Li, QiangQiangLi2025-11-272025-11-272025-02-025th International Conference on Big Data Analytics for Cyber-Physical System in Smart City, BDCPS 2023https://hdl.handle.net/11420/59235High-dimensional multi-objective optimization (MOO) problems have broad application prospects in many fields such as engineering design, resource management, artificial intelligence, etc. They are complex and ever-changing, and there is an urgent need to develop more effective and intelligent solving methods. This article aims to combine the powerful learning ability of neural networks with the global optimization ability of GA (Genetic Algorithm) to design a new type of high-dimensional MOO algorithm. On this basis, this article adopted a deep learning based solution and generation method, combined with the evolutionary mechanism of genetic algorithm, to achieve efficient solution of multiple standard test functions. Especially for optimal solutions under high-dimensional and complex constraints, it has strong robustness and adaptability. At the same time, explore the use of neural networks for parameter adjustment and adaptability, so that they can maintain stability and high efficiency in various optimization environments. Compared with existing high-dimensional MOO methods, the significant decrease in optimal solution distance as the number of iterations increased from 100 to 500 revealed the efficient performance of genetic algorithms supported by neural networks or other intelligent systems in MOO problems. The research results of this article provide a new approach and method for solving high-dimensional MOO problems, promoting the development of intelligent optimization algorithms at both theoretical and practical levels.enGenetic algorithmMulti-objective optimization algorithmNeural networkOptimal solution distanceTechnology::620: EngineeringConference Paper10.1007/978-981-96-0208-7_2Conference Paper