Improved multi-objective decision-making in manufacturing processes through uncertainty quantification and robust pareto front modelling

Manufacturing processes often exhibit complex relationships between input parameters and output responses, posing challenges for optimization and decision-making. Surrogate models are commonly employed to approximate these relationships, enabling efficient exploration of the design space. However, the presence of uncertainty can lead to suboptimal decisions and inaccurate predictions. This paper explores using Probabilistic Graphical Models to represent manufacturing processes, integrating expert knowledge with data-driven approximations of unknown relations. We investigate the impact of aleatoric uncertainty on multi-objective optimization, demonstrating enhanced Pareto front creation and improved decision-making under non-linear process functions. The methodology is applied to a continuous manufacturing case, where probabilistic surrogate sampling generates a more conservative setpoint. Our findings demonstrate that incorporating aleatoric uncertainty into surrogate modelling leads to more reliable and robust decision-making in manufacturing optimization.