[博士生国际交流】贾峰参加高水平国际会议汇报

时 间：2015年12月9日（周三）上午9：00–9：30

地 点：交大曲江校区西五楼南楼205会议室

报 告 人：博1417班---贾峰

学号：4114001029

指导教师：雷亚国 教授

2 参加国际会议信息

会议名称：COMADEM 2015+X CORENDE

会议日期：1-4 December, 2015

会议地点：Buenos Aires, Argentina

会议简介：The Congress intends to invite and integrate people involved in this study area in an open forum. The purpose of the congress is to encourage and facilitate exchange of information and experiences from all regions of the world. COMADEMis the opportunity to get in touch with technologies that continuously improve and enhance the quality, reliability, safety, availability, maintainability and performance of all assets (both physical and human) for as long as possible and to derive maximum benefits with minimum risk. CORENDEis a forum to discuss the improvements in the technologies used in the evaluation of components, systems and structures: non-destructive testing, personnel certification, standards, welding inspection, maintenance and structural testing. The programwill provide an unrivalled opportunity to network with representatives from globally significant research centers, industry leading companies, government organizations, professional bodies and Universities.

Author: Yaguo Lei, Feng Jia, Naipeng Li and Jing Lin

Abstract-Intelligent fault diagnosis of machinery has attracted lots of attention due to its ability in effectively analyzing massive collected signals and providing accurate diagnosis results. In intelligent diagnosis methods, however, the fault features are manually designed depending on prior knowledge about the characteristics of machinery signals and diagnostic expertise, which is time-consuming and labour-consuming. It is interesting to use unsupervised feature learning techniques to directly and adaptively learn fault features from the vibration signals and reduce the need for prior knowledge. Therefore, a novel method is proposed for intelligent fault diagnosis of machinery in this study. In the method, sparse filtering, an unsupervised feature learning technique, is applied to learn representations from the vibration signals of machinery and extract robust features with little prior knowledge. Then softmax regression is employed to identify the health conditions based on these features. The proposed method is validated by a planetary gearbox dataset which involves seven health conditions. The results show that the proposed method obtains fairly high diagnosis accuracies compared with the method using the features designed for gearboxes like energy ratio, sideband index, sideband level factor, etc.

欢迎有兴趣的同学届时参加。