Improved BP Network to Predict Bending Angle in the Laser Bending Process for Sheet Metal

摘要

In this paper BP network was improved based on the Double Chains Quantum Genetic Algorithm (DCQGA). The predicted model of laser bending angle based on our proposed BPN-DCQGA network was set up in the process of sheet metal laser bending. The BPN-DCQGA network was trained and verified through the sample data result from experimental data, and it is proved that our proposed network has enhanced the convergence rate, gained higher train efficiency and stronger capability to find optimal solution, so as to predict the bending angle more accurately. Moreover, based on the mentioned model, parameters optimization system of laser bending was found, with strong robustness and capability to predict the bending angle as well as optimize the process parameters. This system will largely benefit for the manufacture and drive the application of laser bending.