Forming Force Prediction in Two Point Incremental Forming Using Backpropagation Neural Networks in Combination With Genetic Algorithms

Abstract

Process feasibility and material formability are two of the most concerning factors which are studied in Incremental Sheet Forming (ISF) process. They are two important matters under study of the scientific community with the final goal of improving the process performances and to make it reliable to allow its use for rapid prototyping and for small batch production. In recent years, the application of neural network techniques to forming processes has been a research topic for optimizing and predicting process parameters and ISF process has not been an exception. Unfortunately, there are some drawbacks in using Backpropagation (BP) neural networks, such as easily falling into local minimum point. Because of that, recently Genetic Algorithms (GA) are applied to BP networks in order to optimize them. This paper describes the application of BP networks and GA applied to BP networks for the prediction of the forming force in ISF process. These models have been developed using experimental data from several tests. The two point incremental forming (TPIF) tests were performed on a horizontal CNC milling machine and DC04 deep drawing sheets steel 0.8 mm thick were worked. The reliability of the proposed approach has been evaluated by testing the trained networks with unseen experimental data and the comparison between the two BP and GA-BP models are presented. All the details are accurately discussed in the paper.