系统工程与电子技术
繫統工程與電子技術
계통공정여전자기술
SYSTEMS ENGINEERING AND ELECTRONICS
2014年
4期
643-649
,共7页
强跟踪滤波%状态和参数联合估计%平方根容积卡尔曼滤波%故障参数
彊跟蹤濾波%狀態和參數聯閤估計%平方根容積卡爾曼濾波%故障參數
강근종려파%상태화삼수연합고계%평방근용적잡이만려파%고장삼수
strong tracking filter%state and parameter joint estimation%square-root cubature Kalman filter (SCKF)%fault parameter
针对非线性系统中不可观测故障参数估计问题，提出基于多重渐消因子强跟踪平方根容积卡尔曼滤波（multiple fading factors strong tracking square-root cubature Kalman filter，MSTSCKF）的状态和参数联合滤波算法。MSTSCKF 基于强跟踪滤波器理论框架，通过引入多重渐消因子实时调整增益矩阵，克服平方根容积卡尔曼滤波（square-root cubature Kalman filter，SCKF）在故障参数变化函数未知或者突变时滤波精度下降甚至发散的缺点，并兼具 SCKF 在非线性拟合精度和数值稳定性等方面的优点。仿真结果表明，相比 SCKF 和强跟踪无迹卡尔曼滤波（unscented Kalman filter，UKF），本文提出的方法具有更高的估计精度。
針對非線性繫統中不可觀測故障參數估計問題，提齣基于多重漸消因子彊跟蹤平方根容積卡爾曼濾波（multiple fading factors strong tracking square-root cubature Kalman filter，MSTSCKF）的狀態和參數聯閤濾波算法。MSTSCKF 基于彊跟蹤濾波器理論框架，通過引入多重漸消因子實時調整增益矩陣，剋服平方根容積卡爾曼濾波（square-root cubature Kalman filter，SCKF）在故障參數變化函數未知或者突變時濾波精度下降甚至髮散的缺點，併兼具 SCKF 在非線性擬閤精度和數值穩定性等方麵的優點。倣真結果錶明，相比 SCKF 和彊跟蹤無跡卡爾曼濾波（unscented Kalman filter，UKF），本文提齣的方法具有更高的估計精度。
침대비선성계통중불가관측고장삼수고계문제，제출기우다중점소인자강근종평방근용적잡이만려파（multiple fading factors strong tracking square-root cubature Kalman filter，MSTSCKF）적상태화삼수연합려파산법。MSTSCKF 기우강근종려파기이론광가，통과인입다중점소인자실시조정증익구진，극복평방근용적잡이만려파（square-root cubature Kalman filter，SCKF）재고장삼수변화함수미지혹자돌변시려파정도하강심지발산적결점，병겸구 SCKF 재비선성의합정도화수치은정성등방면적우점。방진결과표명，상비 SCKF 화강근종무적잡이만려파（unscented Kalman filter，UKF），본문제출적방법구유경고적고계정도。
For unmeasured fault parameter estimation of nonlinear system,a state and parameter joint esti-mation algorithm based on multiple fading factors strong tracking square-root cubature Kalman filter (MSTSCKF)is presented.Under the basic theory framework of strong tracking filter,MSTSCKF introduces the multiple fading factors to adjust gain matrix in real time and avoids the problem that square-root cubature Kalman filter (SCKF)decreases in accuracy and even diverges when the changing function of fault parameters is unknown or fault parameters abruptly change.Meanwhile,MSTSCKF combines high nonlinear curve fitting and numerical stability of SCKF.The simulation results indicate that higher estimation accuracy is obtained compared with SCKF and strong tracking unscented Kalman filter (UKF).
