A Statistical Model for Gears Noise Prediction in Gearbox

This paper develops a statistical model for gears noise prediction in gearbox. Experimental study is made to detect the relationship between the errors in gear manufacturing and transmission gear noise by means of online noise test bench. Based on the experimental statistics, a multiple linear regression model is developed showing the relationship between the errors in gear manufacturing and transmission gear noise. Ridge regression is used to eliminate the problem of colinearity. Experiments show that the predicted value is almost consistent with the actual value. It provides basis to reduce gear noise and optimize gear micro-parameters.