By using genetic algorithms and radius basis function (GARBF) neural network, the predicting model of friction coefficient has been established based on a measured database with five sliding velocities of 40, 55, 70, 85, 100 m/s and four different normal pressures of 0.1333, 0.4667, 0.60 and 0.7333 MPa. The modeling results confirm the feasibility of the GARBF network and its good correlation with the experimental results. The predictive quality of the GARBF network can be further improved by enlarging the training datasets and by optimizing the network construction. A well-trained GARBF modeling is expected to be very helpful for selecting composite component under different working conditions, and for predicting tribological properties. Finally, by using GARBF modeling data to predict analysis, the results show that the friction coefficients of these composites were increased with the increase in material thermal capability at some region.
Artificial Neural Network and Genetic Algorithms for Discontinuously Reinforced Aluminum Composites Frictional Behaviour Prediction
Qiu, M, Zhang, Y, & Zhu, J. "Artificial Neural Network and Genetic Algorithms for Discontinuously Reinforced Aluminum Composites Frictional Behaviour Prediction." Proceedings of the World Tribology Congress III. World Tribology Congress III, Volume 1. Washington, D.C., USA. September 12–16, 2005. pp. 223-224. ASME. https://doi.org/10.1115/WTC2005-63159
Download citation file: