ANALYSIS OF THE IMPACT OF GENETIC ALGORITHM PARAMETERS ON THE EFFECTIVENESS OF SOLVING THE PROBLEM OF ALLOCATING LIMITED RESOURCES IN SCENARIO-BASED ECONOMIC CONDITIONS
Abstract
The article examines the impact of genetic algorithm parameters on the efficiency of solving the constrained resource allocation problem under scenario-based economic conditions. A discrete optimization model is considered, incorporating a set of alternatives, budget constraints, and logical dependencies between decision components. An experimental sensitivity analysis of the main algorithm parameters is conducted, including population size, crossover and mutation probabilities, elitism degree, and constraint-handling mechanisms. The study identifies the patterns of parameter influence on convergence and search performance, enabling the development of well-founded strategies for tuning evolutionary methods in complex economic environments.
References
Shelekhov I. V., Dziuba O. O. (2009) Doslidzhennia vplyvu parametriv hibrydnoho alhorytmu na funktsionalnu efektyvnist navchannia systemy pidtrymky pryiniattia rishen [Study of the influence of hybrid algorithm parameters on the functional efficiency of decision support system training]. Radioelektronni i komp’iuterni systemy [Radioelectronic and Computer Systems], no. 3, P. 88–91. Available at: http://nbuv.gov.ua/UJRN/recs_2009_3_18 (in Ukrainian).
Bazhan S. (2023) Vykorystannia henetychnykh alhorytmiv iz zastosuvanniam vypadkovykh protsesiv pry rozviazuvanni zadach optymizatsii [Use of genetic algorithms with random processes for solving optimization problems]. Grail of Science, no. 33, P. 253–261. Available at: https://doi.org/10.36074/grail-of-science.10.11.2023.40 (date of access: 31.10.2025).
Two-Archive evolutionary algorithm for constrained multiobjective optimization / K. Li et al. IEEE transactions on evolutionary computation. 2019. Vol. 23, no. 2. P. 303–315. URL: https://doi.org/10.1109/tevc.2018.2855411 (date of access: 31.10.2025).
Composite differential evolution for constrained evolutionary optimization / B.-C. Wang et al. IEEE transactions on systems, man, and cybernetics: systems. 2019. Vol. 49, no. 7. P. 1482–1495. URL: https://doi.org/10.1109/tsmc.2018.2807785 (date of access: 31.10.2025).
Configuring two-algorithm-based evolutionary approach for solving dynamic economic dispatch problems / F. Zaman et al. Engineering applications of artificial intelligence. 2016. Vol. 53. P. 105–125. URL: https://doi.org/10.1016/j.engappai.2016.04.001 (date of access: 31.10.2025).
Kharchenko V. V., Kurdenko O. V. (2025) Zastosuvannia evrystychnykh metodiv optymizatsii dlia rozviazannia zadach rozpodilu resursiv v umovakh nevyznachenosti [Application of heuristic optimization methods for solving resource allocation problems under uncertainty]. Sotsialno-ekonomichnyi stan v umovakh voiennoho chasu: Materialy II Mizhnarodnoi naukovo-praktychnoi konferentsii, Sumy, May 31, 2025. Sumy, P. 46–49. Available at: https://doi.org/10.64076/eecsr250531 (in Ukrainian).
Balancing objective optimization and constraint satisfaction in constrained evolutionary multiobjective optimization / Y. Tian et al. IEEE transactions on cybernetics. 2021. P. 1–14. URL: https://doi.org/10.1109/tcyb.2020.3021138 (date of access: 31.10.2025).
A dual-population algorithm based on alternative evolution and degeneration for solving constrained multi-objective optimization problems / J. Zou et al. Information sciences. 2021. Vol. 579. P. 89–102. URL: https://doi.org/10.1016/j.ins.2021.07.078 (date of access: 31.10.2025).
Vladyslav Kharchenko. Listing of a MATLAB program to study the influence of genetic algorithm parameters. Zenodo. URL: https://doi.org/10.5281/zenodo.17488972 (date of access: 31.10.2025).
Шелехов І. В., Дзюба О. О. Дослідження впливу параметрів гібридного алгоритму на функціональну ефективність навчання системи підтримки прийняття рішень. Радіоелектронні і комп’ютерні системи. 2009. № 3. С. 88–91. URL: http://nbuv.gov.ua/UJRN/recs_2009_3_18.
Бажан С. Використання генетичних алгоритмів із застосуванням випадкових процесів при розв’язуванні задач оптимізації. Grail of science. 2023. № 33. С. 253–261. URL: https://doi.org/10.36074/grail-of-science.10.11.2023.40 (дата звернення: 31.10.2025).
Two-Archive evolutionary algorithm for constrained multiobjective optimization / K. Li та ін. IEEE transactions on evolutionary computation. 2019. Т. 23, № 2. С. 303–315. URL: https://doi.org/10.1109/tevc.2018.2855411 (дата звернення: 31.10.2025).
Composite differential evolution for constrained evolutionary optimization / B.-C. Wang та ін. IEEE transactions on systems, man, and cybernetics: systems. 2019. Т. 49, № 7. С. 1482–1495. URL: https://doi.org/10.1109/tsmc.2018.2807785 (дата звернення: 31.10.2025).
Configuring two-algorithm-based evolutionary approach for solving dynamic economic dispatch problems / F. Zaman та ін. Engineering applications of artificial intelligence. 2016. Т. 53. С. 105–125. URL: https://doi.org/10.1016/j.engappai.2016.04.001 (дата звернення: 31.10.2025).
Харченко В. В., Курденко О. В. Застосування евристичних методів оптимізації для розв’язання задач розподілу ресурсів в умовах невизначеності. Соціально-економічний стан в умовах воєнного часу : Матеріали II Міжнар. науково-практ. конф., м. Суми, 31 трав. 2025 р. Суми, 2025. С. 46–49. URL: https://doi.org/10.64076/eecsr250531.
Balancing objective optimization and constraint satisfaction in constrained evolutionary multiobjective optimization / Y. Tian та ін. IEEE transactions on cybernetics. 2021. С. 1–14. URL: https://doi.org/10.1109/tcyb.2020.3021138 (дата звернення: 31.10.2025).
A dual-population algorithm based on alternative evolution and degeneration for solving constrained multi-objective optimization problems / J. Zou та ін. Information sciences. 2021. Т. 579. С. 89–102. URL: https://doi.org/10.1016/j.ins.2021.07.078 (дата звернення: 31.10.2025).
Vladyslav Kharchenko. Listing of a MATLAB program to study the influence of genetic algorithm parameters. Zenodo. URL: https://doi.org/10.5281/zenodo.17488972 (дата звернення: 31.10.2025).
This work is licensed under a Creative Commons Attribution 4.0 International License.
