Cost estimation for sheet metal parts using multiple regression and artificial neural networks: A case study

Increasing competition in sheet metal operations has urged those companies to search for tools that generate accurate cost estimates within a short time period. The requirement for on-line generation implies that the underlying cost estimate needs to be generated without extensive process planning first. Analysis has been conducted on developing a less-detailed method, based on a brief analysis of the CAD-file. Cost formulas are composed by applying regression techniques and neural networks. A case study is used to compare both methods. The results obtained indicate that neural networks give better results but are still mainly considered black boxes.