基于测量值模糊贴近度时空融合的多传感器融合方法

针对实际多传感器目标跟踪系统中,传感器测量值中的野值对状态估计的不利影响,该算法利用概率源合并理论和非负矩阵特征向量理论,算出k时刻空间上各传感器测量值与其他传感器测量值的综合贴近度,同时将k个时刻的空间上融合的模糊贴近度进行时间融合,以此确定每个传感器的权重。仿真结果表明,该算法有效地抑制了测量野值对测量融合值的不利影响,提高了系统的跟踪准确度。

Multi-sensor Data Fusion Method Based on Measurement Fuzzy Clingy Degree Spatial-Temporal

In actual target tracking system of the sensors data fusion,for the bad influences of the measurement values including outliers to state estimate,the algorithm calculates the fuzzy clingy degree of measurement value of arbitrary two sensors on space based on the theories of probability source combination and characteristic vector of non-negative matrix,at the same time the synthetic fuzzy clingy degree of each sensor measurement value and other sensors measurement values is fused in time,and the weight of each sensor.The simulation results show that the modified algorithms can restrain the bad influences of outliers and improve the tracking accuracy.