Fault Diagnosis of CNC Machine Using Hybrid Neural Network

被引：0

作者：

Du, Juan ^{[1
]}

Yan, Xianguo ^{[1
]}

Wei, Nasha ^{[1
]}

机构：

[1] Taiyuan Univ Sci & Technol, Coll Mech Engn, Taiyuan 030024, Shanxi, Peoples R China

来源：

MEASURING TECHNOLOGY AND MECHATRONICS AUTOMATION IV, PTS 1 AND 2 | 2012年 / 128-129卷

关键词：

CNC machine; Fault diagnosis; BP neural network; ART neural network;

D O I：

10.4028/www.scientific.net/AMM.128-129.865

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 ;

摘要：

This paper proposed a technology of fault diagnosis for CNC machine based on hybrid neural network. Before the fault diagnosis was made, the fault patterns were firstly obtained from technical manuals and field experience to build a sample set, and then they were classified and coded according to the rule. In order to improve diagnosis speed, the fault diagnosis system was designed as a hybrid neural network system which consists of two-grade neural networks. When the fault pattern was input the system, the fault was first classified by the first-grade BP network, and according to the fault type, the corresponding second-grade ART network was activated to perform fault diagnosis. In this paper, the train algorithms of two kinds of neural networks were programmed by MATLAB. Comparing with the traditional diagnosis method, the presented technology possesses advantages of automatic fault diagnosis and ability for self-learning and self-organization.

引用

页码：865 / 869

页数：5