基于改进RBF网络的加工中心主轴热误差建模研究

Research on Modeling of Machining Center Spindle Thermal Error Based on Improved RBF Network

加工中心的主轴热误差一直是影响加工精度的主要因素,针对这一问题,本文提出了一种基于改进PSO优化RBF神经网络的机床热误差建模方法 ,对主轴热误差进行精确预测。首先使用优化的自组织映射(Self-organizingMaps,SOM)神经网络算法和相关系数法筛选关键测温点。然后分别使用RBF神经网络和改进的RBF神经网络建立热误差模型,使用MATLAB软件进行实验仿真,并对其结果进行对比。结果表明:改进后的RBF神经网络模型其预测精度较改进前有明显提高,也为后续误差补偿提供科学理论依据。

The machining center thermal error of the spindle is the main factors affecting the machining precision.To solve this problem,this paper proposes a method of thermal error modeling for machine tools based on RBF neural network optimized by improved PSO and forecasts thermal error of the spindle accurately.Firstly,the optimized SOM neural network algorithm and correlation coefficient method are used to select the critical temperature measurement points.Then the RBF neural network and the improved RBF neural network are used to establish the thermal error model respectively. Using MATLAB software to make simulation experiment and compare there results.The results show that the prediction accuracy of the improved RBF neural network model is better than that before the improvement.It also provides a scientific theoretical basis for subsequent error compensation.