机床直线轴温度测点筛选与热误差预测方法

Method of Temperature Measuring Point Screening and Thermal Error Prediction for Feed Axis of Machine Tool

为揭示数控机床直线进给轴温度场信息与热误差之间的非线性映射关系,提高热误差预测模型效率和精度,提出了基于鹈鹕优化算法(POA)与随机森林(RF)的直线轴热误差预测方法。首先,采用Featts聚类算法将温度测点进行聚类分组,通过Spearman相关性分析计算温度与热误差之间的相关程度及热敏感点;其次,使用POA优化算法对RF模型参数进行寻优,确定RF网络的最佳决策树数量及叶子数等;最后,在三轴铣床加工中心上进行实验验证。结果表明,使用Featts聚类算法与Spearman相关分析方法提高了模型精度,有效避免温度测点间的多重共线性。与传统的BP神经网络及RF网络相比,POA-RF预测网络的均方根误差分别降低了46%和43%。

To reveal the nonlinear mapping relationship between the temperature field information of the feed axis of the CNC machine tool and the thermal error,and to improve the efficiency and accuracy of thermal error prediction model,a thermal error prediction method of linear feed axis based on pelican optimization algorithm(POA)and random forest(RF)was proposed.First,the Featts clustering algorithm was used to cluster and group the temperature measurement points,Spearman correlation analysis was used to calculate the correlation degree between temperature and thermal error and heat sensitive points.Secondly,POA optimization algorithm was used to optimize the parameters of the random forest model to determine the optimal number of decision trees and leaves of the RF network Finally,experiments were carried out on a three-axis milling center.The results show that the Featts clustering algorithm and Spearman correlation analysis are used to improve the accuracy of the model and effectively avoid multicollinearity among temperature measurement points.Compared with traditional BP neural network and RF network,the root-mean-square error of POA-RF prediction network is reduced by 46%and 43%respectively.