摘要
为了减少电主轴的热误差,提高数控机床的加工精度,考虑热变形不仅与自身若干期的滞后值有关,还与当前温升及其滞后值有关,提出采用多元自回归方法建立电主轴热误差模型。首先将观测序列进行差分处理,剔除线性趋势项,然后利用Akaike判据获得自回归模型的阶数,用最小二乘法求得自回归模型的系数,最后用建立的自回归模型预测电主轴热误差,并通过试验验证该模型的有效性。试验结果表明基于位移的热误差自回归模型比基于温度的热误差多元线性回归模型有更好的精度。
In order to reduce the thermal error of the motorized spindle and improve the manufacturing accuracy of NC machine tool, the thermal error model based on multivariate autoregressive method is proposed. This model con- siders not only the relationship between thermal deformation and its lagged value but also the relationship between thermal deformation and the present and lagged value of temperature rise. The linear trends of observed series are eliminated by numerical difference. The order of multivariate autoregressive （MVAR） model is adopted by using Akaike information criterion. The coefficients of the MVAR model are determined by the least square method. The established MVAR model is then used to forecast the thermal error and the experiments have shown the validity and robustness of this model. The results indicate that the displacement-based thermal error autoregressive model has much better accuracy than the temperature-based multiple linear regression model.
出处
《机械科学与技术》
CSCD
北大核心
2012年第9期1526-1529,共4页
Mechanical Science and Technology for Aerospace Engineering
基金
国家科技重大专项项目(2009ZX04001-015)
甘肃省自然科学基金项目(1010RJZA043)资助
关键词
热误差
多元自回归模型
电主轴
预测
thermal error
multivariate autoregressive model
motorized spindle
forecasting
