论文部分内容阅读
在分析径向基函数网络（ＲａｄｉａｌＢａｓｉｓＦｕｎｃｔｉｏｎＮｅｔｗｏｒｋ，ＲＢＦＮ）的基础上，提出了一种更适合于目标识别的基于模糊聚类的径向基函数网络（ＦｕｚｚｙＣｌｕｓｔｅｒｉｎｇＢａｓｅｄＲａｄｉａｌＢａｓｉｓＦｕｎｃｔｉｏｎＮｅｔｗｏｒｋ，ＦＣＢＲＢＦＮ）。这种网络利用模糊聚类方法，根据训练样本的空间分布确定网络的结构，利用聚类结果中的隶属度函数值控制高斯核函数形状参数。理论分析还表明，此径向基函数网络具有比一般径向基函数网络更强的泛化能力。利用一种战场侦察雷达获取的回波数据进行实验，结果表明，基于模糊聚类的该径向基函数网络的分类结果优于一般径向基函数网络。
Based on the analysis of Radial Basis Function Network (RBFN), a fuzzy clustering based radial basis function network (FCBRBFN) based on fuzzy clustering is proposed. The network uses fuzzy clustering method to determine the network structure according to the spatial distribution of training samples. The shape parameter of Gaussian kernel function is controlled by the membership function value in the clustering result. Theoretical analysis also shows that this radial basis function network has more generalization ability than the general radial basis function network. Experiments on the echo data acquired by a battlefield reconnaissance radar show that the classification results of the radial basis function network based on fuzzy clustering are superior to the general radial basis function networks.