Tunnel shaped parts with truncated pyramidal shapes were formed using Single Point Incremental Forming (SPIF) on a Stewart platform. The accuracy behavior of these parts was characterized by an error prediction response surface generated using Multivariate Adaptive Regression Splines (MARS). This response surface predicted over forming for low wall angle parts and under forming for higher wall angle parts. It is based on geometrical parameters associated with features on the part geometry and was used to compensate for inaccuracies in the part geometry. Feature detection was found to work well for tunnel shaped parts using similar thresholds as container shaped parts, while the maximum deviations were found to be lower at a wall angle of 60° compared to a part with wall angle 40°.
|Title of host publication||International Conference on Intelligent Computing for Sustainable Energy and Environment & International Conference on Intelligent Manufacturing and Internet of Things: Proceedings|
|Place of Publication||Chongqing, China|
|Number of pages||10|
|Publication status||Early online date - 05 Sep 2018|
|Name||Communications in Computer and Information Science|