摘要
为提高数控成形铣齿生产率、降低成本和避免安全隐患,需要对刀具的磨损状态进行准确预测。首先基于电流监测法搭建了数控成形铣刀的磨损电流监测系统,然后确定BP神经网络中用于刀具磨损诊断的输入特征量和目标特征量,并应用Matlab软件对样本数据进行归一化处理和神经网络训练,最后利用遗传算法对BP神经网络模型进行优化。测试结果表明,刀具磨损状态预测率达92.78%以上,具有一定的工程应用价值。
In the process of CNC shaping milling,the prediction of the state of tool wear has important applica- tion significance to improve productivity, reduce scrap rate and avoid security risks. Based on the current moni- toring method,the detection system of the wear current of CNC forming milling cutter is set up. Then the input characteristic quantity and target characteristic quantity of BP neural network for tool wear diagnosis are meas- ured, and the sample data were normalized and trained by the Matlab software. At last,the genetic algorithm is used to optimize the BP neural network. The network test results show that the prediction rate of tool wear con- dition is more than 92.78% . This has certain engineering application value.
出处
《河南理工大学学报(自然科学版)》
CAS
北大核心
2017年第5期66-71,共6页
Journal of Henan Polytechnic University(Natural Science)
基金
国家"863"计划项目(2013AA040103)
国家自然科学基金资助项目(51175153/E050903)
关键词
遗传算法
BP神经网络
电流监测
刀具磨损
genetic algorithm
BP neural network
monitoring current
tool wear