人工免疫网络的改进及其在免疫神经控制中的应用

针对人工免疫网络(AIN)搜索能力不强的问题,通过引入新生抗体的进化环节,用遗传算子代替克隆变异作为主要的搜索手段,增强了网络的搜索能力.改进网络被用于免疫神经控制中抽取控制经验,形成知识库,并指导遗传算法在线搜索.仿真结果表明,改进后人工免疫网络的搜索能力得到较大提高;通过引入经验指导,能加快遗传算法的收敛速度,进一步增强了免疫神经控制的实时性.

Improvement of artificial immune network and its application in immunized neurocontrol

The artificial immune network(AIN)has strong self-organizing ability, but its search performance isn't very good. To resolve this problem, genetic operator was applied to act as the main search approach instead of clone hypermutation by introducing the evolution of antibody into algorithm. The improved AIN was applied to extract knowledge, form knowledge base and conduct on-line search of the neural network model of compensate controller in immunized neurocontrol. Simulation results confirm that the search ability of the improved AIN is enhanced; and by introducing the knowledge base, the genetic algorithm can find the optimal model more quickly.