An Algorithm of SelectING Input Measurement Fusion

For measurement fusion of a linear time-invariant system, in past literatures, it´s impliedly consented that all the input measurements will contribute to enhance the fusion precision. In the suspicion of this hypothesis, this paper analysis the fusion principle based on kalman filtering framework and discusses the impact of quantity and quality of the input measurements on fusion accuracy. Based on the conclusion, a new fusion method called selecting input measurement fusion (SIMF) is proposed. The procedure of SIMF is divided into two steps. First, select input measurements by calculating estimated error of each input measurement and selecting smaller ones compared with a threshold. Second, fuse as usual. Theoretical analysis and computer simulation shows that SIMF can effectively improve the accuracy compared with the original algorithm.