Abstract: The variable-dimension filter with input estimation estimates the maneuver input with the least-square criterion and the time that the object begin to maneuver respectively. It has overcome the defect of the input estimation and the variable-dimension filter. The multilevel multi-sensor system can give more precise results than the single sensor .We propose a new information fusion way for multi-level multi-sensor system based on variable filter with input estimation.

摘要： 带输入估计变维滤波利用最小二乘法对系统未知输入进行估计，同时对机动运行开始时刻给出估计，从而有效地克服了输入估计算法和变维滤波各自在系统模型单一和机动运行开始时刻估计不精确方面的缺陷。考虑到多传感器信息融合系统可给出比单传感器更为精确的结果，基于带输入估计变维滤波，将系统状态融合和确定性输入融合相结合,提出了一种多传感器带输入估计变维滤波融合算法。系统仿真结果表明，该算法可以有效地提高估计精度,适用于机动目标跟踪。