產品的多樣化造就了廢水對應的複雜性，加上排放標準日趨嚴格及自動連續監測管理辦法的實施，使廢水生物處理系統產出穩定良好的放流水水質受到了極大的考驗，因此有必要發展一套放流水水質異常早期預警機制。硝化-脫硝程序為處理含氮廢水的主要方式之一，維繫這套系統功能穩定的除了COD降解菌外，還有氨氧化菌、亞硝酸鹽氧化菌、脫硝菌等。本研究以支持向量機(Support vector machines)、多層感知器(Multilayer perceptron)、隨機森林(Random forests)等機器學習模式，結合水質參數、關鍵菌群密度等指標，建立早期預警系統。透過Weka平台，以2270筆廢水場操作紀錄及93筆實場採樣資料，並模擬HRT的概念做「數據平移」，分別訓練出能預測不同天數後出流水質的「歷史資料模型」及「實場採樣模型」。在歷史資料模型中表現最佳的為未經數據平移及特徵選取的隨機森林演算法，其準確率（Accuracy）達90.1%，F1 score為0.94亦是同一個模型最高。93天的實場採樣資料中，新增了11種特徵，這些特徵可以分為兩種類型：微生物指標及水質參數。以這93天資料所建立的「實場採樣模型」，在加入微生物參數後，比起僅加入新水質參數，無論是準確率、F1 score、Precision還是Recall，都有較佳的表現，因此在本研究中，微生物參數比新水質參數還要來的重要，也證實了加入微生物參數可以提升模型表現的假設。因此廢水系統的HRT為5~8天，所以預期模型在預測一個HRT後的出流水會有最佳的表現，歷史資料模型的結果顯示，各模型的預測能力在伯仲之間。實場採樣模型的結果則較符合預期，模型在預測8天後的反應槽狀況有最佳的表現，平均準確率達86.02%。本研究以機器學習模式來預測硝化-脫硝程序穩定度的能力，礙於資料量上的差異，實場採樣模型雖無法達到歷史資料模型最佳的準確率（90.1%），但證實了加入微生物參數可提升模型的表現。

Nitrification-denitrification process is one of the main ways to treat nitrogenous wastewater. Due to the complexity in the process control, its operation highly relies on professional experiences to ensure a stable system. With the implementation of strict discharge standards and under on-line surveillance of effluent quality, the stability of the nitrification-denitrification process has been greatly challenged. Therefore, it is essential to develop an early warning system for predicting the anomaly of effluent quality. The machine learning model has been proved a useful tool with its relatively high accuracy to deal with complicated systems. In this study, several supervised machine learning models, including support vector machines, multilayer perceptron, and random forest, were applied for predicting the stability of the nitrification-denitrification process. Among the models, the random forest possessed the highest performance in terms of accuracy (90.1%) and F1 score (0.94) for the historical data of a wastewater treatment plant. New influential parameters (total organic carbon (TOC), total nitrogen (TN), ammonia, nitrite, nitrate, conductivity) and microbial parameters (densities of ammonia-oxidizing archaea/bacteria (AOA, AOB), nitrite-oxidizing bacteria (NOB), denitrifiers) were included in a dataset of 93 sampling points to train the models. These data were processed by a “data shifting” method to simulate the hydraulic retention time of the reactor. The results revealed that microbial parameters played a more critical role than influential parameters to achieve a high prediction accuracy. The results also confirmed that the model performance could be improved by adding microbial parameters.

Abeling, U. and Seyfried, C. (1992) Anaerobic-aerobic treatment of high-strength ammonium wastewater-nitrogen removal via nitrite. Water Science and Technology 26(5-6), 1007-1015.
Anderson, J. (1964) The metabolism of hydroxylamine to nitrite by Nitrosomonas. Biochemical Journal 91(1), 8.
Anthonisen, A., Loehr, R., Prakasam, T. and Srinath, E. (1976) Inhibition of nitrification by ammonia and nitrous acid. Journal (Water Pollution Control Federation), 835-852.
Assefi, P., Ghaedi, M., Ansari, A., Habibi, M. and Momeni, M. (2014) Artificial neural network optimization for removal of hazardous dye Eosin Y from aqueous solution using Co2O3-NP-AC: Isotherm and kinetics study. Journal of Industrial and Engineering Chemistry 20(5), 2905-2913.
Awolusi, O.O., Nasr, M., Kumari, S. and Bux, F. (2016) Artificial intelligence for the evaluation of operational parameters influencing nitrification and nitrifiers in an activated sludge process. Microb Ecol 72(1), 49-63.
Bagheri, M., Mirbagheri, S., Ehteshami, M. and Bagheri, Z. (2015) Modeling of a sequencing batch reactor treating municipal wastewater using multi-layer perceptron and radial basis function artificial neural networks. Process Safety and Environmental Protection 93, 111-123.
Boser, B.E., Guyon, I.M. and Vapnik, V.N. (1992) A training algorithm for optimal margin classifiers, pp. 144-152, ACM.
Bothe, H., Jost, G., Schloter, M., Ward, B.B. and Witzel, K.-P. (2000) Molecular analysis of ammonia oxidation and denitrification in natural environments. FEMS microbiology reviews 24(5), 673-690.
Braker, G., Fesefeldt, A. and Witzel, K.-P. (1998) Development of PCR primer systems for amplification of nitrite reductase genes (nirK and nirS) to detect denitrifying bacteria in environmental samples. Applied and Environmental Microbiology 64(10), 3769-3775.
Breiman, L. (2001) Random forests. Machine Learning 45(1), 5-32.
Caranto, J.D. and Lancaster, K.M. (2017) Nitric oxide is an obligate bacterial nitrification intermediate produced by hydroxylamine oxidoreductase. Proceedings of the National Academy of Sciences 114(31), 8217-8222.
Cortes, C. and Vapnik, V. (1995) Support-vector networks. Machine Learning 20(3), 273-297.
Čuhel, J. and Šimek, M. (2011) Effect of pH on the denitrifying enzyme activity in pasture soils in relation to the intrinsic differences in denitrifier communities. Folia microbiologica 56(3), 230-235.
Daims, H., Lebedeva, E.V., Pjevac, P., Han, P., Herbold, C., Albertsen, M., Jehmlich, N., Palatinszky, M., Vierheilig, J., Bulaev, A., Kirkegaard, R.H., von Bergen, M., Rattei, T., Bendinger, B., Nielsen, P.H. and Wagner, M. (2015) Complete nitrification by Nitrospira bacteria. Nature 528(7583), 504-509.
De Boer, W. and Kowalchuk, G. (2001) Nitrification in acid soils: micro-organisms and mechanisms. Soil Biology and Biochemistry 33(7-8), 853-866.
Delnavaz, M., Ayati, B. and Ganjidoust, H. (2010) Prediction of moving bed biofilm reactor (MBBR) performance for the treatment of aniline using artificial neural networks (ANN). Journal of hazardous materials 179(1-3), 769-775.
Ducey, T.F., Shriner, A.D. and Hunt, P.G. (2011) Nitrification and denitrification gene abundances in swine wastewater anaerobic lagoons. Journal of Environment Quality 40(2), 610.
Ehrich, S., Behrens, D., Lebedeva, E., Ludwig, W. and Bock, E. (1995) A new obligately chemolithoautotrophic, nitrite-oxidizing bacterium, Nitrospira moscoviensis sp. nov. and its phylogenetic relationship. Archives of Microbiology 164(1), 16-23.
Elnekave, M., Celik, S.O., Tatlier, M. and Tufekci, N. (2012) Artificial neural network predictions of up-flow anaerobic sludge blanket (UASB) reactor performance in the treatment of citrus juice wastewater. Polish Journal of Environmental Studies 21(1).
Erguder, T.H., Boon, N., Wittebolle, L., Marzorati, M. and Verstraete, W. (2009) Environmental factors shaping the ecological niches of ammonia-oxidizing archaea. FEMS microbiology reviews 33(5), 855-869.
Ernfors, M., von Arnold, K., Stendahl, J., Olsson, M. and Klemedtsson, L. (2007) Nitrous oxide emissions from drained organic forest soils––an up-scaling based on C: N ratios. Biogeochemistry 84(2), 219-231.
Fan, M., Hu, J., Cao, R., Ruan, W. and Wei, X. (2018) A review on experimental design for pollutants removal in water treatment with the aid of artificial intelligence. Chemosphere 200, 330-343.
Feng, G., Sun, W., Zhang, F., Karthik, L. and Li, Z. (2016) Inhabitancy of active Nitrosopumilus-like ammonia-oxidizing archaea and Nitrospira nitrite-oxidizing bacteria in the sponge Theonella swinhoei. Sci Rep 6, 24966.
Ford, D.L., Churchwell, R.L. and Kachtick, J.W. (1980) Comprehensive analysis of nitrification of chemical processing wastewaters. Journal (Water Pollution Control Federation), 2726-2746.
Francis, C.A., Roberts, K.J., Beman, J.M., Santoro, A.E. and Oakley, B.B. (2005) Ubiquity and diversity of ammonia-oxidizing archaea in water columns and sediments of the ocean. Proceedings of the National Academy of Sciences of the United States of America 102(41), 14683-14688.
French, E., Kozlowski, J.A., Mukherjee, M., Bullerjahn, G. and Bollmann, A. (2012) Ecophysiological characterization of ammonia-oxidizing archaea and bacteria from freshwater. Appl. Environ. Microbiol. 78(16), 5773-5780.
Fulweiler, R.W., Emery, H.E., Heiss, E.M. and Berounsky, V.M. (2011) Assessing the role of pH in determining water column nitrification rates in a coastal system. Estuaries and Coasts 34(6), 1095.
Gernaey, K.V., van Loosdrecht, M.C., Henze, M., Lind, M. and Jørgensen, S.B. (2004) Activated sludge wastewater treatment plant modelling and simulation: state of the art. Environmental Modelling & Software 19(9), 763-783.
Goreau, T.J., Kaplan, W.A., Wofsy, S.C., McElroy, M.B., Valois, F.W. and Watson, S.W. (1980) Production of NO2-and N2O by nitrifying bacteria at reduced concentrations of oxygen. Applied and Environmental Microbiology 40(3), 526-532.
Grunditz, C. and Dalhammar, G. (2001) Development of nitrification inhibition assays using pure cultures of nitrosomonas and nitrobacter. Water Research 35(2), 433-440.
Hayatsu, M., Tago, K. and Saito, M. (2008) Various players in the nitrogen cycle: Diversity and functions of the microorganisms involved in nitrification and denitrification. Soil Science and Plant Nutrition 54(1), 33-45.
Hooper, A. (1978) Nitrogen oxidation and electron transport in ammonia-oxidizing bacteria. Microbiology, 299-304.
Hooper, A.B., Arciero, D., Bergmann, D. and Hendrich, M.P. (2004) Respiration in archaea and bacteria, pp. 121-147, Springer.
Huang, Z., Gedalanga, P.B. and Olson, B.H. (2010) Distribution of nitrobacter and nitrospira communities in an aerobic activated sludge bioreactor and their contributions to nitrite oxidation. Proceedings of the Water Environment Federation 2010(15), 2390-2403.
Hunt, P., Matheny, T. and Ro, K. (2007) Nitrous oxide accumulation in soils from riparian buffers of a coastal plain watershed–carbon/nitrogen ratio control. Journal of environmental quality 36(5), 1368-1376.
Kits, K.D., Sedlacek, C.J., Lebedeva, E.V., Han, P., Bulaev, A., Pjevac, P., Daebeler, A., Romano, S., Albertsen, M. and Stein, L.Y. (2017) Kinetic analysis of a complete nitrifier reveals an oligotrophic lifestyle. Nature 549(7671), 269.
Koike, I. and Hattori, A. (1975) Growth yield of a denitrifying bacterium, Pseudomonas denitrificans, under aerobic and denitrifying conditions. Journal of General Microbiology 88(1), 1-10.
Kowalchuk, G.A. and Stephen, J.R. (2001) Ammonia-oxidizing bacteria: a model for molecular microbial ecology. Annual Reviews in Microbiology 55(1), 485-529.
Krovvidy, S., Wee, W.G., Summers, R.S. and Coleman, J.J. (1991) An AI approach for wastewater treatment systems. Applied Intelligence 1(3), 247-261.
Kruse, M., Zumbrägel, S., Bakker, E., Spieck, E., Eggers, T. and Lipski, A. (2013) The nitrite-oxidizing community in activated sludge from a municipal wastewater treatment plant determined by fatty acid methyl ester-stable isotope probing. Systematic and applied microbiology 36(7), 517-524.
Leininger, S., Urich, T., Schloter, M., Schwark, L., Qi, J., Nicol, G.W., Prosser, J.I., Schuster, S. and Schleper, C. (2006) Archaea predominate among ammonia-oxidizing prokaryotes in soils. Nature 442(7104), 806.
Ludzack, F. and Ettinger, M. (1962) Controlling operation to minimize activated sludge effluent nitrogen. Journal (Water Pollution Control Federation), 920-931.
Luo, G., Xu, G., Tan, H., Gao, J. and Liu, W. (2016) Effect of dissolved oxygen on denitrification using polycaprolactone as both the organic carbon source and the biofilm carrier. International Biodeterioration & Biodegradation 110, 155-162.
Mahinthichaichan, P., Gennis, R.B. and Tajkhorshid, E. (2018) Bacterial denitrifying nitric oxide reductases and aerobic respiratory terminal oxidases use similar delivery pathways for their molecular substrates. Biochimica et Biophysica Acta (BBA)-Bioenergetics 1859(9), 712-724.
Manu, D. and Thalla, A.K. (2017) Artificial intelligence models for predicting the performance of biological wastewater treatment plant in the removal of Kjeldahl Nitrogen from wastewater. Applied Water Science 7(7), 3783-3791.
Mei, R., Kim, J., Wilson, F.P., Bocher, B.T. and Liu, W.-T. (2019) Coupling growth kinetics modeling with machine learning reveals microbial immigration impacts and identifies key environmental parameters in a biological wastewater treatment process. Microbiome 7(1), 65.
Mendoza-Castillo, D.I., Villalobos-Ortega, N., Bonilla-Petriciolet, A. and Tapia-Picazo, J.C. (2014) Neural network modeling of heavy metal sorption on lignocellulosic biomasses: effect of metallic ion properties and sorbent characteristics. Industrial & Engineering Chemistry Research 54(1), 443-453.
Nicol, G.W., Leininger, S., Schleper, C. and Prosser, J.I. (2008) The influence of soil pH on the diversity, abundance and transcriptional activity of ammonia oxidizing archaea and bacteria. Environmental Microbiology 10(11), 2966-2978.
Nielsen, P.H. (2017) Microbial biotechnology and circular economy in wastewater treatment. Microbial biotechnology 10(5), 1102-1105.
Nilsson, I., Ohlson, S., Häggström, L., Molin, N. and Mosbach, K. (1980) Denitrification of water using immobilized Pseudomonas denitrificans cells. European journal of applied microbiology and biotechnology 10(4), 261-274.
Nogueira, R., Melo, L.s.F., Purkhold, U., Wuertz, S. and Wagner, M. (2002) Nitrifying and heterotrophic population dynamics in biofilm reactors: effects of hydraulic retention time and the presence of organic carbon. Water Research 36(2), 469-481.
Oh, J. and Silverstein, J. (1999) Oxygen inhibition of activated sludge denitrification. Water Research 33(8), 1925-1937.
Okabe, S., Hiratia, K., Ozawa, Y. and Watanabe, Y. (1996) Spatial microbial distributions of nitrifiers and heterotrophs in mixed‐population biofilms. Biotechnology and bioengineering 50(1), 24-35.
Park, S., Bae, W., Chung, J. and Baek, S.-C. (2007) Empirical model of the pH dependence of the maximum specific nitrification rate. Process Biochemistry 42(12), 1671-1676.
Pauleta, S.R., Dell’Acqua, S. and Moura, I. (2013) Nitrous oxide reductase. Coordination Chemistry Reviews 257(2), 332-349.
Payne, W. (1973) Reduction of nitrogenous oxides by microorganisms. Bacteriological Reviews 37(4), 409.
Peng, Y., Chen, Y., Peng, C., Liu, M., Wang, S., Song, X. and Cui, Y. (2004) Nitrite accumulation by aeration controlled in sequencing batch reactors treating domestic wastewater. Water Science and Technology 50(10), 35-43.
Peng, Y. and Zhu, G. (2006) Biological nitrogen removal with nitrification and denitrification via nitrite pathway. Applied Microbiology and Biotechnology 73(1), 15-26.
Picioreanu, C., Van Loosdrecht, M. and Heijnen, J. (1997) Modelling the effect of oxygen concentration on nitrite accumulation in a biofilm airlift suspension reactor. Water Science and Technology 36(1), 147.
Randall, C. and Buth, D. (1984) Nitrite build-up in activated sludge resulting from temperature effects. Journal (Water Pollution Control Federation), 1039-1044.
Rotthauwe, J.-H., Witzel, K.-P. and Liesack, W. (1997) The ammonia monooxygenase structural gene amoA as a functional marker: molecular fine-scale analysis of natural ammonia-oxidizing populations. Applied and Environmental Microbiology 63(12), 4704-4712.
Sahu, J., Acharya, J. and Meikap, B. (2009) Response surface modeling and optimization of chromium (VI) removal from aqueous solution using Tamarind wood activated carbon in batch process. Journal of hazardous materials 172(2-3), 818-825.
Singh, K.P., Gupta, S., Singh, A.K. and Sinha, S. (2010) Experimental design and response surface modeling for optimization of Rhodamine B removal from water by magnetic nanocomposite. Chemical Engineering Journal 165(1), 151-160.
Smith, M.A. (2018) Structural, spectroscopic, and mechanistic insights into the three phases of nitrification metallobiochemistry, Cornell University.
Spang, A., Poehlein, A., Offre, P., Zumbrägel, S., Haider, S., Rychlik, N., Nowka, B., Schmeisser, C., Lebedeva, E.V. and Rattei, T. (2012) The genome of the ammonia‐oxidizing Candidatus Nitrososphaera gargensis: insights into metabolic versatility and environmental adaptations. Environmental Microbiology 14(12), 3122-3145.
Tay, J.-H. and Zhang, X. (1999) Neural fuzzy modeling of anaerobic biological wastewater treatment systems. Journal of Environmental Engineering 125(12), 1149-1159.
Tiedje, J., Sørensen, J. and Chang, Y.-Y. (1981) Assimilatory and dissimilatory nitrate reduction: perspectives and methodology for simultaneous measurement of several nitrogen cycle processes. Ecological Bulletins, 331-342.
Tiedje, J.M. (1982) Denitrification. Methods of soil analysis. Part 2. Chemical and microbiological properties (methodsofsoilan2), 1011-1026.
Tiedje, J.M. (1988) Ecology of denitrification and dissimilatory nitrate reduction to ammonium. Biology of anaerobic microorganisms 717, 179-244.
Turing, A.M. (1995) Lecture to the london mathematical society on 20 february 1947. MD COMPUTING 12, 390-390.
Vajrala, N., Martens-Habbena, W., Sayavedra-Soto, L.A., Schauer, A., Bottomley, P.J., Stahl, D.A. and Arp, D.J. (2013) Hydroxylamine as an intermediate in ammonia oxidation by globally abundant marine archaea. Proceedings of the National Academy of Sciences 110(3), 1006-1011.
Van Kessel, M.A., Speth, D.R., Albertsen, M., Nielsen, P.H., den Camp, H.J.O., Kartal, B., Jetten, M.S. and Lücker, S. (2015a) Complete nitrification by a single microorganism. Nature 528(7583), 555.
van Kessel, M.A., Speth, D.R., Albertsen, M., Nielsen, P.H., Op den Camp, H.J., Kartal, B., Jetten, M.S. and Lucker, S. (2015b) Complete nitrification by a single microorganism. Nature 528(7583), 555-559.
Vapnik, V. (2013) The nature of statistical learning theory, Springer science & business media.
Wang, X., Kvaal, K. and Ratnaweera, H. (2019) Explicit and interpretable nonlinear soft sensor models for influent surveillance at a full-scale wastewater treatment plant. Journal of Process Control 77, 1-6.
Winkler, M.K.H., Bassin, J.P., Kleerebezem, R., Sorokin, D.Y. and Van Loosdrecht, M.C.M. (2012) Unravelling the reasons for disproportion in the ratio of AOB and NOB in aerobic granular sludge. 94(6), 1657-1666.
Xi, X., Cui, Y., Wang, Z., Qian, J., Wang, J., Yang, L. and Zhao, S. (2011) Study of dead-end microfiltration features in sequencing batch reactor (SBR) by optimized neural networks. Desalination 272(1-3), 27-35.
Yao, Q. and Peng, D.-C. (2017) Nitrite oxidizing bacteria (NOB) dominating in nitrifying community in full-scale biological nutrient removal wastewater treatment plants. 7(1).
Yasmin, N.S.A., Wahab, N.A., Anuar, A.N. and Bob, M. (2019) Performance comparison of SVM and ANN for aerobic granular sludge. Bulletin of Electrical Engineering and Informatics 8(4), 1392–1401.
You, S.J., Hsu, C.L., Chuang, S.H. and Ouyang, C.F. (2003) Nitrification efficiency and nitrifying bacteria abundance in combined AS-RBC and A2O systems. Water Research 37(10), 2281-2290.
Zaghloul, M.S., Hamza, R.A., Iorhemen, O.T. and Tay, J.H. (2018) Performance prediction of an aerobic granular SBR using modular multilayer artificial neural networks. Sci Total Environ 645, 449-459.
Zumft, W.G. (1997) Cell biology and molecular basis of denitrification. Microbiol. Mol. Biol. Rev. 61(4), 533-616.
史慧萍 (2000) 好氧脫硝菌之篩選及其特性之研究, 國立臺灣大學, 台北市.
朱敬平, 鍾裕仁, 許國恩, 黃欣栩, 江家菱, 王郁萱, 許家銓, 黃育德, 胡雁翠, 烏春梅 and 吳碩傳 (2012) 事業廢水特性調查及污染管制措施研議計畫, 中華民國行政院環保署, 財團法人中興工程顧問社, Taiwan.
施曼, 张维国, 李江叶, 严少华 and 高岩 (2018) CO 2 浓度升高对水体硝化, 反硝化作用的影响研究进展. 应用生态学报 29(12), 4239-4247.
唐存宏 (2013) 工業廢水氨氮處理概述, 經濟部工業局.
張冠甫, 徐樹剛, 黃盟舜, 李茂松 and 張王冠 (2016) 氨氮廢水處理與回收技術及案例, 經濟部工業局.
許智堯 (2014) 類神經網路運用於工廠廢水處理之導電度預測, 元智大學, 桃園縣.
馮冠仁 (2006) 以支持向量機和特徵選取為基礎之階層式後設資料分類系統, 國立暨南國際大學, 南投縣.
黃子彥 (2018) 隨機森林演算法於UAV影像農作物分類之研究, 國立中興大學, 台中市.
黃家輝 (2006) 動態進流對A2O系統去除碳、氮、磷生物營養鹽影響之研究, 國立暨南國際大學, 南投縣.
蔡詩怡 (2011) 以探索性資料分析方法發展心臟血管疾病臨床輔助預知模型, 資訊管理研究所.
鄭羽展 (2018) 無氧-好氧活性污泥系統處理含氮工業廢水之穩定性與微生物體分析, 國立成功大學, 台南市.
謝忠和 (2012) 以機器學習技術改造傳統臨床決策輔具工具— 應用隨機森林、支援向量機器與類神經網路於急性闌尾炎診斷與乳癌風險評估, 國立陽明大學, 台北市.