摘要
针对自动化加工中切屑生成过程的监控问题,研究了径向基函数神经网络在切屑形态图像识别中的应用,提出了面积比,欧拉数,伸长度等切屑形态图像的几何特征.以这些特征作为神经网络的输入矢量,利用径向基函数网络（RBF）,采用了递推最小二乘法训练该网络,并开发了相应的计算机程序.试验证明:此法具有良好的实时处理性和适应性,识别率达到90％.
Aiming at the problem of process monitoring of chip generating in automatic machining, chip shape recognition using radbas neural network was studied. Area ratio feature, Euler number feature and disperse degree feature etc. Geometry features of chip shape image were put forward and those features were as inputting vector of neural network. Radbas neural network was adopted and training the network using RLS. On the basis of experiment the calculating program was developed and it has excellent real time processing capability and adaptability. The recognizing ratio of the system reaches to 90%.
出处
《哈尔滨理工大学学报》
CAS
2002年第5期79-82,86,共5页
Journal of Harbin University of Science and Technology
