摘要
研究和分析了自组织映射(SOM)神经网络的结构和算法,把SOM网络应用在旋转机械故障诊断中,利用振动传感器拾取振动信号,通过对时域、频域的分析来提取特征。通过对输入样本的"聚类",实现对故障的自动分类。这种故障的分类通过MATLAB更容易实现可视化的界面。仿真结果表明该方法可以对故障进行有效、准确地诊断,从而为旋转机械的故障诊断提供了一种新的途径。
The construction and algorithm of aself-organizing maps neural network are researc hed and analyzed.The SOM network applies to the fault diagnosis of rotating machinery.The use of vibration sensors pick up vibration signals,and through the time domain,frequency domain analysis to extract features.To realize the automatic classification diagnosis through th＂eclustering＂of input sample. This classification of fault is easier to achieve visual interface.The results show that the method can effectively and accurately diagnose the faults.Therefore,a new way is provided for the rotating machinery common fault diagnosis.
出处
《机械设计与制造》
北大核心
2010年第11期44-46,共3页
Machinery Design & Manufacture
基金
陕西省自然科学基金项目(2007E218)
陕西省教育厅自然科学专项(09JK559)
