ARTIFICIAL NEURAL NETWORKS BASED PREDICTION OF FRICTION COEFFICIENT

摘要

This paper investigates the utilization of artificial neural networks to predict the friction coefficient on the contact face between the workpiece and fixture element taking into account factors such as surface roughness, applied normal load, workpiece material, fixture element type etc. Experimental data was used to train and validate the neural network model. Different neural network architectures were tested. The predicted results compares well with experimental results. The approach ensures estimation of the friction coefficient in real time which is needed for simulation of the machining process in general and for fixture configuration selection and optimization in particular.