货车滚动轴承的声发射故障诊断系统研究-测试计量技术及仪器专业论文.docxVIP

河南科技大学硕士学位论文SYSTEM 河南科技大学硕士学位论文 SYSTEM RESEARCH ON FAULT DIAGNOSIS T0 THE ROLLING BEARING OF FREIGHT CAR BY AE SIGNAL AB STRACT The rolling bearing 3 52226X2—2Z is one of the mechanical parts of freight car，which i s easiest to be damaged．Its running state di rectly influences the security of railway．In order to discovery bearing fault earlier，the research in the previous stage brought forward a diagnosis method with acoustic emi ssion and applied it to fault diagnosi S of the rolling bearing．To recognize bearing condition，wavelet analysi s was applied to withdraw the peak-value parameter of AE signal s in time． domain and frequency—domain．Thi s method offers a way 0f fault diagnosi s．However，energy di stribution of AE signal i S wider，the research fo r 900Hz一1．4KHz of AE signals in the earlie r period have a certain limitation in obtaining effective information of the signal s In view of above question，this article has carried on the repeated experiments for 3 52226X2—2Z bearing with the typical condition (normal，inne r—ring fault and roller fault)．With contrastive analy sis of time—domain and frequency—domain of AE signal，this article proposed one kind Of damage examination of Wavelet Probabili stic Neural 一，Network(WPNN)．This method makes use of wavelet packet analy sis to 一’‘decompose the energy of the signal in different frequency band s and recognize s the condition of rolling bearing of the freight vehicle through Probabili stiC Neural Network． The results of research indicate that the condition recognition of rolling bearing，based on wavelet packet analysis to the energy II 摘要di 摘要 di stribution in different frequency band s，is feasible．At the same time， the PNN in the fault diagnosis has not onlY higher training speed and stronger robu stness，but al SO realizes easily in the project．The WPNN enhances the diagnosi S veracity and intelligence of freight ca r． KEY WORDS：Rolling Bearing，Acoustic Emission，Wavelet Packet AnalysiS，Probabili stic Neural