The atmospheric aerosols of biological origin are called primary biological aerosol particles (PBAPs) or bioaerosols, encompassing a broad range of solid airborne particles derived from biological organisms. Pathogenic bioaerosols (e.g. viruses, bacteria, fungi) are one of the categories of PBAPs that can infect or spread toxins to humans. Monitoring these pathogens is required to protect public health and national security against disease vectors spread naturally or through nefarious intent (e.g., bioterrorism). To address the problem associated with the early detection of bioaerosol hazards, it is critical to analyze the physical characteristics of the aerosols involved, in relation to the atmospheric transport and dispersion processes of the biological aerosol hazard from source to receptor and the observation of this aerosol hazard at the sensor immersed in a natural and complex background aerosol.
A simulation environment for the bioaerosol using open-source computational fluid dynamics framework software OpenFOAM is applied to simulate the contaminant dispersion of the bioaerosol agent, which in conjunction with a stochastic time series model for the natural background aerosol interference, is used to generate the requisite datasets needed to test various algorithms vis-a-vis their performance for agent bioaerosol trigger detection on a network of sensors. Datasets for four different cases involving different intensity factors and modes are assigned to the background particle size distribution to train and evaluate the capability of the selected algorithms.
- Concentration distribution and time series at probe point
- PSD types and corresponding number concentration time series
The developed simulation environment is applied to complement the training and evaluating of the bioaerosol trigger detection system. Based on the state-of-the-art anomaly detection algorithms and sensor fusion techniques, the trigger system is applied to the complex temporal-spatial behaviour of the bioaerosol hazards to reveal the hidden anomaly patterns of target bioaerosols immersed in a complex natural bioaerosol background interference.
- ROC curves of the anomaly detection algorithms of the case III PSD
