基于自动特征提取的旋转机械故障诊断方法  

Fault Diagnosis Method of Rotating Machinery Based on Automatic Feature Extraction

作　　者：陈丹峰 冯坤 刘万宇 唐嘉 郭思辰 CHEN Danfeng;FENG Kun;LIU Wanyu;TANG Jia;GUO Sichen(Technical Information Center of Henan Electric Power Co.,Ltd.,SPIC,Zhengzhou Henan 450001,China;Huazhong University of Science and Technology,Wuhan Hubei 430074,China)

机构地区：[1]国家电投集团河南电力有限公司技术信息中心,河南郑州450001 [2]华中科技大学,湖北武汉430074

出　　处：《湖北电力》2024年第5期30-37,共8页Hubei Electric Power

基　　金：国家重点研发计划(项目编号:2022YFB4100700)。

摘　　要：传统的旋转机械专家诊断系统通常依赖人工选择故障征兆,容易受到经验和主观因素影响,而数据驱动方法面临故障数据不完全或稀缺的问题,尤其是在面对数据不足或不平衡的情况下,结合先验知识能够实现更加精准和可靠的故障诊断。为了解决这些问题,提出了一种基于自动特征提取的旋转机械故障诊断方法,旨在提高汽轮机等旋转设备的故障诊断自动化程度和准确性。提出的方法结合了无量纲参数、傅里叶变换、小波变换等,能够高效提取振动信号中的关键特征,进而识别故障类型,估算故障概率,并提供相应的维修建议。燃气轮机实验数据和汽轮机组实际运行数据验证表明,燃气轮机的诊断准确率达到94.49%,汽轮机组运行数据的诊断准确率达到82.23%,说明提出的方法具有较好的准确性和通用性。Traditional expert diagnostic systems for rotating machinery usually rely on manual selection of fault symptoms,which is easily affected by experience and subjective factors,while data-driven method has the problem of incomplete or scarce fault data,especially in the case of insufficient or unbalanced data,thus needing to combine prior knowledge to achieve more accurate and reliable fault diagnosis.In order to solve the aforementioned problems,this paper proposes a fault diagnosis method of rotating machinery based on automatic feature extraction,which aims to improve the automation degree and accuracy of fault diagnosis for rotating equipment.The proposed method,combined with dimensionless parameters,Fourier transform and wavelet transform,can efficiently extract key features of vibration signals,identify fault types,estimate fault probabilities,and even provide corresponding maintenance suggestions.The experimental data of gas turbine and the actual operation data of turbine units show that the diagnostic accuracy of gas turbine reaches 94.49%and that of steam turbine operation data reaches 82.23%,indicating that the proposed method has good accuracy and universality.

关 键 词：旋转机械 自动特征提取 汽轮机 燃气轮机 

分 类 号：TH17[机械工程—机械制造及自动化]