基于改进KDE算法的安全阀泄漏源定位  

Leakage source locating of safety valve based on improved KDE algorithm

作　　者：李树勋[1,2] 侯建军 康云星 王志辉 LI Shuxun;HOU Jianjun;KANG Yunxing;WANG Zhihui(School of Petrochemical Technology,Lanzhou University of Technology,Lanzhou 730050,China;Machinery Industry Pump and Special Valve Engineering Research Center,Lanzhou University of Technology,Lanzhou 730050,China)

机构地区：[1]兰州理工大学石油化工学院,兰州730050 [2]兰州理工大学机械工业泵及特殊阀门工程研究中心,兰州730050

出　　处：《振动与冲击》2021年第11期277-284,共8页Journal of Vibration and Shock

基　　金：国家自然科学基金(51569012)。

摘　　要：针对难以对安全阀微小泄漏孔进行准确定位的问题,提出了一种基于改进近似核密度估计(kernel density estimation,KDE)算法的泄漏源定位方法。该算法引入信噪比(signal-to-noise ratio,SNR)追踪和相关性检测(coherence test,CT)相结合的方法提取受环境噪声影响较小且泄漏声源能量占优的时频支撑域,使得KDE算法应用于安全阀泄漏源定位的前提成立。在此基础上,通过分频融合处理抑制高频混叠,计算近似核密度函数,消减近似核密度函数三维谱图基底。通过MATLAB软件仿真计算,根据改进KDE谱图峰值得到安全阀泄漏源定位坐标。结果表明,工作压力分别为0.72 MPa,0.75 MPa和0.77 MPa的安全阀泄漏定位结果相对误差都在10%以内,通过声发射测试和改进KDE算法相结合的方法可以较为准确地实现安全阀泄漏源定位,且随着安全阀工作压力的增大,定位精度也相应提高。Aiming at the problem of it being difficult to accurately locate tiny leakage holes of safety valve,a leakage source locating method based on improved approximate kernel density estimation(KDE)algorithm was proposed.With this algorithm,firstly,the combination of signal-to-noise ratio(SNR)tracking and coherence test(CT)was introduced to extract the time-frequency support domain less affected by environmental noise and with dominant leakage sound source energy so that the KDE algorithm could be applied in the safety valve leakage source locating.Then,the high frequency aliasing was suppressed by using the frequency division fusion processing,the approximate kernel density function was calculated,and 3-D spectrogram bases of the approximate kernel density function were reduced.Through simulation computing with the software MATLAB,according to the peak value of the improved KDE spectrogram,locating coordinates of safety valve’s leakage source were obtained.The results showed that relative errors of safety valve leakage locating results under working pressures of 0.72 MPa,0.75 MPa and 0.77 MPa,respectively are all less than 10%;the method of combining acoustic emission testing and improved KDE algorithm can more accurately locate leakage source of safety valve;with increase in working pressure of safety valve,the locating accuracy increases accordingly.

关 键 词：安全阀 泄漏源定位 信噪比(SNR)追踪 相关性检测(CT) 分频处理 

分 类 号：TH134[机械工程—机械制造及自动化] TH89