Comparative study of response surface methodology and hybrid back-propagation network for optimizing friction coefficient for textured surface under cavitation conditions

Abstract

Purpose

The purpose of this study is to establish a friction coefficient prediction model using texture parameters and then using the optimal texture parameters to obtain the minimum friction coefficient.

Design/methodology/approach

Based on texture technology and the cavitation phenomenon conditions, a test scheme based on two-factor and five-level texture parameters is designed using central composite design and then the response surface methodology and hybrid back-propagation genetic algorithm (BP-GA) models are used to establish a friction coefficient prediction model and optimize the friction coefficient.

Findings

The result indicates that the values predicted using two methodologies agree well with the experimental data, but the hybrid BP-GA model is superior to the response surface methodology model in both prediction and optimization.

Originality/value

Two methodologies are used to study the influence of the texture parameters on the friction coefficient under the cavitation condition. It is expected that the result can be used to obtain optimum texture parameters to reduce the friction coefficient.

Citation

Mao, Y. and Zeng, L. (2018), "Comparative study of response surface methodology and hybrid back-propagation network for optimizing friction coefficient for textured surface under cavitation conditions", Industrial Lubrication and Tribology, Vol. 70 No. 5, pp. 856-864. https://doi.org/10.1108/ILT-06-2016-0137