基金项目:国家自然科学基金(11704229);山东省重点研发计划(2019GGX101034)收稿日期:2021-03-12修回日期:2021-05-24第40卷第2期计算机仿真2023年2月文章编号:1006-9348(2023)02-0321-05基于CEEMDAN和小波熵的心音信号去噪算法研究刘倩1*,徐彦2,梁春燕1,袁玉英1(1.山东理工大学计算机科学与技术学院,山东淄博255049;2.山东省淄博市中医医院心内科,山东淄博255300)摘要:针对传统心音去噪算法可能丢失部分重要心音信息问题,提出了一种自适应噪声完备经验模态分解(CEEMDAN)和小波熵结合的心音信号去噪算法。算法通过CEEMDAN将心音信号自适应分解成多个本征模态函数(IMFs),基于各阶本征模态的能量分析判别信噪分界点,对含噪IMF分量采用小波熵自适应阈值去噪后,与信号IMF分量重构,得到去噪后的心音信号。仿真结果显示,在不同信噪比条件下,上述算法均能明显提高心音信号的信噪比,降低均方根误差,优于其它传统去噪算法,具有良好地抑制噪声能力。关键词:自适应噪声完备经验模态分解;小波熵;心音信号;噪声中图分类号:TP391.9文献标识码:BResearchOnHeartSoundSignalDenoisingalgorithmBasedontheCombinationofCEEMDANandWaveletEntropyLIUQian1*,XUYan2,LIANGChun-yan1,YUANYu-ying1(1.SchoolofComputerScienceandTechnology,ShandongUniversityofTechnology,ZiboShandong255049,China;2.DepartmentofCardiovascularMedicine,ZiboHospitalofTraditionalChineseMedicine,ZiboShandong255300,China)ABSTRACT:Thetraditionalheartsounddenoisingalgorithmmaylosepartofimportantinformationwhilefilteringoutthenoise.Aheartsoundsignaldenoisingmethodbasedonthecombinationofcompleteensembleempiricalmodedecompositionwithadaptivenoise(CEEMDAN)andwaveletentropywasproposed.Theheartsoundsignalswaswereadaptivelydecomposedintomultipleintrinsicmodefunctions(IMFs)byCEEMDAN.Basedontheenergy,theboundaryofsignalsandnoiseswasdetermined.ThewaveletentropyadaptivethresholddenoisingwasadoptedforthenoisyIMFs.Finally,thedenoisedheartsoundsignalswiththerestoftheIMFscomponentswaswerereconstructed.Thesimulationresultsshowthatunderdifferentsignal-to-noiseratioconditions,thealgorithminthispapercansig-nificantlyimprovethesignal-to-noiseratioandrootmeansquareerrorofthehearts...