基于DBO-BP的机床圆度误差预测模型  

Optimization of Machine Tool Roundness Error Prediction Model Based on DBO-Optimized BP Neural Network

作　　者：谌相志 封志明[1] 宋繁 周转南 陈恒安 CHEN Xiangzhi;FENG Zhiming;SONG Fan;ZHOU Zhuannan;CHEN Heng’an(School of Mechanical Engineering,Xihua University,Chengdu 610039,China)

机构地区：[1]西华大学机械工程学院,四川成都610039

出　　处：《机械》2025年第3期29-36,共8页Machinery

摘　　要：针对机床几何误差测量及补偿繁琐费时等一系列问题,对机床几何误差预测进行研究,提出基于蜣螂算法(DBO)优化BP神经网络的机床圆度误差预测模型。基于球杆仪测量CNV-1050机床的几何误差实验数据,通过引入蜣螂优化算法优化BP神经网络的网络结构参数,建立了机床圆度误差预测模型,然后利用BP神经网络和蜣螂优化BP神经网络模型对机床的圆度误差进行预测对比,得出DBO-BP对圆度误差的预测效果远高于优化前的BP神经网络对圆度误差的预测,最后对机床几何误差进行了补偿,使机床圆度误差降低在工作允许范围内。结果表明,优化的DBO-BP神经网络建立的预测模型相较于原模型具有更强的预测能力和较强的鲁棒性,且在精度与稳定性上都较高,具有一定工程应用价值。Aiming at a series of problems such as the tedious and time-consuming measurement and compensation of machine tool geometric error,the machine tool roundness error prediction model based on BP neural network through Dung Beetle Optimizer(DBO)is proposed.Based on the geometric error experimental data of CNV-1050 machine tool measured by the ball meter,this paper introduces the Dung Beetle Optimizer to optimize the network structure parameters of BP neural network,establishes the roundness error prediction model of machine tool,and compares the performance the BP neural network and the DBO-optimized BP neural network model inpredicting the roundness error of machine tool.The results show that the DPO-BP performs significantly better than the unoptimized BP neural network.Finally,the geometric error of the machine tool is compensated,and the roundness error of the machine tool is reduced within the allowable working range.The findings indicate that the prediction model developed using the optimized DPO-BP neural network exhibits superior predictive power and resilience compared to the previous model with enhanced precision and consistency,and is valuable for engineering applications.

关 键 词：数控机床 几何误差 圆度误差 蜣螂优化算法 BP神经网络 

分 类 号：TH161.21[机械工程—机械制造及自动化]