A stack fusion model for material removal rate prediction in chemical-mechanical planarization process

摘要: In chemical-mechanical polishing process of wafers, the accurate prediction average material removal rate is vital for estimation time, which may significantly optimize production efficiency while maintaining acceptable quality. this study, a new stacking fusion model proposed, offers precise way to predict based on indirect sensor data collected from wafer process. Through procedure feature creation, expansion, and encoding, were transformed into multi-dimensional information. Then, through importance evaluation by following steps selection, subset that effective was selected. The accuracy primary learner optimized via establish highly non-linear mapping relationship between features rate. Compared with other weighted models neural network models, method presented improved precision under several working conditions. promising in terms becoming embedded machines enable an online real-time