基于改进型免疫遗传算法对网格中独立任务调度问题的研究

Independent task scheduling based on improved immune genetic algorithm in grid

在研究现有任务调度算法的基础上,借鉴生物免疫系统原理中抗体多样性产生及保持机理,定义了基于抗体的矢量距离、亲合力及浓度相关的选择概率,这样可以在进化过程中保留优秀个体,同时抑制抗体群陷于同一极值而停止进化的早熟现象;另一方面,提出父子竞争(PCC)交叉算子和基于浓度调节机制的变异概率,提高抗体群进化速度的同时保留优秀抗体.仿真实验结果表明,算法与其它调度算法比较,更能有效地实现资源的分配,可以成功应用于网格环境下独立任务调度.

Based on the research of existing scheduling algorithms,this paper introduces the mechanism of producing and preserving the diversity of antibodies in organismal immune system into evolutionary algorithm,the selection probability based on the distance vector,affinity and concentration is defined,so it can retain high quality anti-bodies and inhibit their prematurity in the process of evolution.On the other hand,this paper presents PCC（Father and Son Competition） crossover and mutation probability based on concentration adjustment mechanism improving the speed of evolution and retaining high quality antibodies.Simulation results show that this algorithm is more effective to the allocation of resources compared with other algorithms,it can be successfully applied to the independent task scheduling in grid.