摘要
为解决发动机异响信号低信噪比和时域波形图对故障特征表现不直观的问题,提出了一种免疫进化网络的诊断算法。以信号频谱差异作为故障的特征信息,将异响信号通过小波包分解方法构建成抗原群体,利用免疫进化网络的训练和学习,取得了较好的识别效果,实现了发动机的异响诊断。应用结果表明,该免疫神经网络具有优越的识别性能,在识别一般抗原的基础上,还可以有效识别由于设备状态波动引起的特异抗原,为故障诊断提供了一种准确、有效的分析手段。
In order to solve the low signal to noise ratio and the representation of fault feature being not so intuitive in time domain chart of engine abnormal sound signal,an immune evolution network diagnosis algorithm is proposed.With the signal spectrum difference as the fault characteristic information, for this abnormal sound signal, through wavelet packet decomposition method to build antigen group,it uses the immune neural network training and learning,and has achieved good effect,and realizes the engine abnormal sound diagnosis. Application results show： the immune neural network has the superior recognition performance,on the basis of general antigen recognition,also can effectively identify the equipment state fluctuations caused by specific antigen,for fault diagnosis provides an accurate and effective analysis means.
出处
《机械设计与制造》
北大核心
2015年第7期172-176,共5页
Machinery Design & Manufacture
基金
陕西省科技厅工业攻关项目(2014K05-47)
宝鸡市科技局工业攻关项目(13KG3-6)
宝鸡文理学院重点科研项目(ZK12021
YK1511)
关键词
免疫进化网络
发动机异响信号
识别方法
故障诊断
小波包频带分解
Immune Evolution Network
Engine Abnormal Sound Signal
Recognition Method
Fault Diagnosis
Decomposition of Wavelet Packet Frequency Band
