2024年第3期仪表技术与传感器InstrumentTechniqueandSensor基金项目:广西自然科学基金项目(2018GXNSFAA294121)收稿日期:2023-08-24基于旋转变分模态分解的IMU角速度去噪算法覃舒娴,覃晓兰,刘运毅广西大学计算机与电子信息学院摘要:惯性测量单元(IMU)的应用中,角速度的噪声误差积累对姿态解算性能有较大的影响。针对角速度中存在的噪声,提出一种融合了变分模态分解(VMD)和角速度旋转三维分解的去噪算法。首先通过坐标系旋转获得角速度在不同虚拟轴的输出,再利用VMD提取合适分量重构虚拟轴信号。VMD的非线性重构使得各个虚拟轴的残留误差相对独立,最终多个虚拟轴的反向旋转回到原始坐标系后通过独立信号的均值合并能有效消除IMU中角速度的噪声。基于EuRoC数据集的实验结果表明:该算法降噪效果显著,均方根误差降低70%~85%,且能有效平衡三轴误差。关键词:变分模态分解;旋转重构;角速度去噪中图分类号:TP212文献标识码:AIMUAngularVelocityDenoisingAlgorithmBasedonRotationalVariationalModeDecompositionQINShuxian,QINXiaolan,LIUYunyiSchoolofComputer,ElectronicsandInformation,GuangxiUniversityAbstract:Intheapplicationofinertialmeasurementunit(IMU),theaccumulationofangularvelocitynoiseerrorsgreatlyaf-fectstheattitudecalculationperformance.Aimingatthenoiseexistingintheangularvelocity,adenoisingalgorithmwaspro-posed,whichcombinedvariationalmodedecomposition(VMD)andangularvelocityrotationalthree-dimensionaldecomposition.Firstly,theoutputofangularvelocityindifferentvirtualaxeswasobtainedbyrotationofcoordinatesystem,andthenthevirtualaxissignalswerereconstructedbyextractingappropriatecomponentsusingVMD.ThenonlinearreconstructionofVMDmadetheresidualerrorsofeachvirtualaxisrelativelyindependent.Finally,afterthereverserotationofmultiplevirtualaxesbacktotheoriginalcoordinatesystem,themeanvalueofindependentsignalscaneffectivelyeliminatetheangularvelocitynoiseinIMU.TheexperimentalresultsbasedonEuRoCdatasetshowthatthealgorithmhasaremarkableeffectonnoisereduction,rootmeansquareerrorisreducedby70%~85%,andthetriaxialerrorcanbeeffectivelybalanced.Keywords:variationalmodedecomposition;rotationalreconstruction;angularvelocitydenoising0引言惯性测量单元(inertialme...
