光学精密工程
光學精密工程
광학정밀공정
OPTICS AND PRECISION ENGINEERING
2010年
2期
377-383
,共7页
混合滤波%惯性导航%零速检测%GPS缺失
混閤濾波%慣性導航%零速檢測%GPS缺失
혼합려파%관성도항%령속검측%GPS결실
hybrid unscented particle filter(HUPF)%integrated navigation system (INS)%ZUPT%Non-linear%GPS output
分析了Unscented卡尔曼滤波(UKF)和Unscented粒子滤波(UPK)算法的特点,提出了一种混合粒子滤波算法(HUPF),以改善车载系统在运行时间长、运动状态变化频繁时导航系统不稳定的非线性特点,并使导航系统能在GPS信号缺失的情况下继续稳定工作.首先,通过零速检测(ZUPT)方法确定车载系统不同的动态特性.然后,针对检测结果选用混合滤波算法,即根据检测结果确定在不同的动态特性下采用UKF或者UPF算法对车载系统特性变化进行不同处理.最后,根据实验结果对不同方法进行了比较.结果表明,采用本文提出的滤波方法,车载导航系统在不同的动态特性下,特别是在GPS缺失的情况下能有效地降低载体在不同动态特性下误差的影响,提高车载系统动态定位精度,将由非线性误差导致车载系统误差积累造成的影响减少了55%,处理速度提高大约2倍.
分析瞭Unscented卡爾曼濾波(UKF)和Unscented粒子濾波(UPK)算法的特點,提齣瞭一種混閤粒子濾波算法(HUPF),以改善車載繫統在運行時間長、運動狀態變化頻繁時導航繫統不穩定的非線性特點,併使導航繫統能在GPS信號缺失的情況下繼續穩定工作.首先,通過零速檢測(ZUPT)方法確定車載繫統不同的動態特性.然後,針對檢測結果選用混閤濾波算法,即根據檢測結果確定在不同的動態特性下採用UKF或者UPF算法對車載繫統特性變化進行不同處理.最後,根據實驗結果對不同方法進行瞭比較.結果錶明,採用本文提齣的濾波方法,車載導航繫統在不同的動態特性下,特彆是在GPS缺失的情況下能有效地降低載體在不同動態特性下誤差的影響,提高車載繫統動態定位精度,將由非線性誤差導緻車載繫統誤差積纍造成的影響減少瞭55%,處理速度提高大約2倍.
분석료Unscented잡이만려파(UKF)화Unscented입자려파(UPK)산법적특점,제출료일충혼합입자려파산법(HUPF),이개선차재계통재운행시간장、운동상태변화빈번시도항계통불은정적비선성특점,병사도항계통능재GPS신호결실적정황하계속은정공작.수선,통과령속검측(ZUPT)방법학정차재계통불동적동태특성.연후,침대검측결과선용혼합려파산법,즉근거검측결과학정재불동적동태특성하채용UKF혹자UPF산법대차재계통특성변화진행불동처리.최후,근거실험결과대불동방법진행료비교.결과표명,채용본문제출적려파방법,차재도항계통재불동적동태특성하,특별시재GPS결실적정황하능유효지강저재체재불동동태특성하오차적영향,제고차재계통동태정위정도,장유비선성오차도치차재계통오차적루조성적영향감소료55%,처리속도제고대약2배.
Combined the Unscented Kalman Filtering (UKF)and the Unscented Particle Filtering (UPK), a Hybrid Unscented Particle Filtering(HUPF) was presented to improve the instability of the Micro Integrated Navigation System(MINS) for a vehicle when it was in a long-time moving, changes of dynamic performance and the case of GPS signal loss. Firstly, the Zero velocity Updates (ZUPT) was used to determine the dynamic characteristics of different vehicle systems. Then, according to the test results, the UKF or UPK algorithms were chosen to treat the performance of the vehicle under different dynamic characteristics. Finally, based on the experimental results, different ways were compared. It is indicate that the HUPF algorithm combined with ZUPT has reduced effectively the general error of the MINS of the vehicle under different dynamic characteristics, especially in the case of GPS signal loss. Obtained results show that the positioning accuracy and the error accumulation of the vehicle have improved by 55% and the processing speed has increased by nearly twice as compared with that of general ways.