Otimização multiobjetivo de fluxos rodoviários em terminais intermodais de grãos com inteligência artificial e simulação

Abstract

The growing demand for logistical efficiency in the Brazilian agribusiness sector imposes significant challenges on the operation of intermodal grain terminals, especially under increasing pressure for sustainability, fairness, and responsible resource management. This dissertation proposes an integrated approach based on Discrete Event Simulation (DES) coupled with multiobjective optimization algorithms, with emphasis on the Non-dominated Sorting Genetic Algorithm II (NSGA-II), to enhance operational performance while mitigating social and environmental impacts. The proposed model simultaneously considers five conflicting objectives: (i) truck dwell time, (ii) unloading throughput, (iii) asset utilization, (iv) average operator fatigue, and (v) inequality in waiting times. In addition to computational modeling, the study incorporates ethical and human-centered criteria into the evaluation of solutions, aligning with the principles of Industry 5.0 and algorithmic governance. The results demonstrate the feasibility of identifying non-dominated solutions that balance efficiency, distributive justice, and sustainability. Despite progress in the literature on simulation and artificial intelligence applied to logistics, a significant research gap remains: few studies offer validated, multidimensional models that support real-world decision-making in complex operational settings such as intermodal grain terminals. This study contributes to bridging this gap by proposing and applying a replicable, ethically grounded, and technically robust model with potential for practical and institutional impact.