Indian Journal of Science and Technology
DOI: 10.17485/ijst/2018/v11i45/138198
Year: 2018, Volume: 11, Issue: 45, Pages: 1-10
Original Article
Milton Muñoz1 , Iván Amaya2 and Rodrigo Correa1*
1 Universidad Industrial de Santander, UIS, Santander, Colombia; [email protected], [email protected]
2 Tecnológico de Monterrey, Monterrey, Nuevo León, México; [email protected]
*Author for correspondence
Rodrigo Correa,
Universidad Industrial de Santander, UIS, Santander, Colombia; [email protected]
Background/Objectives: We present an alternative for estimating drying curves and diffusion coefficients of coffee beans (Castilla variety), based on global optimization strategies. Methods: Four optimization algorithms were tested for adjusting drying curves. Based on the parameters that were found, we determined the diffusion coefficient. Algorithms were tuned up on 11 non-linear systems, prior to using them for adjusting the curves. Their performance were assessed through error dispersion analysis, as well as through the number of evaluations of the objective function and run time. Findings: On non-linear systems, Particle Swarm Optimization (PSO) and Drone Squadron Optimization (DSO) exhibited the best performance in terms of error. When used for estimating drying curves, PSO, DSO and Genetic Algorithms (GA) achieved determination coefficients beyond 0.99. Even so, GA had the lowest run time. Applications: Our experiments offer an alternative with excellent precision for estimating parameters of the drying function and of its diffusion coefficients for different coffee beans.
Keywords: Coffee drying curve, Diffusion coefficient, Metaheuristic
Subscribe now for latest articles and news.
