基于脊线信息增强与特征融合的瞬时转频估计  

Rotational speed estimation based on the ridge enhancement and feature fusion

作　　者：江星星[1] 吴楠 石娟娟[1] 沈长青 李川[2] 朱忠奎[1] JIANG Xingxing;WU Nan;SHI Juanjuan;SHEN Changqing;LI Chuan;ZHU Zhongkui(Soochow University,School of Urban Rail Transportation,Suzhou 215131,China;Chongqing Key Laboratory of Manufacturing Equipment Mechanism Design and Control,Chongqing Technology and Business University,Chongqing 400067,China)

机构地区：[1]苏州大学城市轨道交通学院,江苏苏州215131 [2]重庆工商大学制造装备机构设计与控制重庆市重点实验室,重庆400067

出　　处：《振动与冲击》2018年第20期166-172,共7页Journal of Vibration and Shock

基　　金：国家自然科学基金(51705349; 51605319);中国博士后科学基金(2017M621811);江苏省自然科学青年基金项目(BK20160318)

摘　　要：转速获取是变工况设备健康诊断的前提。在不便安装速度传感器的情况下,基于振动信号时频分析获取转频是最常用的途径。然而,由于时频分析方法的自身特性和采集的振动信号中往往包含大量的背景噪声,导致得到的时频分布能量聚集性差、部分时段转速信息微弱等问题,很难提取到完整、准确的转频信息。为解决这一问题,提出一种脊线信息增强与特征融合的转速估计方法。采用幅值累加平方策略对时频分布特征进行增强;从信号低频区域和共振频带分别预估计出转频信息;最后,建立基于概率分布和局部波动特性的信息融合准则,以确定脊线融合位置以及融合结果,实现转频的准确估计。轴承故障实验信号验证说明:相比于传统的转频提取方法,提出的方法能够显著地改善能量微弱的转速信息的识别结果。One of the key steps for rotating equipment fault diagnosis under variable speed condition is to acquire the accurate shaft instantaneous frequency(IF).The most effective way to realize it is based upon the Time-frequency distribution(TFR)analysis.However,its effectiveness is often undermined by the TFR’s poor energy concentration and weak shaft IF-related signal segmentations caused by strong background noise and interferences.A speed estimation method based on the ridge enhancement and feature fusion was,therefore,proposed in this paper.The TF ridge candidates were enhanced firstly by the amplitude sum-square method,and then the TF ridge candidates in lower band and the resonance band were searched,respectively.The fusion criterion based on probability distribution and local fluctuation was,finally,established to acquire the accurate estimation of shaft IF ridge.Compared with conventional methods,the proposed method yields more accurate IF approximation results and is less sensitive to noise.The experiment results validate the effectiveness of the proposed method for shaft IF estimation.

关 键 词：时频分布 转频估计 特征融合 微弱信息 

分 类 号：TH165.3[机械工程—机械制造及自动化] TN911.7[电子电信—通信与信息系统]