【摘要】 为了提高数控机床的加工精度,文章以精密四轴数控平台为研究对象,采用PT100、激光干涉仪等仪器对X、Z轴的温度、定位误差进行测量与分析,研究精密四轴数控平台定位误差与温度之间的变化规律。运用支持向量回归机建立X、Z轴的热误差模型,利用网格搜索法对支持向量回归机热误差模型进行参数寻优,确定惩罚参数c和核函数参数g的最优参数值。在热平衡状态下,根据BP神经网络、支持向量回归机热误差模型分别计算出X、Z轴定位误差的预测值与测量值对比曲线,对比曲线和数据分析表明支持向量回归机的预测精度较高,其X、Z轴拟合偏差带宽均不超过0.6μm。依据支持向量回归机热误差模型的预测数据进行补偿实验,数控平台X轴的定位误差降低了89.55%,Z轴定位误差降低了85.67%。实验结果证明支持向量回归机建模方法具有较高的预测精度、泛化能力、补偿精度和鲁棒性。更多还原

【Abstract】 To improve the accuracy of CNC machine tools,precision four axis CNC platform for the study,Several kinds of instruments such as laser interferometer,temperature sensor were used to repeatedly measure and analyze the temperature and positioning errors of X,Z axes,found that change rules of between position error and temperature change for the four axis CNC precision in X,Z axes.The thermal error model of X,Z axes was established based on support vector regression,the grid search method for support vector regression thermal error model parameter optimization,and then to determine the penalty parameter c and kernel function parameter g optimal parameter values.In the thermal equilibrium state,according to the BP neural network,support vector regression machine thermal error model to calculate the prediction data comparing curves of X,Z-axis positioning error.Comparison curves and data analysis showthat the prediction accuracy of support vector regression machine is high,and the fitting bandwidth of X,Z-axis value was verified to be less than 0.6μm.A compensation experiment was carried out according to X,Z-axis positioning error of prediction data,and CNC platform positioning errors of X-axis reduced by 89.55%,and the Z-axis positioning errors are respectively reduced by 85.67%.The experimental results demonstrate that support vector regression modeling method has higher prediction accuracy and generalization ability,compensation accuracy and robustness.更多还原