首页 | 本学科首页   官方微博 | 高级检索  
     检索      

难加工材料切削状态的识别——模式识别技术的应用
引用本文:张幼桢,潘良贤,刘肇发,吴殿宗.难加工材料切削状态的识别——模式识别技术的应用[J].航空学报,1983,4(1):87-94.
作者姓名:张幼桢  潘良贤  刘肇发  吴殿宗
作者单位:南京航空学院 (张幼桢,潘良贤,刘肇发),南京航空学院(吴殿宗)
摘    要: 本文介绍模式识别技术在难加工材料切削状态识别中的应用,这是实现适应性控制的第一步。根据切削试验中切削力随机信号处理的结果,可以看出在自相关图和自功率谱图与切削现象之间存在着内在联系。在此基础上分别建立了有无积屑瘤以及将切屑分成三类的判别函数。利用这些判别函数来识别切削状态的结果表明,模式识别技术在切削加工中的应用是成功的。

关 键 词:
收稿时间:1982-04-01;

RECOGNITION OF CUTTING STATES FOR DIFFICULT-TO-CUT MATERIALS——APPLICATION OF PATTERN RECOGNITION TECHNIQUE
Zhang Youzhen,Pan Liangxian,Liu Zhaofa,Wu Dianzong.RECOGNITION OF CUTTING STATES FOR DIFFICULT-TO-CUT MATERIALS——APPLICATION OF PATTERN RECOGNITION TECHNIQUE[J].Acta Aeronautica et Astronautica Sinica,1983,4(1):87-94.
Authors:Zhang Youzhen  Pan Liangxian  Liu Zhaofa  Wu Dianzong
Institution:Nanjing Aeronautical Institute
Abstract:Application of pattern recognition technique to recognition of cutting states for difficult-to-cut materials is introduced in this paper.It is the first step for realization of the adaptive control in cutting processes.A pattern dichotomizer can be adopted to judge whether BUE appears.A more general model of pattern classifier must be designed in order to classify chip shapes into three categories.The major problem in classifier design is the determination of discriminant function or weight vector.First of all,it is necessary to select the components or the features of the pattern vector properly by experience.The results of processing random signals in cutting experiments show that there exist relation-shaps between cutting phenomena and autocorrelation function or auto power spectrum.Therefore both the cutting parameter and the character of autocorrelation function or that of auto power spectrum can be selected as the components of pattern vector.It is possible to find out the discriminant function or weight vector by learning process with the aid of electronic computer.There were more than 60 different cutting states observed in cutting experiments.The discriminant functions determined by several training patterns have been used to check the remaining patterns,and the results show that the recognition of cutting states is basically successful and the rate of success is greater than 70%.The next step of our work will be to enhance the accuracy of recognition.
Keywords:
本文献已被 CNKI 等数据库收录!
点击此处可从《航空学报》浏览原始摘要信息
点击此处可从《航空学报》下载免费的PDF全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号