基于小波神经网络的工程陶瓷动态车削力预测  

Prediction of Engineering Ceramics Dynamic Cutting Force Based on Wavelet Neural Network

作　　者：李琛[1] 尤轲 田俊超 马廉洁[1] 

机构地区：[1]东北大学秦皇岛分校控制工程学院

出　　处：《工具技术》2015年第9期81-84,共4页Tool Engineering

基　　金：国家自然科学基金(51275083)

摘　　要：以小波分析和BP神经网络为基础,构建了小波神经网络预测模型。使用CA6140车床对氟金云母陶瓷进行了干车削试验,并用三向测力仪测量了切削过程的切削力变化趋势。基于小波包中的Wpbmpen函数对切削力信号进行了降噪处理,切削力信号在降噪后有明显改善,能更形象地表达出切削力的变化趋势。基于小波神经网络对切削力进行了预测,结果表明:小波神经网络预测值、信号降噪处理值和试验值都非常相近,说明切削力在预测过程中具有一定的可靠性,小波神经网络预测前对切削力信号的降噪处理是合理的。The wavelet neural network prediction model was constructed based on wavelet analysis and BP neural net- work. Fluorinephlogopite ceramics for dry cutting experiment was carried out by the CA6140 lathe, and the trend of cutting force was measured by three to the dynamometer. The signal of cutting force was improved obviously after noise reduction, and the trend of cutting force can express vividly. The cutting force signal was in noise reduction processing based on the wavelet packet wpbmpen function. The cutting force was forecasted based on wavelet neural network. The resuhs showed that the wavelet neural network, signal de-noising processing, estimated values and experimental values were very close. The cutting force in the process of prediction had certain reliability, and wavelet neural network prediction of cutting force signal noise reduction before prediction was reasonable.

关 键 词：车削力 预测 小波神经网络 工程陶瓷 小波降噪 

分 类 号：TG501[金属学及工艺—金属切削加工及机床] TH161[机械工程—机械制造及自动化]