基于自适应无味卡尔曼滤波的机床选点温升快速辨识方法研究（英文）  

A novel method for fast identification of a machine tool selected point temperature rise based on an adaptive unscented Kalman filter

作　　者：Chen-hui XIA Jian-zhong FU Yue-tong XU Zi-chen CHEN 

机构地区：[1]State Key Lab of Fluid Power Transmission and Control, Department of Mechanical Engineering,Zhejiang University

出　　处：《Journal of Zhejiang University-Science A(Applied Physics & Engineering)》2014年第10期761-773,共13页浙江大学学报（英文版）A辑（应用物理与工程）

基　　金：Project supported by the National Nature Science Foundation of China(No.51175461);the Science Fund for Creative Research Groups of National Natural Science Foundation of China(No.51221004);the Specialized Research Fund for the Doctoral Program of Higher Education(No.20120101110055),China

摘　　要：研究目的:为了缩短机床温升试验时间,提出一种机床热特性快速辨识方法,利用较短时间的温度采样数据即可准确预测出完整的温升曲线,进而获得热平衡时间及稳态温度等热特性参数。创新要点:提出了基于自适应无味卡尔曼滤波的机床选点温升快速辨识方法,其中最短辨识时间判据可以有效解决如何寻找准确辨识热特性参数的最短采样时间问题,而自适应无味卡尔曼滤波则可以实时调整参数,防止外界因素对辨识的干扰。研究方法:由于无味卡尔曼滤波在非线性状态预测和参数辨识上具有优势,所以本文将无味卡尔曼滤波算法应用到机床选点温升辨识上。为了防止辨识过程中的发散退化等问题,将无味卡尔曼滤波发展为自适应无味卡尔曼滤波(图1)。在快速辨识方法上提出了最短辨识时间判据(图2)。文章中又将此算法应用到实际的立式加工中心温升辨识上,证明了该算法的可行性及有效性(图5和6)。最后又将带有自适应调整过程的无味卡尔曼滤波算法和不带调整过程的算法做了对比,显示了自适应调整过程对辨识算法的重要性(图6和11)。重要结论:基于自适应无味卡尔曼滤波的机床选点温升快速辨识方法可以准确快速地辨识出温升曲线,获取热特性参数,将原来394 min的热平衡试验时间缩短,只需28 min即可得到温升变化情况。A novel method is presented for fast identification of a machine tool selected point temperature rise, based on an adaptive unscented Kalman filter. The major advantage of the method is its ability to predict the selected point temperature rise in a short period of measuring time, like 30 min, instead of 3 to 6 h in conventional temperature rise tests. A fast identification algorithm is proposed to predict the selected point temperature rise and the steady-state temperature. An adaptive law is applied to adjust parameters dynamically by the actual measured temperature, which can effectively avoid the failure of prediction. A vertical machining center was used to validate the effectiveness of the presented method. Taking any selected point, we could identify the temperature rise at that point in 28 min. However, if the method was not used, it took 394 min to obtain the temperature rise curve from the start-up of the machine tool to the time when it reached a steady-state temperature. The root mean square error（RMSE） between the estimated and measured temperatures in the period of 394 min was 0.1291 °C, and the error between the estimated and measured steady-state temperatures was 0.097 °C. Therefore, this method can effectively and quickly identify a machine tool selected point temperature rise.

关 键 词：温升 快速辨识 自适应无味卡尔曼滤波 机床 

分 类 号：TG502.1[金属学及工艺—金属切削加工及机床]