盲源分离理论在振动筛轴承故障诊断中的应用

发布时间：2018-12-18 11:32

【摘要】：随着我国经济建设和科学研究事业的进一步发展，筛分机械设备所涉及的领域与应用变得越来越广泛，对于有原材料生产以及应用的领域，都可以看到筛分机械设备，而在这些筛分机械设备中，最常见和常用的设备就是振动筛。在煤炭工业部门、水利水电部门、交通工业部门、化工部门甚至在环卫部门都已经应用到了振动筛。可以看出振动筛在各个行业部门起着至关重要的作用。而振动筛的轴承部分对于振动筛的正常工作有着重要的作用，其工况不仅影响该机器设备本身的安全稳定运行，而且还会对后续生产造成直接影响，，故障严重时会造成重大经济损失，甚至造成机毁人亡的事故，因此对轴承进行故障检验技术与分析技术显得更加迫切。故障诊断技术是一门新发展的科学领域，还没有形成较为完整的科学体系。因此对研究的目的、内容范畴的理解，往往与工程应用背景，乃至工程技术人员的专业不同而有很大的差异，所以对现有的故障理论方法还有一些不足之处与难题，而最关键也是最困难的问题之一就是故障特征信号的特征提取。可以这么说，特征提取是当前故障诊断方面中的一个瓶颈问题，它对于故障诊断的准确性和早期预报的可靠性有着很大的关系。而盲源分离理论为振动信号的处理、故障诊断的识别提供了积极地方法。但是正如其他算法一样，它也有自身的限制，其一就是观测数必须大于振动源数，如果不能满足这一前提条件，那么分离最终会造成失败。针对这一限制，本文提出了基于集合平均经验模态分解的盲源分离算法(EEMD-BSS)，该算法能很好的克制这一限制，使得在观测数小于振动源数的情况下也能较好的分离出故障数据，从而达到分离的目的。最后本文分别使用传统的盲源分离算法和改进的EEMD-BSS算法对轴承的内外圈实验故障数据进行了多通道与单通道的故障特征的分离，都较好的完成了分离任务，说明算法的有效性。
[Abstract]:With the further development of economic construction and scientific research in our country, the fields and applications of screening machinery and equipment have become more and more extensive. For the fields where raw materials are produced and applied, we can see screening machinery and equipment. In these screening mechanical equipment, the most common and commonly used equipment is vibrating screen. Shakers have been used in coal, hydropower, transportation, chemicals and even sanitation. It can be seen that the vibrating screen plays a vital role in all sectors of the industry. The bearing part of the vibrating screen plays an important role in the normal operation of the vibrating screen. Its working conditions not only affect the safe and stable operation of the machine itself, but also have a direct impact on the subsequent production. When the fault is serious, it will cause great economic loss, even cause the accident of machine destruction and death, so it is more urgent to carry on the fault inspection and analysis technology to the bearing. Fault diagnosis technology is a newly developed field of science and has not yet formed a relatively complete scientific system. Therefore, the understanding of the purpose and content category of the research is often different from the engineering application background and even the engineering technicians' specialty, so there are still some deficiencies and difficulties in the existing fault theory and methods. One of the most critical and difficult problems is feature extraction of fault feature signals. It can be said that feature extraction is a bottleneck problem in fault diagnosis at present. It has a great relationship with the accuracy of fault diagnosis and the reliability of early prediction. Blind source separation theory provides an active method for vibration signal processing and fault diagnosis. However, like other algorithms, it has its own limitations. One is that the number of observations must be greater than the number of vibration sources. If this precondition is not satisfied, separation will eventually lead to failure. In order to overcome this limitation, a blind source separation algorithm (EEMD-BSS) based on set average empirical mode decomposition (EMD) is proposed in this paper. The fault data can be separated better when the number of observations is less than the number of vibration sources, so as to achieve the purpose of separation. Finally, the traditional blind source separation algorithm and the improved EEMD-BSS algorithm are used to separate the multi-channel and single-channel fault data of the bearing's inner and outer ring experiment respectively. The effectiveness of the algorithm is illustrated.
【学位授予单位】：西安建筑科技大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TH165.3

【参考文献】