基于预测策略的动态多目标免疫优化算法

Dynamic Multi-Objective Immune Optimization Algorithm Based on Prediction Strategy

为了有效解决动态多目标优化问题,文中提出了一种新的基于预测策略的动态多目标免疫优化算法.该算法首先采用相似性检测算子较好地检测到环境的变化.同时利用前几个时刻的最优非支配抗体解集建立新的预测模型来预测产生新时刻的初始抗体种群,进一步提高了算法对环境变化的反应能力.此外,通过引入基于两种不同的父代个体选择策略而改进的差分交叉算子来加快算法的收敛速度.文中采用几个典型的标准测试问题验证算法的有效性,实验结果表明,提出的相似性检测算子的预测模型可以提高算法的跟踪能力,而改进的差分交叉算子能够提高算法的收敛性能.

In this paper,a new dynamic multi-objective immune optimization algorithm based on prediction strategy is proposed for solving dynamic multi-objective optimization problems effectively.Firstly a similarity detection operation is used to detect the environment change.Then,a new forecasting model,which is established by the non-dominated antibodies in previous optimum locations,is adopted to generate the initial antibody population in order to improve the ability of responding to the environment change.Moreover,an improved differential evolution crossover operator based on two different selection strategies is introduced to speed the convergence of algorithm.The proposed algorithm is validated on several benchmark testing problems,the experimental result shows that the forecasting model based on the similarity detection operation can improve the tracking ability and the improved differential crossover operation can enhance the convergence.