Data-Driven Evaluation of Predictive Models for Estimation of Springback in Backward Metal Flow Forming Process
摘要
Springback is common defect when soft materials are formed during backward metal flow forming process. Springback is a combinational effect of roller geometry, process parameters and material formed during the process. This research focus on determination of springback by different soft computing models such as gradient boost, XGboost, adaboost, random forest and back propagation neural network. The models are statistically evaluated to determine the best model. A pictorial representation is given below.