关键词: 大中型化工装置, 无人机(UAV), 粒子群优化算法(PSO), 泄漏源定位, 主动嗅觉

Abstract:

In order to address frequent hazardous gas leaks in large and medium-sized chemical plant areas, this study proposes a leak source localization method based on a multi-strategy improved PSO(MSPSO) algorithm, leveraging the collaborative capabilities of a small number of UAVs. First, considering the physical constraints UAVs face during actual movement, an acceleration control strategy was integrated into PSO algorithm. Simultaneously, the chemical plant area was divided into distinct zones to more accurately simulate the UAVs' flight states during the search process. Second, an upwind search strategy was introduced based on diffusion characteristics of leak sources, utilizing wind direction information to accelerate the search process. Third, to prevent UAVs from getting stuck in pseudo-leak sources, Cauchy mutation perturbations and simulated annealing mechanisms were employed to enhance the UAVs' ability to escape local optima. Finally, a three-dimensional simulation environment for large and medium-sized chemical plant areas was established to compare and analyze the performance of various swarm intelligence algorithms in simulated scenarios. The results indicate that MSPSO exhibits faster convergence and higher localization success rates, with performance better meeting the leakage source localization requirements of large-to-medium-scale chemical plant areas.

Key words: large and medium-sized chemical plants, unmanned aerial vehicles(UAV), particle swarm optimization(PSO), leak source localization, active olfaction