智慧型手機應用做為空氣品質監測系統之未來潛力研究

Abstract

智慧型手機應用做為空氣品質監測系統之未來潛力研究 學生：孫偉碩 指導教授：白曛綾 國立交通大學工學院永續環境科技學程 摘 要 空氣品質監測Sensor包含影像，影像除與能見度有關外，能見度又與民眾之視覺感受有關，且能見度好壞又受PM2.5濃度高低影響，其中能見度高低又與死亡率有關，而智慧型手機即是一種Sensor，因此，本研究將探討以智慧型手機應用做為空氣品質監測系統之未來潛力。 本研究採用LPV-3能見度儀於嘉義所做之能見度監測實驗(僅統計能見度與影像之相關性)，及環保署監資處於金門-海景、陽明-山景、萬華-都市、崙背-鄉村及台東-乾淨背景(僅統計影像與PM2.5之相關性)等5站空品站所增設之Belfort 能見度儀，並統計其監測之（1）能見度與影像、（2）影像與PM2.5、（3）PM2.5與能見度之相關性，統計結果R2依序為：（1）嘉義- LPV-3: 能見度與影像為0.81、（2）金門-海景:依序為0.81、0.72及0.87、（3）陽明-山景:依序為0.83、0.27及0.5、（4）萬華-都市:依序為0.53、0.02及0.1、（5）崙背-鄉村:依序為0.92、0.71及0.67及（6）台東-乾淨背景: 影像與PM2.5為0.83。 由以上統計結果得知，（1）影像與能見度具一定之相關性（R2平均為0.8）、（2）影像與PM2.5之相關性與地表粗糙度及大氣混和均勻程度有關，粗糙度越小且大氣混和越均勻其相關性就越高及（3）能見度儀與拍攝儀需架設同一地點，其監測數據才具代表性。 本研究所探討使用智慧型手機上傳照片僅是影像來源之一，未來可評估延伸路口監視器、行車紀錄器等，任何有鏡頭的地方都可作為影像來源，以擴增巨量資料範圍，並利用大數據技術統計分析且分類其影像與能見度及PM2.5之級距，未來更能搭配APP軟體及穿載式裝置所提供環境監測功能，由雲端處理中心及影像大數據圖庫資料運算，連結地理資訊定位系統運算後，並由雲端中心提供慢跑者空氣品質最佳之路徑，並於跑完步後可將環境資訊及照片上傳並分享至社群網站，不僅可增加朋友之間的互動亦可擴增影像圖庫資料之收集，使雲端資料庫所回饋之訊息將越來越準確。 未來更期盼能藉由智慧型手機將空污監測系統，由監測何時會超標提升能預期何時會超標之功能，並由系統自動發送預測空品資訊，再進一步則由系統自動提供如何改善之建議方案，使智慧型手機應用於空氣污染監測系統變得更智慧。 關鍵字: 智慧型手機,影像 ,能見度, PM2.5 ,空氣品質監測系統

Feasibility study on application of smart phones as an air quality monitoring system Student：Weiso Sun Advisor：Hsunling Bai Degree program of Environmental Technology for Sustainability College of Engineering National Chiao Tung University

ABSTRACT This study aims to examine the potential of integrating a smart phone mobile software application “APP” in air quality monitoring system. Measurement data were obtained from LPV-3 visibility meter in the city of Chai-I as well as from the Environmental Protection Administration Executive Air Quality Monitoring Stations, which use the Belfort visibility meter across five different cities and settings as variables; these were Kinmen (sea), Yangming (mountain terrain), Wanhua Taipei (metropolitan), Qunlung (countryside), and Taitung (clean surrounding) to analyze the following correlations: Visibility vs. Digital Image; Image Index vs. PM2.5 and PM2.5 vs. Visibility. The resultant correlation coefficients (R2) by order of the analysis were: 1. Chai-I (LPV-3): R2= 0.81 (Visibility vs. Image Index only); 2. Kinmen (sea): R2= 0.81, 0.72, 0.87; 3. Yangming (mountain terrain): R2 = 0.83, 0.27, 0.5; 4. Wanhua Taipei (metropolitan): R2= 0.53, 0.02, 0.1; 5. Qunlung (countryside): R2 = 0.92, 0.71, 0.67; 6. Taitung (clean surrounding): R2= 0.83 (Image Index vs. PM2.5 only). From the above result, we conclude that there is significant casual relationship between Image Index vs. visibility, with R2 being 0.8 on average. The correlation between image index and PM2.5 is dependent on terrain roughness and atmospheric uniformity, particularly higher correlation in the setting of lower terrain roughness and a more stable atmospheric uniformity. The visibility meter and the image index system need to be set up in the same location in order to capture representative and comparable data. This concept can thus be applied to traffic cameras, dash camera in motor vehicles, or any other instruments with build-in camera, a visual sensor which has the function of digitizing images and uploading them to a virtual cloud processing center. Images are analyzed and categorized by image analysis index and global positioning system, factor in visibility measures and PM2.5. In our daily life, this information is particularly valuable for people such as joggers, enabling them to get a snapshot of air quality instantaneously using smart phone build-in camera, mobile analytical software application or other wearable environmental monitoring device, to work out the route with better air quality. This information can be shared in social networks amongst friends, and can also become a source of centralized data collection. We expect the accuracy of this air quality monitoring system will improve overtime with more quality data collected for the centralized database, and hopefully in the future, this information will provide useful feedback for air quality monitor.

Keywords: Smart phone, Image, Visibility, PM2.5 , Air quality monitoring system