摘要
超声多普勒流量计的应用环境复杂多变,因此提高测量精度和误差,精确提取回波信号非常重要。提出了融合变分模态分解(Variational Mode Decomposition,简称VMD)和奇异谱分析(Singular Spectrum Analysis,简称SSA)的降噪模型,以更好地改善回波信号信噪比。该方法首先利用柯西变异算子产生随机迭代过程,克服了海鸥算法(Seagull Optimization Algorithm,简称SOA)容易陷入局部最优的特性;其次,采用包络谱熵值作为适应度函数,自适应优化VMD参数组合,同时引入云相似度值作为有效模态分量(Intrinsic Mode Functions,简称IMF)筛选的标准;最后,针对中低频的二次谐波振荡现象,引入SSA加以解决。通过构造超声波模拟信号和走车实验数据实例,与小波阈值、经验模态分解(Empirical Mod Decomposition,简称EMD)等方法对比,分析CVSOA-VMD-SSA降噪效果。结果表明:对于仿真信号而言,CVSOA-VMD-SSA能克服模态混叠及SOA易陷入局部最优解问题,更有效地抑制噪声干扰,相较于EMD-SSA、SOA-VMD-SSA,信噪比最高达30.78 dB,均方根误差最低达0.01;对于走车实验而言,采用多组信号统计分析,确定云相似度阈值为0.6,对比不同走车流速探测精度,CVSOA-VMD-SSA误差最小,范围在0.01~0.03 m/s,该结果为实际工程应用提供理论支撑。
The application environment of the ultrasonic Doppler flowmeter is complex and variable,and the accurate extraction of the echo signal is crucial to improve its measurement accuracy and error.In order to better filter out the noise in the echo signal,a noise reduction model integrating variational mode decomposition(VMD)and singular spectrum analysis(SSA)is innovatively proposed.Firstly,this method uses the Cauchy variation operator(CV)to generate a random iteration process to overcome the seagull optimization algorithm(SOA)easy to fall into the local optimal solution problem;secondly,adopts the envelope spectrum entropy value as the fitness function,optimizes the VMD parameter combination to overcome the subjective problem,and introduces cloud similarity value as the criterion of the intrinsic mode functions(IMF)screening;Finally,for the medium and low-frequency oscillation phenomenon after transformer mode decomposition,SSA introduces secondary filtering to further improve the signal-to-noise ratio.The noise reduction effect of CVSOA-VMD-SSA is analyzed by analyzing ultrasonic simulation signals and experimental data examples and comparing them with wavelet threshold(WD)and empirical Mod decomposition(EMD).The results show that,for the simulation signals,The CVSOA-VMD-SSA can overcome the modal stacking and the SOA can easily fall into the local optimal solution problem,more effectively suppressing the noise interference as compared to the EMD-SSA,SOA-VMD-SSA,The signal to noise ratio is up to 30.78 dB,and the lowest RMS error is 0.01.For the automatic car-walking experiment,using a multiple-group signal statistical analysis,To determine a cloud similarity threshold of 0.6,And comparing the detection accuracy of different vehicle flow rates,CVSOA-VMD-SSA with minimal error,Range is in the 0.01~0.03 m/s,Provide theoretical support for practical engineering applications.
作者
赵军华
戴聪聪
李丛
冯阳
邓权
张清波
ZHAO Jun-hua;DAI Cong-cong;LI Cong;FENG Yang;DENG Quan;ZHANG Qing-bo(Shenzhen Hongdian Technology Co.,Ltd.,Instrument Development Center,Shenzhen 518000,Guangdong)
出处
《节水灌溉》
北大核心
2023年第9期16-24,共9页
Water Saving Irrigation
基金
城区内涝风险监测预警系统关键技术研发(KCXFZ202002011007040)
黄河南岸(内蒙段)基于生态安全地下水承载能力提升灌溉模式和管理机制研究(NSK2021-Z1)
5G工业互联网核心产品技术攻关及产业化项目(XMHT20210101008)。
关键词
超声多普勒
柯西变异
海鸥算法
变分模态分解
奇异谱分析
ultrasonic doppler
cCauchy variation
seagull algorithm
variational mode decomposition
singular spectrum analysis
