Artificial Neural Network for Predicting Wear Properties of Brake Lining Materials
Many factors influence the wear of friction material performance such as formulation, manufacturing condition and operating regimes, and so on. In this paper, the wear rate variation has been modeled by means of artificial neural network, the network have been developed with all these relevant factors taking into consideration. 16 influence factors and wear rate selected as input and output respectly, 16 [10-8]2 1 is regarded as the best architecture of neural network, the Levenberg-Marquardt algorithm is used for training the network. The result shows that the model is valid to predict the wear property, as well as that it is useful for optimizing the formulation and manufacturing conditions, the relatively excellent combination of the ingredients and the appropriate manufacturing condition parameters can be obtained by this approach.
Liangchi Zhang, Chunliang Zhang and Zichen Chen
J. H. Han and Q. S. Wu, "Artificial Neural Network for Predicting Wear Properties of Brake Lining Materials", Advanced Materials Research, Vols. 328-330, pp. 237-240, 2011