Abstract: The ventilation efficiency is low and energy consumption is high in high-altitude tunnel construction. To address these challenges, a ventilation data simulation and optimization system based on generative adversarial networks (GANs) is developed. This system uses GANs technology to generate high-precision construction environment data, replacing traditional computational fluid dynamics simulation methods. By learning the distribution characteristics of multidimensional data in the construction environment, it generates simulated data in real-time. Subsequently, genetic algorithms and particle swarm optimization are combined to optimize the parameter configuration of the ventilation system, enhancing the system′s response speed and stability. Experimental results show that: (1) The GANs-based ventilation optimization strategy significantly improves ventilation efficiency and reduces energy consumption, with ventilation efficiency increasing by 12% and energy consumption decreasing by 8%. (2) The error between GANs-generated data and actual measurement data is less than 15%, demonstrating high accuracy and adaptability. (3) Tests conducted in various construction environments confirm the effectiveness and robustness of the optimization strategy under complex conditions, especially in extreme conditions. (4) By integrating GANs technology, the optimization efficiency of the tunnel construction ventilation system is significantly improved, reducing safety risks during construction and providing new technological support for intelligent management of high-altitude tunnel construction.

Key words: tunnel construction ventilation, generative adversarial networks, ventilation optimization, genetic algorithm, particle swarm optimization, data simulation