Deep Drawing Investigation Using Neural Network and Data Mining

Nowadays better production with minimum effort and lower cost is vital requirement of industrial production. So all proved solutions in decreasing cost and enhancing quality has also close competition with each other. Data mining and neural networks are both eminent methods in estimating the objective functions in manufacturing, so this research is aimed to develop an integrated methodology based on Finite Element Method (FEM) - data mining in first part, FEM-neural networks in the next to approximate the results of considered process parameters. In this methodology, the FEM simulation is first used to create training cases for two mentioned algorithms, and the well-trained algorithms (Support Vector Machine in data mining method and Levenberg-Marquardt in Neural Network method) are used to predict the wall thickness. Comparison between methods is done and data mining shows good accuracy but neural network shows better precision. To validate the FEM results, related mold was made and FEM shows good conformity with experiment.

Keywords:

Data Mining , Deep Drawing , FEM , Levenberg-Marquardt Neural Network , Support Vector Machine