一种多运动模式下自适应阈值零速修正算法

传统基于微机电惯性测量单元（MEMS-IMU）的惯性导航系统（INS）引入零速修正（ZUPT）算法校正器件的累积误差。但由于ZUPT算法零速判定阈值为固定值，只适合单一运动模式，当室内行人运动轨迹包含多种运动模式时，定位精度下降。对此，提出了一种多运动模式下自适应阈值ZUPT算法。分析了室内行人包括静止、走、跑、上楼和下楼5种运动模式零速判定阈值的选取，实现了利用随机森林（RF）算法对5种运动模式的分类识别，并根据识别结果对ZUPT算法零速判定阈值进行自适应调整。为了验证本文算法的可行性和有效性，利用MATLAB软件平台对实测数据进行处理，并与传统定位算法进行了比较。3组实验结果表明，当室内行人运动轨迹包括多种运动模式时，相比传统固定阈值的ZUPT算法，引入自适应调整阈值的ZUPT算法可使定位算法的定位精度提高73.83%。

Adaptive threshold zero-velocity update algorithm under multi-movement patterns

Zero-velocity update (ZUPT) algorithm is imported to calibrate device's cumulative error in traditional inertial navigation system (INS) which is based on micro-electro-mechanical system inertial measurement unit (MEMS-IMU).The positioning accuracy will be reduced when the movement trajectory of indoor pedestrian contains multi-movement patterns,because the zero-velocity determination threshold is fixed and only suitable for a single movement pattern.An adaptive threshold ZUPT algorithm under multi-movement patterns was proposed.The selection of zero-velocity determination threshold of indoor pedestrian's five movement patterns including Still,Walk,Run,Upstairs and Downstairs was analyzed.Classification and recognition of five movement patterns using random forest (RF) algorithm were realized.And the zero-velocity determination threshold of ZUPT was adaptively adjusted according to the recognition results.In order to verify the feasibility and validity of the algorithm,the test data was disposed and was compared with traditional positioning algorithm through MATLAB software platform.The three groups of test results show that,when there are multiple movement patterns in an indoor pedestrian trajectory,the positioning accuracy of positioning algorithm can be improved by 73.83％ when ZUPT algorithm with adaptively adjusted threshold is imported,compared with traditional positioning algorithm with fixed threshold.