| 摘要: | 過去傳統的環境監測與數據分析，大多利用昂貴的大型設備或人工採樣後，進到標準實驗室進行儀器分析，在耗時且耗人力的前提下，又無法及時掌握污染特徵進行溯源。故自21世紀開始，部分國家、政府及研究單位利用物聯網及低成本感測設備進行環境污染監測，比起過往已可發揮24小時連續監測及提高數據在時間及空間的解析度，但也因開發初期對感測設備品質掌握度不一，同步衍生布建成本無法掌握及疏於維運，雖有大量數據卻無法妥善應用分析之窘境。
有鑑於此，本研究參考「毒災應變決策支援模式(TDSS)」及國外「低成本感測器布建及應用案例」，以台灣環境污染事件應變情境為主軸下，研發一套完整「水質感測物聯網布建及數據應用規範」，從感測設備的品保品管及雲端校正模式，配合布建場域優化選址及低成本維運，設計水質自動連續監測預警分析模式，設定預警規則並自動判別環境異常或儀器異常。透過「疑似污染源頭分析模組」，結合動靜態跨域資訊迅速溯源至上游，以物聯網技術發出通知警訊並啟動稽查程序；建立「污染熱區分析模組」評估未來數小時內污染物的影響範圍，即時展開應變行動，達到快速預警水污染防治之成效。
本研究以桃園市作為實際驗證場域，布建100台國產化小型水質監測設備，驗證「水質感測物聯網快速反應支援系統」之整體解決方案(Total Solution)。結果顯示導入酸鹼值、導電度與溫度三合一感測晶片，其量測範圍與精度/誤差可控制在pH 3 – pH 12 ± pH 0.5 (R2=0.978)、90 - 10500μS /cm±13.5％ (R2=0.990)，以及10 - 85℃ ± 0.5℃ (R2=0.990)。COD+SS感測模組之COD與SS量測範圍分別可達0.1-300 mg/L與0.1-100 mg/L，精密度與準確度可達10%與40%。在自然水體、工業區排水及農渠灌溉水路可透過上述物聯網布建達到24小時污染預警(時間序列分析)及污染溯源(地理空間分析)成效，以實際案例驗證水質感測物聯網快速反應支援系統之有效性，成功發現過往無法察覺之排放源及污染事件。
;Previously, environmental monitoring and related data analysis had been carried out via deployment of large-scale equipment or manual sampling, as well as analysis with equipment at standard laboratory. It was a time- and manpower-consuming process incapable of timely grasping of polluting characteristics for source tracing. Therefore, since the turn of the 21st century, government agencies and research institutions in some countries have started to employ IoT (Internet of Things) and low-cost sensing devices in monitoring environmental pollution around the clock with better time and data resolution. Initially, the approach was ridden with multiple problems such as unstable sensing-device quality, high deployment cost, and inadequate maintenance causing wasted massive collected data which were not analyzed and utilized properly.
To rectify the flaws, this study puts forth a complete "norms on deployment of water-quality sensing IoT and data application" in reference to "toxic accident decision support system (TDSS) and foreign cases for the deployment and applications of low-cost sensors, based on the scenario of polluting incidents and responses in Taiwan. The norms cover sensing devices′ quality assurance and control, along with a cloud-end calibration model, optimized deployment siting, low-cost maintenance, continuing water-quality monitoring and alerting model, and early-warning rules and automatic determination for environmental or device abnormality. The norms set alerting rules and call for automatic determination of environmental or device abnormality, as well as rapid polluting-source tracing, via utilization of "suspicious polluting source analytical model" and cross-region data, before sending alarm via IoT and triggering auditing procedure. Another feature is "polluting hot area analytical model," capable of forecasting pollution-affected area in the next several hours, as basis for emergency response and prevention of pollution spread.
This study selected Taoyuan City as the verification field, where 100 small water-quality monitoring devices were deployed, for testing the effectiveness of the total solution of the "Water quality sensing IoT rapid response support system." The results show that the measurement range and accuracy/error of the pH, conductivity and temperature sensor chip can be controlled within pH3–pH12 ± pH0.5 (R2=0.978), 90 - 10500μS/cm±13.5% (R2=0.990), and 10 - 85℃ ± 0.5℃ (R2=0.990). The COD and SS measurement ranges of the COD+SS sensor module can reach 0.1-300 mg/L and 0.1-100 mg/L respectively, with precision and accuracy of 10% and 40%. The IoT-based network is capable of pollution alerting (time-sequence analysis) and source tracing (geospatial analysis) within 24 hours for water body, industrial-area drainage channel, and agricultural irrigation channel. The network succeeded in detecting three polluting cases and tracing their sources timely, outperforming previous systems. |
