多项式回归理论在机床热误差建模中的应用

Application of polynomial regression theory in thermal error modeling of machine tool

通过理论分析,利用多项式回归理论中的增广样本相关系数,结合复相关系数的方法剔除与因变量和其他自变量相关系数均很低的自变量,建立机床热误差补偿模型.该方法与逐步线性回归热误差建模法相比,可避免出现温度变量耦合现象,缩短建模时间.通过残差及拟合图验证该模型具有较高的回归精度.

Based on the theoretical analysis and by using the duplicate correlation coefficient together with that of augmentation sample correlation coefficient of polynomial regression, those independent variables could be rejected when they had low correlation coefficients with other independent variables and dependent variables, a robust thermal error compensation model was established. Compared with stepwise regression, this method could avoid to the coupling of temperature variables and reduce the modeling time. The model was confirmed by residual errors and fitting chart to have a higher regression precision.