关键词: 数据融合, 自熵, 限定记忆, 冗余信息

Abstract: In the case that any prior knowledge was unknown, a new fusion algorithm was proposed in order to deal with the defect of the existing algorithm. In the algorithm, for better fusion accuracy, the mean and entropy based on multi-sensor support degree were used to excavate the redundant information sufficiently, and then the weight coefficient of sensor could be determined. In addition, a limited memory fusion was used to avoid data saturation when the old measurement information was increasing, and ensured delicacy degree when the measurement was changing. To verify the effectiveness of this algorithm, three algorithms were used to detect the sample data. The simulation result shows the weight distribution gained through the new algorithm is more effective, and the accuracy of fusion can be further improved.

Key words: data fusion, entropy, limited memory, redundant information