The lubrication analysis of the piston skirt-liner system based on design of experiment and particle swarm optimization-support vector regression

Abstract

Good design for the piston skirt can improve efficiency without harming emissions, while it simultaneously reduces wear and improves the reliability of the engine. In this paper, the finite element method is applied to solve the Reynolds equation to analyze the piston skirt-liner system lubrication, and the side motion of piston skirt solved with Runge–Kutta is coupled with lubrication. Some factors that have an effect on the lubrication and dynamic characteristics are selected as variables, such as the clearance, offset distance of the piston pin, bump position, curvature, the length of piston skirt, and the ellipticity of the piston skirt. The orthogonal experimental design which includes six factors with five levels is used to analyze the dominance of these structure factors, and the effects of all the responses for the structural parameters to the friction loss are also identified, and the orthogonal samples simulated by the experiment are conducted to do the regression for the piston skirt. Some regression models are introduced to predict the friction loss, the precise comparison of all the models are compared, and the larger error appears in these prediction models, and then the particle swarm optimization-support vector regression is also used to predict the friction loss, and the results agree well with each other. All the analyses are very useful to provide guidance for the design of the piston skirt-liner during the development of the engine.