energy plays a vital role in the sustainability of industrial processes, and despite advancements in renewable energy development, fossil fuels remain the dominant source of global energy supply. Among these, coal—particularly in the form of coke—holds a significant position in the steel industry due to its abundance, cost-effectiveness, and efficiency. Accordingly, this dissertation investigates the impact of intrinsic properties and preparation parameters of coal on coke quality indices using statistical methods and mextaheuristic algorithms. In the first phase, experimental data from the National Iranian Steel Company and relevant literature were utilized to design and conduct laboratory tests on three types of coking coal, focusing on their intrinsic characteristics. Subsequently, behavioral modeling was carried out using both linear and nonlinear regression methods in univariate and multivariate formats, resulting in the development of 18 models, from which two nonlinear multivariate models were selected as final representations. Following the modeling phase, Design of Experiments methods were applied to assess the influence of preparation parameters on coke quality. For optimal blend formulation, Gradient Boosting was employed to identify key variables, with ash content emerging as the most influential factor (importance score ranging from 66% to 85%). Support Vector Regression revealed a strong relationship (R² = 0.81) between ash content and coke porosity. Finally, an optimization model was developed by defining an objective function under three categories of constraints (coal quality, coke quality, and operational limits), and the Whale Optimization Algorithm was employed to derive an optimal coal blend. The outcome was a composition of three coal types with a minimum required amount and a consumption rate of 31%.