基于ICEEMDAN-VMD与结合注意力机制的简单循环单元的水轮机故障诊断  

Fault Diagnosis of Water Turbine Based on ICEEMDAN-VMD and Combined Attention Mechanism with Simple Recurrent Unit

作　　者：王淑青[1] 盛世龙 王云鹤 翟宇胜 陈开元 WANG Shu-qin;SHENG Shi-long;WANG Yun-he;ZHAI Yu-sheng;CHEN Kai-yuan(School of Electrical and Engineering,Hubei University of Technology,Wuhan 430068,Hubei Province,China;School of Power and Mechanical Engineering,Wuhan University,Wuhan 430072,Hubei Province,China)

机构地区：[1]湖北工业大学电气与电子工程学院,湖北武汉430068 [2]武汉大学动力与机械学院,湖北武汉430072

出　　处：《中国农村水利水电》2024年第11期166-173,178,共9页China Rural Water and Hydropower

摘　　要：基于水电机组振动信号存在的非平稳和非线性,提出一种结合二次分解和简单循环单元(SRU)与注意力结合(Attention)的方法。首先采用ICEEMDAN和VMD方法对振动信号进行二次分解,将信号分解成本征模态函数(IMF);随后,引入基于SRU的深度学习模型,将提取得到的IMF信号输入到SRU中进行学习得到水电机组故障识别器。为了进一步提升模型的性能,在SRU中加入了注意力机制,使模型能够动态地关注不同IMF的重要信息。结合故障识别器和实时振动信号特征,即可识别水电机组运行状态为:正常、预警或故障类型;最后,为验证该方法的有效性,通过对比试验,结合实际电站机组样本数据,验证了所提方法在挖掘信号特征方面的可行性和优越性。Based on the non-stationary and non-linear characteristics of the vibration signals of hydroelectric units,we propose a method that combines second decomposition,Simple Recurrent Units(SRU),and Attention Mechanism.Firstly,we use ICEEMDAN and VMD methods to perform second decomposition on the vibration signals,decomposing the signals into Intrinsic Mode Functions(IMFs).Subsequently,we introduce a deep learning model based on SRU,and the extracted IMF signals are input to the SRU for learning to obtain a fault recognizer for hydroelectric units.To further enhance the model′s performance,the Attention Mechanism is incorporated into the SRU to allow the model to dynamically focus on important information from different IMFs.By combining the fault recognizer with real-time vibration signal features,the operational state of hydroelectric units can be identified as normal,warning,or a specific fault type.Finally,to validate the effectiveness of this method,comparative experiments are conducted using actual data samples from power plant units,confirming the feasibility and superiority of the proposed method in extracting signal features.

关 键 词：水电机组 VMD SRU 注意力机制 ICEEMDAN 故障诊断 

分 类 号：TV734.1[水利工程—水利水电工程] TK05[动力工程及工程热物理]