一种适用于功能电路的自组织神经网络模型构建

Construction of a Self-organizing Neural Network Model Suitable for Function Circuits

借鉴信息在生物体神经系统内的传递过程及电子电路的构造特点,提出了一种适用于功能电路的自组织神经网络模型及构建方法。首先提出了神经元、神经网络模型,分析了信号传递过程;然后描述了功能电路到神经网络的映射关系,给出了神经网络自组织特性实现方法;最后以2×2乘法器为功能电路,验证了此模型对功能电路的承载性及自组织特性。结果表明:模型与电子电路具有较好的对等映射关系,在权值变换方面也可极大降低硬件实现的难度。同时,该模型可望为复杂电磁环境下电子系统的仿生防护研究提供一定的理论和实验参考。

Based on the transfer process of information in the biological nervous system and the characteristics of the circuit structure, a serf-organizing neural network model suitable for function circuits is proposed and con- structed. Firstly, the nerve cell, the neural network model are proposed, and the information transfer process is analysed. Then, the mapping relations between function circuit and neural network are described, and the im- plementation method of self-organizing neural network is presented. Finally, the carrying capacity and self-orga- nizing feature of the proposed model is verified by a 2 × 2 multiplier. Experiment results demonstrate that the proposed model has good mapping relations with the electronic circuit, and it can greatly reduce the difficulty of hardware implementation in the weight change. Meanwhile, this model can provide certain theoretical and exper- imental references for the bionic protection of electronic system under the complex electromagnetic environment.