A New Method of Multi-Sensor Data Fusion

For the problem of multi-sensor data fusion on the decision level, an evidence fusion method based on the modified support degree distribution method is proposed. First, starting from the viewpoint of classical evidence fusion formula, the reason why confidence paradox is generated is analyzed and the general rule of generating confidence paradox is achieved; second, with analysis of support degree distribution fusion method, the problems in application of such method in overall reliability and conflict information are presented, and the method is modified through introducing hyper-sphere fitness and conflicting value distribution function; finally, the modified fusion effect is validated with numerical simulation.