| Summary: | 碩士 === 國立交通大學 === 環境工程系所 === 106 === In recent years, the public has increased their awareness on environmental protection, and above all, the matters of PM2.5 have been even more concerned. Although the Environmental Protection Agency of Taiwan has set up a number of air quality monitoring stations that conform to monitoring standards, they still fail to meet the expectations of the public; accordingly, the simple dust sensors for detecting particles are regularly used for instant monitoring devices but may potentially exist inaccurate monitoring data, this study mainly discusses its applicability with precision and accuracy.
This study researched the efficacy of the optical-detected principle of several simple dust sensors. The results showed that Plantower G-series has the highest degree of correlation with DustTrak DRX (R2 = 0.98) at ambient low concentration and its precision and low concentration performance of each piece are better than others; hence the research chooses Plantower PMS7003 (G7) as the follow-up accuracy and field testing.
The results of the comparison between the selected sensors and the EPA-approved FEM instruments show that the average accuracy of the sensors was 33.5±61.3 % and 60.1±99.9 % respectively. After correction, their average accuracy were 77.2±35.2 % and 78.1±22.3 %, respectively under the atmospheric PM2.5 (0-55 μg/m3) and PM10 (5-61 μg/m3). The G7 sensor was compare with BAM-1020 in the large testing chamber, and it is speculated that the detection of PM2.5 concentration from 55 μg/m3 to 200 μg/m3 resulted in the correction value turning point, showing overestimation of low concentration and underestimation of high concentration.
In the 9 months period of smart campus field tests, the parallel comparison between the TEOM(FEM) and the G7 sensors’ results results showed that re-calibration of low cost sensors is necessary after a certain period of usage. And without removal of abnormal values and re-calibration, the results show that the best monthly PM2.5 concentration accuracy was 54.9±67.3 % while the worst monthly average was -148.9±302.7 % in terms of accuracy. This indicates that sensors are feasible to be installed outdoors to enhance the resolution of the region, but abnormal values and sensor performance must be checked periodically. Insects or small animals, rainy weather and sensor sensitivity after long use may all result in the declining accuracy of the sensors.
|
|---|
