Abstract: In order to solve the problems of miscellaneous gas types in the environment of the launch site of super heavy rockets, serious interference of outdoor temperature and humidity in gas concentration monitoring precision, and lack of accurate data support for the assessment of environmental pollution for rocket launching, this paper proposed a design of multi-gas concentration monitoring system. On the basis of completing the hardware design of the system, an optimized BP neural network algorithm based on hybrid genetic algorithms and particle swarm optimization (GA-PSO-BP) was proposed. The concentration monitoring accuracy of volatile organic compounds (VOC) such as CO, SO₂ and CH₄ in the launch site with temperature and humidity compensation was studied. The experimental results show that the maximum concentration error in the node data from the sensing layer at the front end of the system to the measurement and control hall at the back end of the launch site is no more than 1.12%, indicating that the compensation ability is superior. The design of the proposed system is of great significance for the accurate monitoring of multi-gas concentration in the environment of launch site.