自适应PBIL算法求解一类动态优化问题

Adaptive PBIL algorithm for a class of dynamic optimization problems

在不确定环境中,环境的变化总是以一定的概率发生,本文把何时变化看作随机变量,其满足一定的统计规律,由此归纳出一类动态优化问题。对于此类动态优化问题的求解,提出了自适应PBIL(Population-based incremental learning algorithm)算法。算法中利用随机变量的概率自适应地调整当前代群体的概率模型,增加种群多样性,快速适应环境的变化。应用两个动态优化问题进行了仿真实验。实验结果表明,与传统PBIL算法相比,自适应PBIL算法能够快速跟踪最优解的变化。

In an uncertain environment, the environmental changes always occur with probabilities. In this paper the moment when a change occurs is considered as a random variable, which obeys certain distribution, and the dynamic problems possess such features are classified as a class of dynamic optimization problems. Then an adaptive population-based incremental learning （PBIL） algorithm is proposed to solve the class of dynamic optimization problems. This algorithm applies the adaptive probability of random variable to regulate the probable model of the current population. The objectives are to increase the population diversity and to rapidly adapt the environmental changes. Results of case study show that compared with traditional PBIL algorithm, the proposed adaptive PBIL algorithm can track the dynamic solution reliably and accurately.