基于最大平均峭度盲解卷积的直升机故障诊断  

Helicopter Fault Diagnosis by Blind Deconvolution Based on Average Kurtosis Maximization

作　　者：张新[1,2] 赵艺珂[2] 王家序 王景霖[3] ZHANG Xin;ZHAO Yike;WANG Jiaxu;WANG Jinglin(State Key Laboratory of Mechanical Transmission for Advanced Equipment,Chongqing UniversityChongqing,400044,China;School of Mechanical Engineering,Southwest Jiaotong University Chengdu,610031,China;Aviation Key Laboratory of Science and Technology on Fault Diagnosis and Health Management Shanghai,201601,China)

机构地区：[1]重庆大学高端装备机械传动全国重点实验室,重庆400044 [2]西南交通大学机械工程学院,成都610031 [3]故障诊断与健康管理技术航空科技重点实验室,上海201601

出　　处：《振动．测试与诊断》2024年第3期480-485,617,共7页Journal of Vibration,Measurement & Diagnosis

基　　金：国家自然科学基金面上资助项目(52175122);机械传动国家重点实验室开放课题资助项目(SKLMT‑MSKFKT‑202108);四川省自然科学基金面上资助项目(2023NSFSC0362);中国博士后科学基金资助项目(2023M732917);四川省博士后创新人才资助项目(BX202214);中央高校基本科研业务费科技创新资助项目(2682021CX021)。

摘　　要：针对最小熵解卷积(minimum entropy deconvolution,简称MED)应用于故障诊断时倾向于恢复少量主导冲击而非周期性故障冲击的问题,定义一种滤波器系数求解指标——平均峭度,提出了最大平均峭度盲解卷积方法。首先,通过对故障信号进行均等分割,取各分割段信号峭度的均值,得到信号的平均峭度;其次,将平均峭度作为信号盲解卷积指标,求解滤波器系数;最后,完成信号滤波,提取周期性故障冲击。仿真信号与直升机故障诊断案例分析结果表明:所提最大平均峭度盲解卷积方法能从含复杂干扰成分的故障信号中恢复故障冲击序列,为故障诊断提供可靠信息;相比于MED等传统盲解卷积方法,所提方法具有较强的普适性。To solve the problem that the minimum entropy deconvolution(MED)tends to recover a few dominant impacts rather than the periodic fault impact sequence,a blind deconvolution(BD)based on average kurtosis maximization is proposed.This method includes a new BD indicator:average kurtosis.The signal is initially split into several equal segments,and then the average kurtosis is defined as the mean value of the kurtosis of each segment.Subsequently,the BD filter coefficient is solved by iteratively maximizing the average kurtosis of the filtered signal.Finally,the fault impact sequence is extracted by filtering the signal via the obtained coefficient.The results of the simulated signal analysis and helicopter fault diagnosis show that the proposed BD method can accurately extracting periodic fault impact sequences from a signal containing complex interferences.The comparisons with the most popular BD methods in this field further highlight the advantages of the proposed method for gear fault diagnosis.

关 键 词：直升机 齿轮箱 故障诊断 盲解卷积 平均峭度 

分 类 号：V232.8[航空宇航科学与技术—航空宇航推进理论与工程] TH132.41[机械工程—机械制造及自动化]