Método para redução de tempo computacional e melhoria de flexibilidade de demanda no balanceamento de linhas de soldagem robotizadas

Abstract

In this work, an alternative approach is proposed to decrease the computational time expended to reach the solution of a mathematical model in Mixed-Integer Linear Programming present in the literature to perform the production balancing of a robotic manufacturing line in the automotive industry. For this, it is proposed to divide the initial problem into subsets and seek the optimization of each one of these groups, individually. Dividing the problem into parts implies a considerable reduction in the number of variables and restrictions in each group. To verify the method, a practical case from the industry was tested. As a result, a reduction of approximately 10 times in computational time was obtained, with the same cycle time response. In addition, an analysis of the flexibility of the robots in this production line was carried out to absorb possible fluctuations in market demand. For that, some scenarios of volume reduction were proposed by reducing the number of robots. Cases were analyzed in which the robots had different accessibility capabilities to the welding regions. As a result, it has been found that robots with greater flexibility have up to twice the impact on the cycle time of the production line compared to those with very specific welding capabilities.