Cycle Time Reduction in Machining under Modeling Uncertainty

Optimizion of machining operations is important for increased productivity. However, modeling uncertainty precludes the application of traditional optimization methods. Moreover, stochastic optimization methods, which are designed to cope mainly with the random aspect of proceses, are not suitable due to their inability to cope with `modeling bias´. The purpose of this paper is to introduce an uncertainty-tolerant optimization method (UTOM) that can cope with process noise as well as `modeling bias´. Modeling bias, in UTOM, is accounted for by considering a bias range for individual parameters of the model. Noise is incorporated by including random components in the relationships defining individual constraints. UTOM´s main contribution is its utilization of measurements at the end of each cycle to continually improve cycle time based on nonlinear programmin, as well as its use of a monotonicity analysis for determining the extremes of bias and noise so as to guarantee constraint satisfaction. UTOM is investigated in simulation for a turnig operation. The results indicate that it is effective in reducing cycle time despite inaccuracies in the model and the presence of noise in process measurements.