SVR-based analysis on tribological property of ultra high molecular weight polyethylene composites filled with nano-ZnO particles

This study develops support vector regression (SVR) models for describing the complex nonlinear relationship between tribological properties (friction coefficient and wear rate) and experimental factors including load, content of filled nanoparticles and speed of relative sliding for the ultra high molecular weight polyetwearhylene composites filled with nano-ZnO particles (UHMWPE/nano-ZnO). The particle swarm optimization (PSO) algorithm is employed for optimizing the parameters of SVR models and obtaining the optimal process parameters for preparing UHMWPE/nano-ZnO. The comparison of results achieved by SVR and multivariable linear regression (MLR) exhibits the superior simulation accuracy and generalization performance of the SVR approach. Meanwhile, multifactor analysis is adopted for investigation on the significances of each experimental factor and their influences on the tribological properties of UHMWPE/nano-ZnO. This study suggests that the SVR is an efficient and novel approach in development of the UHMWPE/nano-ZnO with lower friction coefficient and perfect wear resistance.