E-CSO-2021-阅读笔记

An Efficient Competitive Swarm Optimizer for Solving Large-Scale Multi-objective Optimization Problems

Li Y., Li L., Lin Q., Ming Z. (2021) An Efficient Competitive Swarm Optimizer for Solving Large-Scale Multi-objective Optimization Problems. In: Huang DS., Jo KH., Li J., Gribova V., Bevilacqua V. (eds) Intelligent Computing Theories and Application. ICIC 2021. Lecture Notes in Computer Science, vol 12836. Springer, Cham. https://doi.org/10.1007/978-3-030-84522-3_6

摘要

The search ability of most existing MOEAs on solving LSMOPs is still weak.

现有的大多数MOEAs在求解LSMOPs时的搜索能力仍然较弱。

To address this issue, an efficient competitive swarm optimizer with a strong exploration ability, denoted as E-CSO, is presented in this paper, which designs a novel three-particle-based particle updating strategy to improve the search efficiency.

为了解决这一问题，本文提出了一种具有较强探索能力的高效竞争群优化器E-CSO，该优化器设计了一种新的基于三粒子的粒子更新策略来提高搜索效率。

1、引言

Nevertheless, most existing MOEAs for solving LSMOPs were proposed based on the division of decision variables, which may encounter some problems when solving some problems with complicated PFs or PSs.

然而，现有的求解LSMOPs的MOEAs大多是基于决策变量的划分而提出的，这在解决一些具有复杂PFs或PSs的问题时可能会遇到一些问题。

In addition, few MOEAs for solving LSMOPs optimize all the decision variables simultaneously, due to the inefficient search ability of existing search operators (i.e., simulated binary crossover and differential evolution).

此外，由于现有搜索算子(即模拟二进制交叉和差分进化)的搜索能力不足，很少有求解LSMOPs的MOEAs同时优化所有决策变量。

We design an efficient competitive swarm optimizer in this paper, denoted as E-CSO, which proposes a novel three-particle-based particle updating strategy with a strong search ability.

本文设计了一种高效的竞争群优化算法E-CSO，提出了一种新的基于三粒子的具有较强搜索能力的粒子更新策略。

2、相关工作

2.1 Competitive Swarm Optimizer(CSO)

In order to enhance the ability of PSO solving problems that have a larger number of decision variables, a novel PSO variant, namely CSO, has been proposed for solving large-scale MOPs, which randomly selects two particles every time and the velocity and position of the worse particle, called loser particle, is updated by the better one, called the winner particle.

为了提高PSO解决决策变量较多的问题的能力，提出了解决大规模MOP的新型PSO变种，即CSO，每次随机选择两个粒子，而更差粒子（称为loser particle）的速度和位置则由更好的粒子更新（称为winner particle）。

3、E-CSO

3.1 算法框架

如图：

主要包括：

• 初始化（line 1-2）：一组均匀分布的参考点$R$、种群$P$
• 更新粒子（line 4）：更新$P$