.png){kind=icon}
.png){kind=icon}
Return
The genetic algorithm based on generative adversarial network for dynamic multi-objective optimization problem and application
.png){kind=icon}
PRE
AI
DOI：10.1016/j.knosys.2025.114939.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Dynamic multi-objective optimization problems, where objectives, constraints, and parameters vary over time, are highly challenging, particularly in rapidly and accurately tracking the Pareto Front. Existing prediction-based methods often rely excessively on correlations between consecutive environments while neglecting the global distribution characteristics of historical solutions, leading to predicted solutions deviating from the true evolutionary direction. To address this issue, we propose a novel dynamic multi-objective evolutionary algorithm based on Generative Adversarial Network(GAN), called GAN-DMOEA. The method first employs clustering to divide historical Pareto sets into high-quality and low-quality solutions, using the high-quality subset to train the GAN so that the generator can learn their latent distribution. Upon environmental changes, a hybrid strategy combining GAN-generated solutions, differential evolution mutation, and random sampling is adopted to balance exploration and exploitation. Experimental results on the DF benchmark test suite demonstrate that, compared with mainstream algorithms, GAN-DMOEA achieves superior performance in terms of diversity and distribution of solutions, thereby significantly enhancing the effectiveness of dynamic multi-objective optimization. Furthermore, the proposed method is applied to the dynamic hydro-thermal power dispatch problem, further validating its practicality and advantages in real-world applications.
Journal
K
IF:
7.6
Papers:
1.2W
Citations:
4.5W