复数小波包能量矩熵的深度GRU轴承故障预测  

Deep GRU Bearing Fault Prediction Based on Complex Wavelet Packet Energy Moment Entropy

作　　者：葛本利 GE Benli(Huainan School of Economics and Technology,Anhui Huainan 232000,China)

机构地区：[1]淮南经济技术学校,安徽淮南232000

出　　处：《机械设计与制造》2024年第12期21-28,共8页Machinery Design & Manufacture

基　　金：安徽省教育信息技术研究2018年度课题(AH2018093)。

摘　　要：为了准确及时的表征轴承的性能退化,实现有效的早期故障预测,提出了一种基于复数小波包能量矩熵的深度门控回归单元轴承早期故障预测方法。首先,将复数小波包能量矩熵定义为一种新的监测指标,用来表征其承载性能的退化。其次,构造了深门控回归单元网络,捕获了所定义的监测指标中隐藏的线性映射关系。最后,提出了一种基于学习速率衰减策略的改进训练算法,以提高所构建的深度模型的预测能力。通过两个实验结果分析可知提出方法具有更高的灵敏度和准确度。In order to accurately and timely characterize the performance degradation of bearing and achieve effective early fault prediction,a depth gated regression unit bearing early fault prediction method based on complex wavelet packet energy moment entropy was put forward.Firstly,the complex wavelet packet energy moment entropy was defined as a new monitoring index to characterize the degradation of its bearing capacity.Secondly,a deep gated regression cell network was constructed to capture the hidden linear mapping relationship in the defined monitoring index.Finally,an improved training algorithm based on learning rate decay strategy was proposed to improve the prediction ability of the depth model.The analysis of two experimental results shows that the proposed method has higher sensitivity and accuracy.

关 键 词：复数小波 能量矩熵 深度门控回归单元 故障预测 

分 类 号：TH16[机械工程—机械制造及自动化] TH133.33