Rapid construction of safe-zone by active flow field control and sparse-sampling-perception-network for public indoor environments suffering from nerve agents

Nerve agents' attacks in public indoor environments significantly threaten citizens' health. The pattern control of nerve agents is pivotal in efficiently decontaminating and rescuing. We propose an intelligent system that constructs safe zones rapidly for people to take refuge through active flow field control and a sparse-sampling perception network (SSPN). In this work, A 336*206*46.5 m large-scale indoor space, modeling a meeting hall suffering from nerve agents, has been utilized as a typical scenario to simulate the decontamination and safe zone construction process: (1) Firstly, numerous kinds of distribution patterns are constructed by active flow field control method. (2) Secondly, 100 real-world scenarios are simulated based on the computational fluid dynamic (CFD) methods and verified by PIV experiments. (3) Thirdly, a sparse-sampling perception network is proposed to perceive the concentration distribution of nerve agents in the whole space under limited local samplings. (4) Fourth, the decontamination and safe zone construction strategy is dynamically adjusted according to the perceived distribution. Extensive experiments are conducted, and the results show that our SSPN achieves the highest accuracy among all algorithms proposed in the article. The RMSE and MARE between predicted and actual results are less than 0.5 and 0.8, and the fitted curve for SPNN is closest to the curve y = x, indicating the model's accuracy and robustness. Furthermore, the safe zone can be constructed within 25 s, and the rescue routes can be designed efficiently. We provide an innovative and efficient rescue reference for public indoor environments suffering from nerve agents.