基于SOM神经网络和复相关系数结合的机床主轴温度测点的优化筛选  

Optimal Selection of Temperature Measuring Points for Spindle of Machine Tool Based on Method of Combining SOM Neural Network and Multiple Correlation Coefficient

作　　者：王战中[1] 孙少华[1] 

机构地区：[1]石家庄铁道大学机械工程学院,河北石家庄050043

出　　处：《石家庄铁道大学学报（自然科学版）》2017年第1期95-98,共4页Journal of Shijiazhuang Tiedao University(Natural Science Edition)

摘　　要：提出了基于SOM神经网络和复相关系数相结合的温度测点的优化算法,并应用于SV-48立式加工中心主轴测温点优化筛选。首先,在主轴上模拟布置温度传感器和Z轴位移传感器,在有限元分析的基础上得到了一系列温度和Z轴热位移仿真数据;然后,将温度数据输入到SOM神经网络聚类分组;最后利用复相关程度法将聚类的温度值与主轴Z轴热误差拟合,确定出机床热敏感点。研究结果表明,该方法简明易懂,有效减少了测温点的数量。A novel optimization algorithm based on method of combining the SOM neural network and multiple correlation coefficient for identifying temperature measuring points is presented, which is applied to the main spindle of SV-48 vertical machining center. Firstly, a series of temperature sensors and Z-direction displacement sensors are placed on the spindle in simulation state and a series of temper- ature simulation data and Z-direction thermal displacement simulation data are obtained; A series of tem- perature simulation data are then clustered by SOM neural network. Finally, the temperature value after clustering and the spindle Z-direction thermal error are fitted by using complex correlation method, then, the thermal sensitive points of the machine tool are determined. The result shows that the method is clear and easy to understand, and effectively reduces the number of temperature measurement points.

关 键 词：SOM神经网络 复相关系数 热敏感点 优化算法 

分 类 号：TG502.15[金属学及工艺—金属切削加工及机床]