Optimization-based parameter identification of a coffee fermentation model using evolutionary algorithms

This paper presents the parameter identification of a coffee fermentation model with different heuristic methods. A discrete-time linear model is proposed to relate dynamical reactions between relevant variables such as pH, glucose concentration, and lactic acid concentration. An optimization-based parameter identification is also proposed using three evolutionary algorithms: i) Particle Swarm Optimization (PSO), ii) Genetic Algorithms (GA), and iii) Differential Evolution (DE). These approaches are used to find the best fitting for experimental data. Results are compared in terms of cost function, curve fitting, solution dispersion, and convergence rate. The best results are obtained with PSO algorithm. The model can be used for monitoring, predict, and improve coffee quality for similar fermentation processes, coffee beans varieties, and agroecological conditions. Copyright (C) 2021 The Authors.