机构地区: 华南理工大学自动化科学与工程学院
出 处: 《计算机工程与设计》 2007年第5期1104-1107,共4页
摘 要: 阐述了免疫系统抗体网络的机理和特点,深入分析了抗体网络与常用的免疫算法和Hopfield神经网络异同。通过不断更新输入模式(抗原)和采用最优保存策略,将基于克隆选择的竞争学习算子、自动生成网络结构、剪枝算子和低频变异用于进化操作,提出一种新的基于抗体网络的免疫算法,用于函数优化问题。实验结果表明新算法可行有效。与常用的免疫算法、Hopfield神经网络优化算法比较,新算法具有较好的全局搜索能力和较快收敛速度。 The theory of the antibody network of immune system is described, and the similarities and differences between antibody network, immune algorithm and Hopfield network is clearly outlined. By updating input pattern (antigen) each iteration and using optimum maintaining scheme, a novel immune algorithm based on antibody network is proposed to solve optimization problem. In which, clone select-based competitive learning operator, automatic generation of the network structure, network pruning operator and small mutation are used to perform evolution. Simulation results indicate the approach is feasible and effective. Comparison with conventional immune algorithm and Hopfield network, it shows that the new algorithm has the advantage ofglobal searching ability and faster convergence rate.
领 域: [自动化与计算机技术] [自动化与计算机技术]