数控机床电主轴热误差的预测方法

为了减少电主轴的热误差,提高数控机床的加工精度,对于时变速度的主轴运转,分别采用多元自回归方法和遗传径向基函数神经网络方法建立电主轴热误差预测模型.根据2种模型对电主轴热变形产生机理的不同表述形式,比较二者的计算效率和拟合精度.研究表明:在相同温升变量的条件下,二者的收敛速度和运算时间相差无几；在预测精度方面2种建模方...

Forecasting method for thermal error of motorized spindle in NC machine tools

In order to reduce the thermal error of the motorized spindle and improve the manufacturing accuracy of NC machine tool,the thermal error forecasting models were proposed based respectively on multivariate autoregressive(MVAR) method and genetic neural network(GANN) method.According to their different representations of generation mechanism of motorized spindle thermal deformation,the computation efficiency and curve fitting precision of these two models were compared.The investigation showed that under the condition of identical temperature rise,MVAR model and GARBF neural network model exhibited almost the same convergence and computation time;relative errors of two models were less than 3%.However,the estimation ranges of two models were different;the MVAR model exhibited higher forecasting precision in short-term prediction and,on the other hand,the GARBF neural network model exhibited higher forecasting precision in mid-long term forecasting.