A congeneric multi-sensor data fusion algorithm and its fault-tolerance

In this paper, a multi-sensor data fusion algorithm was put forward. First, based on the statistics, the standard deviation relating to the fusion value of sensors' measured data was used to estimate sensor's variance. Secondly, by using this estimate variance, a new adaptive weighted fusion algorithm for multi-sensor data fusion, as well as its iterative form, was put forward. Then, a “memory attenuation factor” was introduced to improve this algorithm to weaken the adverse effect from the old data. At last, ample simulation work was presented, in which this algorithm and the adaptive weighted estimate fusion algorithm in some references were applied with sensor no-fault and fault conditions. This result showed that this algorithm effectively captured the sensor fault character, and timely adjusted the sensor fusion weight value, so as to ensure much better fusion result. In all, this algorithm brings satisfying performance in respect of accuracy and tolerance to sensor fault for multi-sensor data fusion.