| Summary: | 碩士 === 國立高雄應用科技大學 === 電子工程系 === 106 === In this paper, the analysis of near-infrared signals in fine particulate matte pollution is proposed. The oscillation of different toxic gas dusts, we concern, can be obtained by general commercially available near-infrared light sensor. We utilize the digital signal processing methods (Fourier transform, Time-Frequency analysis, and multi-signal analysis) to analyze the main ingredient of them.
By converting the measured time-domain data into frequency-domain information based on Fourier transform, the characteristics of the dusts are able to obviously express. However, the Fourier transform fails to keep the time information describing when the polluted gas discovers. To overcome the shortage, we apply time-frequency analysis to avoid the disappearance of the time component. Meanwhile, the independent components analysis is employed to effectively separate mixing gas into simple compounds in order to seek polluted resources.
Finally, we do several experiments compared with data of Environmental Protection Agency to demonstrate our process is work, where the same frequency behaviors are observed in the gas dusts of different combustion media.
|
|---|
