华中科技大学学报（自然科学版）Issue(10)：31-36,6.DOI:10.13245/j.hust.141007
EM-CDKF算法及其SINS初始对准应用
EM-CDKF algorithm and its applications on SINS′initial alignment
摘要
Abstract
As the unknow n priori statistical properties of the nonlinear strap-dow n inertial navigation system (SINS) noises ,based on the central divided Kalman filtering (CDKF) algorithm ,the expecta-tion maximum steepest descent method was presented to develop the adaptive expectation maximum based central divided Kalman filtering (EM-CDKF) algorithm with maximum likelihood criterion ,to evaluate on-line the system noise’s statistical properties .The EM-CDKF algorithm constructs the Log-likelihood function of system noises statistical properties with maximum likelihood criterion ,and transforms the estimation evaluation of system noise statistical properties into the maximum evalua-tion of Log-likelihood function with the expectation maximum steepest descent method ,and with the EM-CDKF algorithm ,the process and measurement noises can be evaluated by online recursive pat-tern .T he simulink experiments of SINS’ large azimuth misalignment angle indicate that ,compared to CDKF algorithm ,the EM-CDKF algorithm can effectively overcome the evaluation precision declining and so that algorithm divergence without the priori statistical properties of the nonlinear SINS noises , and at the same time can online evaluate the system noises statistical properties with recursive pattern .关键词
捷联惯导系统/初始对准/中心差分卡尔曼滤波/极大似然准则/极大期望算法Key words
strap-dow n inertial navigation system/initial alignment/central divided Kalman filtering/maximum likelihood principle/expectation maximization algorithm分类
信息技术与安全科学引用本文复制引用
丁国强,徐洁,周卫东,张志艳..EM-CDKF算法及其SINS初始对准应用[J].华中科技大学学报（自然科学版）,2014,(10):31-36,6.基金项目
国家自然科学基金资助项目（U1204603）；郑州轻工业学院博士基金资助项目（2011BSJJ00048）． （）
