Abstract
Determination of the bending moment and springback coefficient of thin-walled steel and aluminium alloy profiles, as a function of the bending curvature, was the main objective of the work. The time consuming process of determination of distortion of cross section, which affects bending process of thin wall products, was eliminated. This objective was achieved by an application of the artificial neural networks. Multi-layer Perceptron (MLP) was trained using measured process data of profile bending. The MLP had profile geometry and mechanical parameters of material as input, and bending moment as well as spring-back coefficient as output. It was confirmed that this system is a valid alternative for the quick responsible method of main bending parameters determination.