【Applied Energy 最新原创论文】基于压电俘能装置自传感的自供电滚动轴承故障诊断方法研究

•提出了一种基于压电俘能装置自传感的自供电故障诊断方法。

•通过分析压电俘能装置的电学输出信号来实现故障诊断。

•通过自供电滚动轴承故障诊断实验证明方法可行性。

•所提出方法的分类准确率可高达99.057%。

The self-powered fault diagnosis methods provide a new idea to realize fault diagnosis without using wired sensors or battery-supported sensors. However, current self-powered fault diagnosis methods mainly depend on self-powered sensors, most of which may destroy the original structure of mechanical parts or contain connected energy harvesters and sensors, resulting in reduced reliability and stability. In this paper, a self-powered rolling bearing fault diagnosis method with a piezoelectric generator (PEG) for self-sensing is proposed. The PEG can be simply installed without destroying the bearing structure, and its electrical outputs excited by the mechanical vibrations can be further used for fault diagnosis. The energy conversion mechanism of a conventional monostable PEG and the basic knowledge of the convolutional neural network (CNN) model are first introduced, based on which the self-powered fault diagnosis approach is proposed. Experimental work is conducted to identify the parameters of the PEG’s governing equation and verify its effectiveness. The responses of a PEG installed onto a rolling bearing experimental table are further measured and analyzed. Finally, the CNN model is trained based on the electrical outputs of the PEG under different installation positions, different rotation speeds and different healthy statuses of the bearings. The research results show that the average classification accuracies of this method for rotational speeds of 1200 rpm, 3000 rpm and 4800 rpm can achieve 88.389%, 98.99% and 98.431%, respectively, which are quite feasible in practical applications. This work shows a totally new idea of adopting a vibration energy harvester as a self-sensing device to achieve self-powered fault diagnosis, which has wide application potential in the future. 

Keywords

Self-powered rolling bearing fault diagnosis；

Self-sensing；

Piezoelectric generator;

Convolutional neural network；

Fig. 1.    Fig. 1. (a) System structure and (b) the force diagram of the cantilever beam-based monostable PEG.

Fig. 2. (a) the schematic diagram and parameters of the system and (b) the equivalent mass-spring-damping model.

Fig. 4. Flowchart of the self-powered fault diagnosis approach based on the vibration/rotation energy harvester and classification model.

Fig. 5. Experiment setup for measuring the tip displacement and output voltage of the PEG.

Fig. 8. Rolling bearing fault diagnosis experimental table.

Fig. 12. Installation positions of the PEG on the bearing fault diagnosis experimental table.

Fig. 16. Feature visualization via t-SNE: feature representations of images through the last fully-connected layer.

       本研究由深圳大学和宁波大学的研究人员共同完成。通信作者简介：

       赖志慧，博士，深圳大学特聘研究员，深圳市海外高层次人才。现带领团队从事机械动力学领域的研究。主持省部级以上科研项目9项；累计发表学术论文70余篇，其中以第一/通讯作者在Joule、Applied Energy、Mechanical Systems and Signal Processing等期刊发表SCI收录论文40余篇，并多次受邀报告。谷歌学术引用1900余次，H指数24。

第一作者简介：

       石润烨，硕士，深圳大学在读硕士研究生，从事振动能量获取、设备故障诊断领域研究。在Applied Energy、Mechanical Systems and Signal Processing、Journal of Sound and Vibration、Smart Material and Structures等SCI期刊发表多篇研究论文。

关于Applied Energy

点击“阅读原文”

喜欢我们的内容？

点个“赞”或者“再看”支持下吧！