近年來隨著物聯網與感測器技術的發展，水質自動連續監測系統(WQMS)的應用也更為普及，例如為了防止廠商惡意排放廢水而建置之放流水連續監測設備，為了預防魚蝦死亡而設置之魚塭水質即時監測系統。應用於戶外、開放水體之WQMS常因設置環境條件之變動而影響其數據可靠度，其中最為顯著的是溶氧(DO)感測器，許多案例顯示DO感測器探頭或薄膜極易被水中微生物與藻類附著，往往在清理校正後數小時或數日內即無法傳回正確數據，亦使DO感測器遠比其他常用水質感測器(水溫、電導度、pH)需要更高的清理與校正頻率。因此，本研究首先針對DO探頭附著造成的數據失效問題，利用DO感測器探頭附著水槽實驗，使用兩種機器學習模型改善數據品質以降低WQMS的維護頻率，第一種Temporal Fusion Transformer (TFT)用於判斷DO探頭失效時機及預測失效後的數據，第二種Random Forest (RF)用於修補長期DO數據的失效片段。其次，擬定使用兩種模型的時機與策略，制定一套Artificial Intelligence enabled dissolved oxygen sensor (AIDO)方法。最後將AIDO實際應用於戶外、開放水體之WQMS，並進一步分析討論此流程帶來的效益。
本研究結果顯示TFT能有效預測DO探頭失效後的DO趨勢，預測數據與有附著探頭相比，其平均絕對誤差(MAE)由2.21 mg/L降低為1.31 mg/L，R2則由0.81提升為0.85。若進一步以RF修補長期DO數據的失效片段，可將MAE與R2進一步提升為0.78 mg/L及0.96。本研究將AIDO應用於台南運河，從每次感測器清校開始即以TFT判斷DO探頭失效時機，失效後無需立刻清校，失效期1天內可以TFT補正即時數據，清校後再利用RF進行失效片段修復，依照此操作流程循環，最終應用結果顯示AIDO能準確判斷DO探頭失效時機，協助規劃派遣維護人員，並可大幅增加DO監測數據的穩定度與可靠度。

Recently, with the development of the Internet of Things (IOT) and sensor technology, the application of water quality monitoring system (WQMS) has become more and more popular. Dissolved oxygen (DO) is one of the important parameters in water quality monitoring. Many cases show that the DO sensors are susceptible to fouling and are prone to fouling from algal growth and sedimentation. In rapid biofouling conditions, DO sensors often return incorrect data within a few hours or days after cleaning and calibration, which means DO sensors need higher cleaning and calibration frequency than other commonly used water quality sensors (water temperature, conductivity, pH). In this study, two machine learning approaches were developed to improve the data failure problem caused by DO probe fouling. First, Temporal Fusion Transformer (TFT) was used to judge the failure timing of DO probes and predict DO data after probe failure. Second, Random Forest Regressor (RFR) was used to repair the failure segments of long-term DO data. The result shows that TFT can effectively predict the probe failure timing and the DO trend after probe failure. Using RFR can precisely repair failure segments of data. Furthermore, combining two machine learning models to develop a process for maintenance the WQMS. This WQMS maintenance process was practically applied to Tainan Canal WQMS. The result shows that this process can assist in planning the dispatchment of personnel, and greatly increase the stability and reliability of WQMS data.

Adu-Manu, K.S., Tapparello, C., Heinzelman, W., Katsriku, F.A. and Abdulai, J.-D. (2017) Water quality monitoring using wireless sensor networks: Current trends and future research directions. ACM Transactions on Sensor Networks (TOSN) 13(1), 1-41.
Arakeri, M.P., Kumar, B.V., Barsaiya, S. and Sairam, H. (2017) Computer vision based robotic weed control system for precision agriculture, pp. 1201-1205, IEEE.
Atique, U. and An, K.-G. (2019) Reservoir Water Quality Assessment Based on Chemical Parameters and the Chlorophyll Dynamics in Relation to Nutrient Regime. Polish Journal of Environmental Studies 28(3).
Babin, M., Roesler, C.S. and Cullen, J.J. (2008) Real-time coastal observing systems for marine ecosystem dynamics and harmful algal blooms: Theory, instrumentation and modelling, Unesco.
Chaerun Nisa, E. and Kuan, Y.-D. (2021) Comparative Assessment to Predict and Forecast Water-Cooled Chiller Power Consumption Using Machine Learning and Deep Learning Algorithms. Sustainability 13(2), 744.
Correll, D.L. (1998) The role of phosphorus in the eutrophication of receiving waters: A review. Journal of environmental quality 27(2), 261-266.
de Santana, F.B., de Souza, A.M. and Poppi, R.J. (2018) Visible and near infrared spectroscopy coupled to random forest to quantify some soil quality parameters. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 191, 454-462.
Delauney, L., Compere, C. and Lehaitre, M. (2010) Biofouling protection for marine environmental sensors. Ocean Science 6(2), 503-511.
Gubbi, J., Buyya, R., Marusic, S. and Palaniswami, M. (2013) Internet of Things (IoT): A vision, architectural elements, and future directions. Future generation computer systems 29(7), 1645-1660.
Hang, Y., Xiangtong, L., Longqin, X., Jingbin, L., Shuangyin, L., Liang, C., Dachun, F., Jianjun, G. and Liqiao, L. (2021) EMD-RF-LSTM: Combination Prediction Model of Dissolved Oxygen Concentration in Prawn Culture. 智慧农业.
Holmberg, M., Davide, F.A., Di Natale, C., D'Amico, A., Winquist, F. and Lundström, I. (1997) Drift counteraction in odour recognition applications: lifelong calibration method. Sensors and Actuators B: Chemical 42(3), 185-194.
HORIBA Advanced Techno, C., Ltd. (2020) Galvanic vs Optical Dissolved Oxygen Sensors.
Huan, J., Chen, B., Xu, X.G., Li, H., Li, M.B. and Zhang, H. (2021) River Dissolved Oxygen Prediction Based on Random Forest and LSTM. Applied Engineering in Agriculture 37(5), 901-910.
Huan, J., Li, H., Li, M. and Chen, B. (2020) Prediction of dissolved oxygen in aquaculture based on gradient boosting decision tree and long short-term memory network: A study of Chang Zhou fishery demonstration base, China. Computers and electronics in agriculture 175, 105530.
Hussain, C.M. and Kec̦ili, R. (2019) Modern environmental analysis techniques for pollutants, Elsevier.
Huynh, T.M., Nguyen, T.S.V., To, T.D., Nguyen, L.D., Le, H.M.T., Doan, T.C.D. and Dang, C.M. (2018) Electrochemical sensor chips for multiple measurements of dissolved oxygen, pH and oxidation-reduction potential in aquacultural farming. International Journal of Nanotechnology 15(11-12), 1024-1038.
Jesus, G., Casimiro, A. and Oliveira, A. (2021) Using Machine Learning for Dependable Outlier Detection in Environmental Monitoring Systems. ACM Transactions on Cyber-Physical Systems 5(3), 1-30.
Jewson, D. and Taylor, J. (1978) The influence of turbidity on net phytoplankton photosynthesis in some Irish lakes. Freshwater Biology 8(6), 573-584.
Kramer, D.L. (1987) Dissolved oxygen and fish behavior. Environmental biology of fishes 18(2), 81-92.
Kruse, P. (2018) Review on water quality sensors. Journal of Physics D: Applied Physics 51(20), 203002.
Lambrou, T.P., Anastasiou, C.C., Panayiotou, C.G. and Polycarpou, M.M. (2014) A low-cost sensor network for real-time monitoring and contamination detection in drinking water distribution systems. IEEE sensors journal 14(8), 2765-2772.
Li, D., Xu, X., Li, Z., Wang, T. and Wang, C. (2020) Detection methods of ammonia nitrogen in water: A review. TrAC Trends in Analytical Chemistry 127, 115890.
Li, W., Wu, H., Zhu, N., Jiang, Y., Tan, J. and Guo, Y. (2021) Prediction of dissolved oxygen in a fishery pond based on gated recurrent unit (GRU). Information Processing in Agriculture 8(1), 185-193.
Lim, B., Arik, S.O., Loeff, N. and Pfister, T. (2019) Temporal fusion transformers for interpretable multi-horizon time series forecasting. arXiv preprint arXiv:1912.09363.
Liu, Y., Zhang, Q., Song, L. and Chen, Y. (2019) Attention-based recurrent neural networks for accurate short-term and long-term dissolved oxygen prediction. Computers and electronics in agriculture 165, 104964.
Lowe, M., Qin, R. and Mao, X. (2022) A review on machine learning, artificial intelligence, and smart technology in water treatment and monitoring. Water 14(9), 1384.
Mitchell, T.O. (2006) Luminescence based measurement of dissolved oxygen in natural waters. Hach Company, Loveland, CO.
Morais, R., Matos, S.G., Fernandes, M.A., Valente, A.L., Soares, S.F., Ferreira, P. and Reis, M. (2008) Sun, wind and water flow as energy supply for small stationary data acquisition platforms. Computers and electronics in agriculture 64(2), 120-132.
Morrison, G., Fatoki, O., Persson, L. and Ekberg, A. (2001) Assessment of the impact of point source pollution from the Keiskammahoek Sewage Treatment Plant on the Keiskamma River-pH, electrical conductivity, oxygen-demanding substance (COD) and nutrients. Water Sa 27(4), 475-480.
Najafzadeh, M. and Ghaemi, A. (2019) Prediction of the five-day biochemical oxygen demand and chemical oxygen demand in natural streams using machine learning methods. Environmental monitoring and assessment 191(6), 1-21.
Niu, W.-J., Feng, Z.-K., Yang, W.-F. and Zhang, J. (2020) Short-term streamflow time series prediction model by machine learning tool based on data preprocessing technique and swarm intelligence algorithm. Hydrological Sciences Journal 65(15), 2590-2603.
Parra, L., Lloret, G., Lloret, J. and Rodilla, M. (2018) Physical sensors for precision aquaculture: A Review. IEEE sensors journal 18(10), 3915-3923.
Patel, K.K. and Patel, S.M. (2016) Internet of things-IOT: definition, characteristics, architecture, enabling technologies, application & future challenges. International journal of engineering science and computing 6(5).
Pobering, S. and Schwesinger, N. (2004) A novel hydropower harvesting device, pp. 480-485, IEEE.
Roberts, D., Rittschof, D., Holm, E. and Schmidt, A. (1991) Factors influencing initial larval settlement: temporal, spatial and surface molecular components. Journal of Experimental Marine Biology and Ecology 150(2), 203-221.
Samuelsson, O., Björk, A., Zambrano, J. and Carlsson, B. (2018) Fault signatures and bias progression in dissolved oxygen sensors. Water Science and Technology 78(5), 1034-1044.
Singh, K.P. and Gupta, S. (2012) Artificial intelligence based modeling for predicting the disinfection by-products in water. Chemometrics and Intelligent Laboratory Systems 114, 122-131.
Stankevič, V. and Šimkevičius, Č. (2000) Use of a shock tube in investigations of silicon micromachined piezoresistive pressure sensors. Sensors and Actuators A: Physical 86(1-2), 58-65.
Suquet, J., Godo-Pla, L., Valentí, M., Verdaguer, M., Martin, M.J., Poch, M. and Monclús, H. (2020) Development of an environmental decision support system for enhanced coagulation in drinking water production. Water 12(8), 2115.
Suslow, T.V. (2004) Oxidation-reduction potential (ORP) for water disinfection monitoring, control, and documentation.
Taylor, R. (1990) Interpretation of the correlation coefficient: a basic review. Journal of diagnostic medical sonography 6(1), 35-39.
Valera, M., Walter, R.K., Bailey, B.A. and Castillo, J.E. (2020) Machine learning based predictions of dissolved oxygen in a small coastal embayment. Journal of Marine Science and Engineering 8(12), 1007.
van Vliet, M.T., Franssen, W.H., Yearsley, J.R., Ludwig, F., Haddeland, I., Lettenmaier, D.P. and Kabat, P. (2013) Global river discharge and water temperature under climate change. Global Environmental Change 23(2), 450-464.
Wagner, R.J., Boulger Jr, R.W., Oblinger, C.J. and Smith, B.A. (2006) Guidelines and standard procedures for continuous water-quality monitors: station operation, record computation, and data reporting.
Wei, Y., Jiao, Y., An, D., Li, D., Li, W. and Wei, Q. (2019) Review of dissolved oxygen detection technology: From laboratory analysis to online intelligent detection. Sensors 19(18), 3995.
Yang, X.-e., Wu, X., Hao, H.-l. and He, Z.-l. (2008) Mechanisms and assessment of water eutrophication. Journal of zhejiang university Science B 9(3), 197-209.
Yu, Y., Si, X., Hu, C. and Zhang, J. (2019) A review of recurrent neural networks: LSTM cells and network architectures. Neural computation 31(7), 1235-1270.
工業技術研究院 (2021) 111年度高效化智慧水連網研發應用及專案管理計畫服務建議書.
台南市政府 (2019) 『運河水環境改善計畫』整體計畫工作計畫書