Algoritmo genético aplicado ao problema do sequenciamento de produção: o problema do caixeiro viajante no setor de duas rodas do Polo Industrial de Manaus

Abstract

The increasing complexity of assembly lines, driven by the growing variety of motorcycle models, has posed significant challenges to production sequencing in the motorcycle industry. This study proposes an innovative approach to optimize production sequencing through the application of Genetic Algorithms, modeling the problem based on the classical Traveling Salesman Problem (TSP). The research was conducted in a two-wheeler manufacturing company located in the Manaus Industrial Hub, where model changeovers directly impact total production time. Experiments were carried out using three real-world assembly lines, and the results demonstrated the capability of Genetic Algorithms to significantly reduce changeover times, achieving average improvements of approximately 80% compared to randomly generated sequences. The implementation employed the DEAP library in Python, with careful calibration of parameters such as population size, mutation rate, and number of generations. The findings indicate that the proposed approach is effective, robust, and scalable, delivering substantial gains in operational efficiency. This study contributes by highlighting the potential of Genetic Algorithms as a strategic decision-support tool in industrial contexts and reinforces the feasibility of their practical application to complex combinatorial problems.