粒子滤波算法在多传感器测量中的应用

目标跟踪是粒子滤波算法在处理非线性问题的一种典型应用,但由于在线处理能力或传输条件的限制,实际应用中往往无法对多个传感器数据同时处理。据此,给出了一种基于多传感器选优的粒子滤波算法。假设每个时刻可以处理一个测量数据,该算法先采用加权的概率密度函数来评价每个传感器获得的测量值,并用粒子滤波对概率密度函数的加权进行实时更新,基于最大熵标准来选取最优测量数据进行处理。同时,最大熵标准保证了最优似然函数分布最宽,从而缓解粒子衰竭问题。通过数值仿真实验证明,该算法可以选择最优观测数据进行处理,有效降低多传感器测量中粒子滤波在线实时处理性能的要求,也较好地缓解了粒子滤波的"衰竭"问题。

Application of particle filter algorithm in multisensor measurement

Target tracking is one of the typical applications of particle filter algorithm in deal with the nonlinear problems. But due to the limitations of on-line processing and transmitting, it may be infeasible to process the multiple sensor data at one time. A particle filter algorithm based on multisensor prepotency is presented. Suppose that only one measured data can be process in one time, the algorithm uses weighted probability density function （PDF） to estimate the obtained value in each candidate sensors, and real-time updates the weighting of the probability density function by particle filter. The optimal measured data is selected based on the maximum entropy rule. Meanwhile, the maximum entropy ensured the optimal likelihood function is the most widely, so that the failure problem of particle can be relieved. Numerical simulation experiment shows that the algorithm can process the optimal observation data, and effectively reduce the requirement of particle filter of muhisensor in on-line real- time processing. It also relieved failure problem of particle.