工况信息融合的神经网络刀具监控方法研究  

Study on Neural Network Method for Cutting Tool Monitoring Based on Syncretic Machining Signals

作　　者：于广伟[1] 贾建军[1] 

机构地区：[1]西安工业大学

出　　处：《工具技术》2009年第5期60-63,共4页Tool Engineering

摘　　要：基于加工过程中刀具产生的动态信号,利用BP神经网络多输入、多输出和非线性映射的特性,通过融合多种加工特征信号,建立了切削参数与加工动态过程之间的关系模型,实现了刀具在线加工状况的检测与预报。仿真结果表明,基于工况信息融合的神经网络刀具监控方法不但可以减少加工参数变化对刀具状态检测的影响,而且提高了在线检测刀具磨损量的精确度,验证了该方法的有效性。Based on the machining signals generated by cutting tool in the automatization manufacturing system, by means of characteristics of BP neural network, such as muhi-input, multi-output and nonlinear relation, through syncretizing manifold machining signals, the model of relationship between cutting parameters and machining conditions was established, so that the on-line monitoring and prediction of cutting tool state were implemented. The simulating results indicate that this improved neural network method based on syncretic machining signals can not only reduce the influence of machining parameter change on tool monitoring condition, but also improve the precision of on-line monitoring tool wear values, and the validity of this method was proved.

关 键 词：神经网络 刀具磨损 在线监控 非线性映射 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TU712.2[自动化与计算机技术—控制科学与工程]