結合群眾外包的微型PM2.5感測器效能評估研究

Abstract

空氣品質逐漸為人們所重視，相較體積大且昂貴的專業儀器，更多人選擇購買便宜的微型感測器，然而也因為低成本的設計，在對抗環境因素干擾的能力可能不如專業儀器來得好，所以各廠商所製作的微型感測器準確度各有差異。工研院也為了因應日漸增加的感測器需求而研發第一款國產PM2.5感測器，目前已經過環保署的實地場域測試，但尚未有大量佈建的實測資料。本研究以群眾外包的方式將大量的微型感測器佈設於全國各地，以取得大量的實測資料，並對這些資料加以分析，分析內容包括：感測器良率、模組內變異性及變異係數。同時提供感測器使用者的使用經驗回饋，以提供未來感測器研發改善方向。採用群眾外包進行感測器效能評估能夠以數量的優勢同時收集不同地點的資料，達到縮短實驗時間的效果，更能有效減少實驗所需的場地成本。

Air quality is becoming more and more important. Compared to large and expensive professional instruments, more people choose to buy low-cost sensors. However, because of low-cost design, the ability to resist environmental interference may not be as good as professional instruments. ITRI also developed the first domestic PM2.5 sensor in response to the increasing demand for sensors. At present, it has been field evaluated by the Environmental Protection Administration, but there is not a lot of measured data in the field. In this study, a large number of low-cost sensors were deployed throughout the country in a crowdsourcing way to obtain a large amount of measured data in the field and analyze these data. The analysis includes: sensor yield, intra-model variability (IMV), and coefficient of variation (CV). At the same time, we provide feedback on the experience of sensor users to provide future directions for sensor development and improvement. Using crowdsourcing for sensor performance evaluation can collect data from different locations at the same time with the advantage of quantity, achieve the effect of shortening the experiment time, and reduce the site cost required for the experiment.