A Mixed Coding Scheme of a Particle Swarm Optimization and a Hybrid Genetic Algorithm with Sequential Quadratic Programming for Mixed Integer Nonlinear Programming in Common Chemical Engineering Practice

In this paper, mixed integer nonlinear programming (MINLP) is optimized by PSO_GA-SQP, the mixed coding of a particle swarm optimization (PSO), and a hybrid genetic algorithm and sequential quadratic programming (GA-SQP). The population is separated into two groups: discrete and continuous variables...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:Chemical engineering communications Ročník 204; číslo 8; s. 840 - 851
Hlavní autori: Chanthasuwannasin, Manatsanan, Kottititum, Bundit, Srinophakun, Thongchai
Médium: Journal Article
Jazyk:English
Vydavateľské údaje: Philadelphia Taylor & Francis 03.08.2017
Taylor & Francis Ltd
Predmet:
Chemical engineering
Continuity (mathematics)
Fitness
Genetic algorithm
Genetic algorithms
Heat exchanger network
Mixed integer
Mixed integer nonlinear programming
Mixed-coding
Nonlinear programming
Optimization
Particle swarm
Particle swarm optimization
Quadratic programming
Sequential quadratic programming
Variables
ISSN:0098-6445, 1563-5201
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Popis
Shrnutí:In this paper, mixed integer nonlinear programming (MINLP) is optimized by PSO_GA-SQP, the mixed coding of a particle swarm optimization (PSO), and a hybrid genetic algorithm and sequential quadratic programming (GA-SQP). The population is separated into two groups: discrete and continuous variables. The discrete variables are optimized by the adapted PSO, while the continuous variables are optimized by the GA-SQP using the discrete variable information from the adapted PSO. Therefore, the population can be set to a smaller size than usual to obtain a global solution. The proposed PSO_GA-SQP algorithm is verified using various MINLP problems including the designing of retrofit heat exchanger networks. The fitness values of the tested problems are able to reach the global optimum.
Bibliografia:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0098-6445
1563-5201
DOI:10.1080/00986445.2017.1294583