基于模糊RBF神经网络的辐射源识别

针对模糊识别系统的不足,为了提高辐射源识别系统的识别正确率,构建了基于模糊RBF神经网络的辐射源识别系统,提出了一种等价型模糊RBF神经网络的结构和学习算法,采用五层神经网络结构来实现模糊系统的模糊化和规则推理,神经网络的所有节点和参数对应了模糊系统的隶属函数和推理过程.在仿真实验中,分别采用模糊识别系统、并联型模糊RBF神经网络、结构等价型模糊RBF神经网络进行辐射源识别,给出了三种算法在相同噪声环境下的仿真结果,表明等价型模糊RBF效神经网络有较高的正确识别率,具有更强的抗干扰能力,但运算量相对较大.

Emitter Identification Based on Fuzzy RBF Neural Network

In order to overcome shortage of fuzzy identification system and to increase accurate identification degree, a emitter indentification system based on equivalence type fuzzy RBF neural network was constructed. For the network, five-layer RBF neural network was applied to realize fuzzy rule inference, with all of the nodes and parameters corresponded to membership function and inference process of fuzzy system. For the simulation test, fuzzy identification method, parallel type fuzzy RBF neural network and equivalence type fuzzy RBF neural network are adopted, respectively, to identify emitter under the same noise, the simulation results indicate the last one has the highest accurate identification degree and strongest anti-noise ability, but needs more computation amount.