IDENTIFICATION OF MANUFACTURABILITY CONSTRAINTS THROUGH PROCESS SIMULATION AND DATA MINING

With continued demands on high-quality and low-cost products, designers are increasingly required to explore the use of new materials and manufacturing processes. The design for manufacturability rules for unconventional materials and processes will be of great value to improve component manufacturability. Instead of going through years of trial-and-error practice to gain some "rule of thumb" design guidelines, this paper proposed a knowledge-based computational method for manufacturability constraint modeling (MCM) through process simulation, design of experiment, and data mining. With the input of geometric attributes for local critical features of a component, the pre-trained manufacturability constraint model will output the manufacturability prediction and the confidence of the prediction. The 2D visualization of the manufacturability prediction facilitates the interpretation by the human designers, and provides her with concurrent and intuitive manufacturability feedback and directed re-design suggestions. The preliminary result on mild steel stamping process demonstrated the feasibility of the method.