摘要
为了应对传统粒子群算法(PSO)存在的初始位置不均匀、易达到局部最优、搜索精度不高等问题,提出了一种基于改进Sine混沌映射的新型PSO算法。采用一种改进的Sine混沌映射技术代替传统的伪随机数方法生成初始粒子种群,以丰富种群的多样性。在原始基本位置更新公式的基础上增加两种新的位置更新机制,并分别引入一个高斯变异算子,以实现算法勘探性能和开发性能之间的动态平衡,以及在迭代过程中使粒子有效跳出局部最优。在由7个单峰函数、6个多峰函数和10个固定维函数组成的基准测试函数和3个带约束经典工程优化设计问题上对所提出算法开展仿真实验,并与其他几种流行的PSO变体进行对比。仿真结果表明:与其他PSO变体相比,基于改进Sine混沌映射的新型PSO算法具有更快的收敛速度和更高的寻优精度,对于基准测试函数的寻优结果有20个排名第一,约为总测试函数的87%;该算法在压力容器和工字梁设计优化中,综合性能排在第一位,应可用于解决一些实际工程优化问题。
In order to address the problems of uneven initial positions,ease of reaching local optimum,and low search accuracy in traditional particle swarm algorithm(PSO),a novel PSO algorithm based on an improved Sine chaotic mapping is proposed.The improved Sine chaotic mapping technique is used instead of the traditional pseudorandom number method for generating the initial particle population to enrich the population diversity.Two new position update mechanisms are added to the original basic position update formula.A Gaussian mutation operator is introduced to achieve a dynamic balance between the exploration and exploitation performance of the algorithm,as well as to help particles effectively jump out of the local optima during the iteration process.For three classical engineering optimization design problems with constraints,simulation experiments are performed for the proposed algorithm based on a benchmark test function consisting of seven singlepeaked functions,six multipeaked functions and ten fixeddimensional functions.This algorithm is then compared with several other popular PSO variants.Simulation results show that the novel PSO algorithm based on improved Sine chaotic mapping has faster convergence speed and higher optimizationseeking accuracy than those of other PSO variants.For the benchmark test functions,it ranked first in 20 of them,accounting for about 87%of the total testing functions.The proposed algorithm ranked first in the overall performance of pressure vessel and Ibeam design optimization,and can be used to solve some practical engineering optimization problems.
作者
刘磊
姜博文
周恒扬
浦晨玮
钱鹏飞
刘波
LIU Lei;JIANG Bowen;ZHOU Hengyang;PU Chenwei;QIAN Pengfei;LIU Bo(School of Mechanical Engineering,Jiangsu University,Zhenjiang,Jiangsu 212013,China;The State Key Laboratory of Fluid Power and Mechatronic Systems,Zhejiang University,Hangzhou 310027,China;Electrical Engineering Department,King Fahd University of Petroleum and Minerals,Dhahran 31261,Saudi Arabia)
出处
《西安交通大学学报》
EI
CAS
CSCD
北大核心
2023年第8期182-193,共12页
Journal of Xi'an Jiaotong University
基金
国家自然科学基金资助项目(52075223)
中国博士后科学基金资助项目(2021M691308)
江苏省研究生科研与实践创新计划资助项目(KYCX23_3731)。
关键词
粒子群算法
混沌映射
高斯变异
基准函数
工程问题
PSO algorithm
chaotic mapping
Gaussian mutation
benchmark functions
engineering problems
