节水灌溉Issue(1)：1-8,8.DOI:10.12396/jsgg.2025221
基于改进VMD-MSE-SVD的超声多普勒测流信号降噪方法研究
Study on Noise Reduction Method of Ultrasonic Doppler Flow Measurement Signal Based on Improved VMD-MSE-SVD
摘要
Abstract
The application environment for ultrasonic Doppler flow meters is quite complex.To address issues such as low and medium flow rates being affected by noise,leading to low measurement accuracy and significant errors,an innovative noise reduction model has been proposed.This model combines Variational Mode Decomposition(VMD)with Singular Value Decomposition(SVD),based on an improved dung beetle optimizer(ISEDBO),to significantly enhance the signal-to-noise ratio of the echo signals.The method first optimizes the Dung Beetle Optimizer(DBO)using crossover strategies,suboptimal guidance control strategies,and theft behavior enhancement strategies.By comparing different test functions and other algorithms,the superiority of the ISEDBO algorithm is demonstrated.Secondly,the ISEDBO algorithm optimizes the VMD parameter combination,integrating Multi-scale Sample Entropy(MSE)and spectral coefficients to distinguish Intrinsic Mode Functions(IMF).Finally,SVD is used to perform dimensionality reduction and reconstruction on the effective IMF components,further overcoming the secondary harmonic oscillation phenomenon in the mid-low frequency range.Through the processing and analysis of simulation signals and laboratory towing experiments,the feasibility of the method is verified from multiple perspectives.Additionally,the ISEDBO-VMD-SVD noise reduction effect is compared with methods such as particle swarm optimization(PSO),grey wolf optimization(GWO),and dung beetle optimizer(DBO).The results show that,compared to simulated signals,ISEDBO-VMD can effectively suppress noise interference and significantly preserve the original signal characteristics.Compared to PSO-VMD,GWO-VMD,and DBO-VMD,it achieves a signal-to-noise ratio of up to 18.78 dB and a waveform correlation coefficient of up to 0.987.In the comparative towing experiment,statistical analysis of the MSE values for multiple signal groups effectively distinguishes the original signal from background noise.When comparing detection errors at different flow rates,ISEDBO-VMD-SVD has the smallest error,ranging from 0.009～0.02 m/s,which provides a solid foundation for practical water monitoring applications.关键词
超声多普勒/蜣螂优化算法/变分模态分解/样本熵/奇异值分解/测流信号/降噪Key words
ultrasonic doppler/dung beetle optimizer/variational mode decomposition/sample entropy/singular value decomposition/flow measurement signal/noise reduction分类
农业科技引用本文复制引用
沈宏学,商信华,牛亚坤..基于改进VMD-MSE-SVD的超声多普勒测流信号降噪方法研究[J].节水灌溉,2026,(1):1-8,8.基金项目
国家自然科学基金委员会青年基金项目"面向JPEG图像的篡改取证关键技术研究"(62202141). （62202141）
