摘要
紧耦合SINS/GPS导航系统的数据融合过程存在高维状态和非线性混合模型,采用常规的线性卡尔曼滤波很难处理;而一般的非线性粒子滤波存在粒子退化和样本枯竭的问题。对此,提出将容积粒子滤波(CPF)算法应用于SINS/GPS组合导航系统。CPF利用容积卡尔曼滤波算法得到粒子滤波(PF)的重要性密度函数,并将最新量测值融入系统状态的转移过程中,由此产生的预测样本接近于系统状态的真实后验概率的样本。仿真结果表明,CPF算法的估计性能明显优于标准PF。
Due to high dimensional state and nonlinear mixed model exist in data fusion of SINS/GPS tightly coupled navigation system, conventional linear Kalman filter is incompetent; while general nonlinear particle fiher （PF） features particle degradation and sample depletion problems, thus the cubature particle filter （CPF） algorithm is applied in SINS/GPS integrated navigation system. With CPF, the importance density function of PF is obtained by adopting Kalman filtering algorithm, the latest measured values are integrated into the transfer process of system state; the predictive sample produced by this method is very close to real posterior probability sample of system state. The result of simulation indicates that the estimation performance of CPF algorithm is obviously better than standard PF.
出处
《自动化仪表》
CAS
北大核心
2014年第1期6-9,共4页
Process Automation Instrumentation
基金
国家自然科学基金资助项目(编号:60974070)
