Establishing quantitative structure tribo-ability relationship model using Bayesian regularization neural network

Xinlei Gao,Junbo He,Zhan Wang,Kang Dai,Tingting Wang

doi:10.1007/s40544-016-0104-z

Abstract

Abstract Quantitative structure-activity relationship methods are used to study the quantitative structure triboability relationship (QSTR), which refers to the tribology capability of a compound from the calculation of structure descriptors. Here, we used the Bayesian regularization neural network (BRNN) to establish a QSTR prediction model. Two-dimensional (2D) BRNN–QSTR models can flexibly and easily estimate lubricant-additive antiwear properties. Our results show that electron transfer and heteroatoms (such as S, P, O, and N) in a lubricant-additive molecule improve the antiwear ability. We also found that molecular connectivity indices are good descriptors of 2D BRNN–QSTR models.

Highlights

To find effective tribological materials suitable for use under different working conditions, numerous experiments need to be performed, including modifying or synthesizing novel potential materials, testing their tribological performance, and screening several compounds
The performance of the Bayesian regularization neural network (BRNN)–quantitative structure triboability relationship (QSTR) models with 3D Jurs descriptors is poor because all of the three q2 values are less than 0.1, which indicates that 3D Jurs descriptors do not predict the wear scar data well
(2) We analyzed the influence of each descriptor on the antiwear property of lubricant additives, and proposed structural characteristics for improving the antiwear ability

Summary

Introduction

To find effective tribological materials suitable for use under different working conditions, numerous experiments need to be performed, including modifying or synthesizing novel potential materials, testing their tribological performance, and screening several compounds. We still largely rely on luck to find an effective lubricant. This approach is inefficient because the entire development process lacks theoretical guidance, resulting in a lack of direction in the development process. Finding a compound with good tribological properties from numerous possible structures by a random search and/or based on experience is an inefficient use of time and resources. It is well known that the physical, chemical, biological, and tribological properties of materials are determined by their structures. Studies on systematic and scientific methods for linking structures

Methods

Results

Discussion

Conclusion