基于小波网络和证据理论的多源信息融合系统研究

Research on Multi-Source Information Fusion System Based on Wavelet Neural Network and Evidential Theory

在多源信息融合中 ,不确定、不完整和冗余现象普遍存在。为了解决这个复杂问题 ,本文首先利用相空间重构理论对输入的信息在相空间中进行重构 ,以充分提取相关信息。然后在重构相空间上 ,利用模糊理论、小波网络和遗传算法对上述重构信息进行了时间域信息融合。最后 ,利用 D-S证据理论将时间域融合结果进行了空间域信息融合 ,并根据决策规则进行了决策。结果表明 ,据此形成的分布式多源信息融合系统具有良好的目标探测能力。

In order to solve general phenomena of uncertainty, unintegrity and redundancy in the multi source information fusion, firstly using a phase reconstruction theory, input information is reconstructed in the phase space and correlative information in the input information is extracted. Then using fuzzy theory, wavelet neural network and genetic algorithm, the above reconstructed information is fused from the time domain in the reconstructed phase space. Finally, using a D S evidence theory, result of the time domain fusion is fused from the space domain, and decision making is also done by a rule of decision making. Practical application shows that the distributed multi source information fusion system based on the ideology has famous capability for target detection, resisting the environmental disturbance and fault tolerance.