Bioaerosol multi-pollutant source localization in urban microclimates using an improved particle swarm optimization method with restricted search capabilities

In recent years, incidents of chemical and biological pollution in urban environments have occurred sporadically. Urban environments, being vast and intricately populated with diverse buildings and infrastructure, present exceptionally complex air flow fields. Additionally, the covert nature of urban pollution sources combined with rapid pollutant dispersion poses significant challenges for rapid localization. Traditional trajectory inversion methods are limited by their reliance on prior modeling and extensive sensor deployment, but rapid source identification is vital for public health. To address these limitations, this study introduces the LAN-DPSO method (Limited-area Niche Particle Swarm Optimization for Dynamic Systems), a niche-based source localization algorithm. When deployed in dispersed configurations, LAN-DPSO achieved a success rate exceeding 95% in our experiments. Its advantage lies in pre-locating the approximate positions of potential pollution sources within urban areas and accurately pinpointing multiple sources even under unstable airflow conditions. This study establishes urban building complex models using two typical residential areas. Through Computational fluid dynamics (CFD) simulations and wind tunnel experiments, airflow fields and bioaerosol diffusion are simulated and experimentally validated, respectively. The LAN-DPSO is then systematically evaluated using the simulated data and compared with another high-performing method. Results indicate its ability to rapidly and accurately localize multiple pollution sources in urban microclimates. Parameter optimization for varying environmental conditions is critical to LAN-DPSO's performance. LAN-DPSO also exhibits notable cost-efficiency in source tracing, suggesting its potential for practical application in urban environments.