Identification and quantitative analysis for atmospheric bioaerosols based on mixed fluorescence spectra

Using UV lidar to induce bioaerosol fluorescence is an effective method for detecting atmospheric bioaerosols. In this work, vertex component analysis, combined with the genetic algorithm, is proposed to analyze the received fluorescence spectra from UV LIF lidar. The pure components in the mixed fluorescence spectra are extracted using the vertex component analysis method, and then the genetic algorithm is adopted to fit the proportion of each component in the mixed spectra. The echo signals of the UV LIF lidar are simulated using the Voigt function at the excitation wavelength of 266 nm. All components can be correctly identified. Low-frequency noise is one of the significant interference factors when non-biological particles are present in the atmosphere. For three mixed components-bacillus subtilis, escherichia coli, and meadow oat pollen-the genetic algorithm can predict the proportion of each component with high accuracy, even in the presence of low-frequency noise. The stability of the algorithm was also explored with mixtures of four and five components, along with low-frequency noise. The residual predictive deviation (RPD) for all cases ranged from 2.0 to 28.2. It was concluded that the genetic algorithm can effectively predict the proportions in the mixed fluorescence spectra.