中文摘要：

在粗糙集和数据融合基本理论的基础上，研究基于粗糙集理论和神经网络相结合的数据融合方法。由于粗糙集理论能够有效地简化知识，降低特征的维数，将粗糙集理论和神经网络结合起来，利用基于信道容量的知识相对约简算法对输入信息进行简化，剔除冗余信息，从而缩减了神经网络的规模，提高网络的收敛性和融合系统的识别率，达到提高整个融合系统效率的目的。将改进后的融合系统与传统的神经网络融合的效率进行比较，通过实例说明了该方法的有效性。

英文摘要：

Based on rough set and basic theory of data fusion, the data fusion algorithm combining rough set theory with BP neural network is studied. Since rough set theory can effectively simplify information, cut down the tagged dimension. combining rough set theory with neural networks, using channel capacity of knowledge relative reduction algorithms to simplify the input information. Rough set theory is first used to process the sample data,and eliminate the redundant information,then reduce the scale of neural network,improve the identification rate,and efficiency of the whole data fusion system. Effectiveness of the improved algorithm is demonstrated by an example compared with the traditional neural network system.