Guided golden jackal optimization using elite-opposition strategy for efficient design of multi-objective engineering problems

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F61989100%3A27240%2F23%3A10253150" target="_blank" >RIV/61989100:27240/23:10253150 - isvavai.cz</a>

Výsledek na webu

<a href="https://link.springer.com/article/10.1007/s00521-023-08850-0" target="_blank" >https://link.springer.com/article/10.1007/s00521-023-08850-0</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1007/s00521-023-08850-0" target="_blank" >10.1007/s00521-023-08850-0</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Guided golden jackal optimization using elite-opposition strategy for efficient design of multi-objective engineering problems

Popis výsledku v původním jazyce

Multi-objective optimization (MOO) issues that are encountered in the realm of real engineering applications are characterized by the curse of economically or computationally expensive objectives, which can strike insufficient performance evaluations for optimization methods to converge to Pareto optimal front (POF). To address these concerns, this paper develops a guided multi-objective golden jackal optimization (MOGJO) to promote the coverage and convergence capabilities toward the true POF while solving MOO issues. MOGJO embeds four reproduction stages during the seeking process. Firstly, the population of golden jackals is initialized according to the operational search space and then the updating process is performed. Secondly, an opposition-based learning scheme is adopted to improve the coverage of the Pareto optimal solutions. Thirdly, an elite-based guiding strategy is incorporated to guide the leader golden jackal toward the promising areas within the search space and then promote the convergence propensity. Finally, the crowding distance is also integrated to provide a better compromise among the diversity and convergence of the searched POF. To evaluate the MOGJO&apos;s performance, it is analyzed against sixteen frequently utilized unconstrained MOO issues, five complex constrained problems, four constrained engineering designs, and real dynamic economic-emission power dispatch (DEEPD) problem. The experimental results are performed using the generational distance (GD), hypervolume (HV), spacing (SP) metrics to validate the efficacy of the proposed methods, which affirms the progressive and competitive performance compared to thirteen state-of-the-art methods. Finally, the results of the Wilcoxon rank sum test with reference to GD and HV exhibited that the proposed algorithm is significantly better than the compared methods, with a 95% significance level. Furthermore, the results of the nonparametric Friedman test were performed to detect the significant of average ranking among the compared algorithm, where the results confirmed that the proposed MOGJO outperforms the best algorithm among thirteen state-of-the-art algorithms by an average rank of Friedman test greater than 41% while outperforming the worst one, MOALO, by 84% for ZDT and DTLZ1 suits. Additionally, the proposed algorithm saved the overall energy cost and total emission of the DEEPD problem by 1.89%, and 1.48%, respectively, compared with the best existing results and thus, it is commended to adopt for new applications.

Název v anglickém jazyce

Guided golden jackal optimization using elite-opposition strategy for efficient design of multi-objective engineering problems

Popis výsledku anglicky

Multi-objective optimization (MOO) issues that are encountered in the realm of real engineering applications are characterized by the curse of economically or computationally expensive objectives, which can strike insufficient performance evaluations for optimization methods to converge to Pareto optimal front (POF). To address these concerns, this paper develops a guided multi-objective golden jackal optimization (MOGJO) to promote the coverage and convergence capabilities toward the true POF while solving MOO issues. MOGJO embeds four reproduction stages during the seeking process. Firstly, the population of golden jackals is initialized according to the operational search space and then the updating process is performed. Secondly, an opposition-based learning scheme is adopted to improve the coverage of the Pareto optimal solutions. Thirdly, an elite-based guiding strategy is incorporated to guide the leader golden jackal toward the promising areas within the search space and then promote the convergence propensity. Finally, the crowding distance is also integrated to provide a better compromise among the diversity and convergence of the searched POF. To evaluate the MOGJO&apos;s performance, it is analyzed against sixteen frequently utilized unconstrained MOO issues, five complex constrained problems, four constrained engineering designs, and real dynamic economic-emission power dispatch (DEEPD) problem. The experimental results are performed using the generational distance (GD), hypervolume (HV), spacing (SP) metrics to validate the efficacy of the proposed methods, which affirms the progressive and competitive performance compared to thirteen state-of-the-art methods. Finally, the results of the Wilcoxon rank sum test with reference to GD and HV exhibited that the proposed algorithm is significantly better than the compared methods, with a 95% significance level. Furthermore, the results of the nonparametric Friedman test were performed to detect the significant of average ranking among the compared algorithm, where the results confirmed that the proposed MOGJO outperforms the best algorithm among thirteen state-of-the-art algorithms by an average rank of Friedman test greater than 41% while outperforming the worst one, MOALO, by 84% for ZDT and DTLZ1 suits. Additionally, the proposed algorithm saved the overall energy cost and total emission of the DEEPD problem by 1.89%, and 1.48%, respectively, compared with the best existing results and thus, it is commended to adopt for new applications.

Druh

J_imp - Článek v periodiku v databázi Web of Science

CEP obor

—

OECD FORD obor

10200 - Computer and information sciences

Projekt

—

Návaznosti

S - Specificky vyzkum na vysokych skolach

Rok uplatnění

2023

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název periodika

Neural Computing and Applications

ISSN

0941-0643

e-ISSN

1433-3058

Svazek periodika

35

Číslo periodika v rámci svazku

34

Stát vydavatele periodika

US - Spojené státy americké

Počet stran výsledku

32

Strana od-do

—

Kód UT WoS článku

001035531400005

EID výsledku v databázi Scopus

—