摘要
在实际大型动态测量船动态变形测量应用中准确确定捷联惯性姿态测量系统噪声模型和相关性比较困难,采用常规卡尔曼滤波(KF)会导致较大的状态估计误差,甚至使滤波器发散。针对这一问题,在分析扩展遗忘因子递推最小二乘(EFRLS)算法的稳定性基础上将EFRLS估计误差与带系统相关噪声的卡尔曼滤波器估计误差进行了比较,并针对大型测量船动态环境采用EFRLS算法对变形参数进行了估计,仿真结果验证了EFRLS算法在噪声信息未知情况下的有效性。
For the actual application for large survey ship dynamic deformation measurement,it is difficult to establish noise model and to define correlation properties.There exists a large state estimation error using conventional Kalman filter,and it can even make the filter diverging.Aiming at the problem,based on analysis of the stability character of a extended forgetting factor recursive least squares（EFRLS） estimator,estimation errors between EFRLS and Kalman estimators with correlation system noise are compared,and combing dynamic environment of large survey ship deformation parameters are estimated by EFRLS algorithm.Simulation results show the validity of EFRLS under unknown prior noises.
出处
《船舶工程》
CSCD
北大核心
2012年第2期46-49,共4页
Ship Engineering
基金
国家自然科学基金项目(50975049)
国家重点基础发展研究计划(973计划)资助项目(2009CB724002)
关键词
动态变形
惯性测量
EFRLS算法
滤波估计
dynamic deformation
inertial measurement
extended forgetting factor recursive least squares（EFRLS） algorithm
filtering estimation
