改进ESPRIT算法在转子动不平衡信号提取中的应用  

Application of Improved ESPRIT Algorithm on Extraction of Rotor Dynamic Unbalance Signals

作　　者：张淑彦 蔡萍[1] 朱明 ZHANG Shuyan;CAI Ping;ZHU Ming(School of Electronic Information and Electrical Engineering,Shanghai Jiao Tong University;Shanghai United Imaging Healthcare CO.,LTD.)

机构地区：[1]上海交通大学电子信息与电气工程学院 [2]上海联影医疗科技股份有限公司

出　　处：《仪表技术与传感器》2025年第1期92-97,共6页Instrument Technique and Sensor

基　　金：国家重点研发计划项目(2021YFB3203104)。

摘　　要：提取转子动不平衡信号参量时,近频干扰严重影响互相关等传统周期信号处理方法的性能。针对近频干扰影响转子动不平衡信号准确提取的问题,提出一种基于改进的ESPRIT算法的信号参数提取方法,对传统的奇异值定阶法进行改进,实现了对信号模态数的准确识别,继而利用改进的ESPRIT方法识别不平衡信号的幅值相位。实验结果表明:相较于传统互相关法和FFT方法,该方法有效解决了短数据、大近频干扰、非整周期条件下不平衡振动信号参数的准确提取问题,在数据长度大于半拍时,结果即趋于稳定,将幅值测量误差降至3%以下,相位测量误差降至3°以下,提高了检测效率和准确度。When the parameters of rotor dynamic unbalance signal are extracted,near-frequency interference seriously affects the performance of traditional periodic signal processing methods such as cross-correlation.Aiming at the problem that near-frequency interference affects the accurate extraction of rotor dynamic unbalance signal,a signal parameter extraction method based on the improved ESPRIT algorithm was proposed to improve the traditional singular value fixed-order method so that it can accurately identify the signal modes,and the magnitude and phase of the unbalanced signal were identified using the improved ESPRIT method.The experimental results show that,compared with the traditional mutual correlation method and FFT method,this method effectively solves the problem of accurate extraction of the parameters of unbalanced vibration signals under the conditions of short data,large near-frequency interference,and non-integer period,and the result tends to be stabilized when the data length is greater than half a beat,and the amplitude measurement error is reduced to less than 3%,and the phase measurement error is reduced to less than 3°,which improves the detection efficiency and accuracy.

关 键 词：动不平衡 近频干扰抑制 ESPRIT 奇异值 定阶方法 信号参数提取 

分 类 号：TH825[机械工程—仪器科学与技术]