基于全局边际排序和自适应参数调节的多目标粒子群算法

Multi-Objective Particle Swarm Optimization Based on Global Marginal Ranking and Adaptive Parameter Adjustment

摘要: 多目标粒子群优化算法作为一种强大的优化技术，在各个领域得到了广泛的应用。然而，对于大多数现有的算法来说，同时在收敛性和多样性方面表现良好仍然是一项具有挑战性的任务。针对上述挑战，本文提出了一种基于全局边际排序和自适应参数调节的多目标粒子群优化算法(GGRAW)。在多目标粒子群算法中，飞行参数、惯性权重ω、加速度系数c₁和c₂对于实现种群开发和勘探之间的平衡非常重要。该策略利用个体在客观空间中的位置信息，以获得整个种群的优势边际。该方法不仅考虑了种群的分布，还考虑了个体的相关信息。通过自适应参数的调节更加促使种群中的粒子向真实的Pareto前沿移动；将GGRAW算法与3种经典的多目标粒子群算法在ZDT、DTLZ和WFG系列的部分测试函数上进行仿真实验；试验结果表明：相对于其他的几种经典算法，GGRAW算法表现出较好的收敛性和多样性，因此，GGRAW算法可以作为求解多目标优化问题的有效算法。

Abstract: As a powerful optimization technology, multi-objective particle swarm optimization algorithm has been widely used in various fields. However, performing well in both convergence and diversity remains a challenging task for most existing algorithms. To address these challenges, this paper proposes a multi-objective particle swarm optimization algorithm (GGRAW) based on global marginal sorting with adaptive parameter adjustment. In multi-objective particle swarm optimi-zation, flight parameters, inertia weight ω, acceleration coefficients c₁ and c₂ are very important to achieve a balance between population exploitation and exploration. This strategy uses the position information of individuals in objective space to obtain the dominant margin of the whole population. This method not only considers the distribution of the population, but also considers the relevant information of the individual. By adjusting the adaptive parameters, the particles in the population can move towards the real Pareto frontier. The GGRAW algorithm and three classical multi-objective particle swarm optimization algorithms were simulated on some test functions of ZDT, DTLZ and WFG series. The experimental results show that compared with other classical algorithms, GGRAW algorithm shows better convergence and diversity. Therefore, GGRAW algorithm can be used as an effective algorithm for solving multi-objective optimization problems.