Markov随机跳变系统的传感器系统误差估计

 针对Markov随机跳变系统的系统误差估计问题,提出一种基于马尔可夫链蒙特卡罗(MCMC)和最大似然估计相结合的在线系统误差估计方法。利用最大似然估计给出系统误差等效后验概率分布函数,采用Metropolis-Hastings抽样方法从该概率分布函数中进行抽样;利用系统误差估计和状态估计互为因果的关系,采用期望极大化(EM)方法迭代估计出最优的系统误差;分别对时变和时不变系统误差场景进行仿真分析,结果表明,在考虑系统误差统计特性的同时,所提方法对解决目标运动模型难以建立情况下的系统误差估计问题具有可行性和有效性。

Estimation Method for Sensor System Error Based on Markov Stochastic Jump System

In order to resolve the problem of system error in a Markov stochastic jump system,this paper proposes a novel on-line system error estimation method based on Markov chain Monte Carlo(MCMC) and maximum likelihood.It uses a Metropolis-Hastings sampler to sample from an equitable probability density distributing function which is based on the maximum likelihood estimation.Besides,it can iteratively estimate system error by using expectation maximization(EM) based on the causation of system error estimation and state estimation.The paper simulates two scenes which include time-varying and time-invariant system errors,and the simulations show that this method can take into consideration the system error statistical characteristics,and is feasible and effective in estimating system errors to solve the case of the unknown target state model.