基于环境识别策略的多目标自适应粒子群算法及应用

打开文本图片集

关键词：粒子群算法；进化计算；自适应学习；多目标优化；多阶段生产问题中图分类号：TP301.6 文献标志码：A 文章编号：1001-3695（2025）10-011-2980-09doi：10.19734/j.issn.1001-3695.2025.04.0090

Multi-objective adaptive particle swarm optimization algorithm with environment recognition and application

Wu Baotongla，1b，ShuRuoqi2a，2b，Chen Zhixiangla，bt （1.a.Schoolfsinss，bbatorfigatDngentaioDecsionking，unYatnUniesityuo 51006，Cina2oofteligentafcuing，bZoingCieyLboratorofAdacedufacingo&p ment，Guangdong TechnologyCollege，ZhaoqingGuangdong5261oO，China）

Abstract：Toaddress theissuesofthestandardmulti-objectiveparticleswarmoptimizationalgorithm，suchasgetingtrapped inlocal optima，overlyfastconvergence，and low precisionduring theoptimizationprocess，thispaper proposeda multiobjectiveadaptiveparticleswarmalgorithm basedonenvironmental recognition.Theinitialpopulation wasgeneratedusing an optimalpointsetstrategytoensure individualsareuniformlydistributedinthesolutionspace.Itemployedanonlinearinertia weightmechanismandacrossver mutation strategytopreventthealgorithmfromconverging tooquicklyduring thesearch process.Aditionally，itintroducedanadaptivelearningoperatorandanadaptivejumpcolaborationoperatorbasedonenvironmentalrecognition，which facilitateinteractionandlearningamong particlesbyself-identifying thediversitylevelofthe population inthe solution spaceand thecrowdingdegree within the particle'slocal niche.Comparativesimulation experments onmultiplebenchmark functionsshowthattheimprovedalgorithmsignificantlyenhancesbothsearchcapabilityandoptimizationprecision.Finall，apractical multi-stageproductioncase withNP-hardvalidatestheefectiveness of algorithm.

Keywords：particleswarmalgorithm；evolutionarycomputation；adaptive learning；multi-objectiveoptimization；multi-stage production problem

0 引言

多目标优化问题是现代工程与科学领域中的核心挑战之二[1]，其目标是在多个相互冲突的优化目标之间寻找最优解，且该问题广泛存在于生产、交通以及能源[2]等领域内。（剩余20647字）