進階搜尋


   電子論文尚未授權公開,紙本請查館藏目錄
(※如查詢不到或館藏狀況顯示「閉架不公開」,表示該本論文不在書庫,無法取用。)
系統識別號 U0026-1907202009494400
論文名稱(中文) 發展時間序列深度學習與應變管制程序於空氣汙染防治研究
論文名稱(英文) Develop Time Series Deep Learning and Contingency Control Procedures in Air Pollution Prevention
校院名稱 成功大學
系所名稱(中) 工程管理碩士在職專班
系所名稱(英) Institute of Engineering Management (on the job class)
學年度 108
學期 2
出版年 109
研究生(中文) 黃豊哲
研究生(英文) Li-Che Huang
學號 N07051127
學位類別 碩士
語文別 中文
論文頁數 125頁
口試委員 指導教授-李家岩
口試委員-馬瀰嘉
口試委員-楊大和
口試委員-薛欣達
中文關鍵字 細懸浮微粒  深度學習  時間序列預測  空氣汙染防制策略 
英文關鍵字 PM2.5  Deep learning  Time series prediction  Air pollution control Strategy 
學科別分類
中文摘要 由於資訊傳輸能力越來越發達,人們隨時都能得到來自於世界各地的訊息。中國大陸的霧霾對於當地人民,甚至是周遭各國的影響越來越大,世界各國也逐漸地重視霧霾對於人體的危害。霧霾的主要組成成分為二氧化硫、氮氧化物和可吸入顆粒物,其中的「霾」便是我們常聽到的懸浮微粒(Particulate Matter),也是造成空氣汙染的主要原因之一,懸浮微粒除了會造成能見度下降外,長期暴露在這種環境下更會對人體的呼吸系統及心血管系統造成巨大的傷害,甚至導致死亡,其中危害最深的就屬細懸浮微粒(PM2.5),故研究並精準預測PM2.5將成為未來空氣汙染的重要課題之一。
本研究使用行政院環境保護署設於高雄地區的五個空氣品質測站(楠梓、前金、橋頭、仁武、左營)截取2015年1月1日至2019年6月30日每小時紀錄的監測資料,並從中挑選出細懸浮微粒、懸浮微粒、臭氧、一氧化氮、二氧化氮、氮氧化物、二氧化硫、一氧化碳、風速、風向、雨量、溫度及濕度等13項因素,利用這些資料分別進行ARIMA、LSTM、GRU等預測模型的建立。研究結果顯示,利用LSTM所建置出的模型,在未來的10天內,其預測值的MSE平均值為5.82,MAE的平均值為1.83,MAPE的平均值為8.95,是本研究三個模型中最適合用來預測PM2.5濃度的模型,利用這個模型搭配本研究建議的策略,能夠輕易的判斷出防治措施的啟用時機,以及加強處理的時機,對提升空氣品質有著極大的幫助。
英文摘要 As information transmission capabilities become more developed, people can get messages from all over the world at any time. The "smog" in China has an increasing impact on local people and even surrounding countries. Countries around the world have gradually paid more attention to the harm of smog to humans. The main components of smog are sulfur dioxide, nitrogen oxides, and inhalable particulate matter. The "smog" is the particulate matter we often hear, and it is also one of the main causes of air pollution. In addition to the decreased visibility, long-term exposure to this environment will cause great damage to the human respiratory system and cardiovascular system, and even cause death. The most harmful one is fine suspended particulates (PM2.5). Researching and accurately predicting PM2.5 will become one of the important topics of future air pollution.
This study used five air quality stations (Nanzi, Qianjin, Qiaotou, Renwu, Zuoying) of the Environmental Protection Department of the Executive Yuan to intercept every hour from January 1, 2015 to June 30, 2019. Record the monitoring data and select fine suspended particles, suspended particles, ozone, nitric oxide, nitrogen dioxide, nitrogen oxides, sulfur dioxide, carbon monoxide, wind speed, wind direction, rainfall, temperature and humidity from these 13 factors. These data were used to establish prediction models such as ARIMA, LSTM, and GRU. The results of the study show that the model built using LSTM has an average MSE of 5.82, MAE of 1.83, and MAPE of 8.95, which are the three models in this study. It is the most suitable model for predicting PM2.5 concentration. Using this model and the strategy proposed in this study can easily determine the timing of the prevention and control measures and the timing of enhanced treatment, which is of great help to improve air quality.
論文目次 中文摘要 I
Extended Abstract II
致謝 VII
目錄 VIII
表目錄 X
圖目錄 XII
第一章 緒論 1
1.1 研究背景與動機 1
1.2 研究目的 2
1.3 研究範圍與限制 3
1.4 研究流程與論文架構 3
第二章 文獻探討 5
2.1 懸浮微粒 5
2.2 時間序列 10
2.3 深度學習 14
2.4 空氣汙染防制策略 17
2.5 文獻探討小結 21
第三章 研究方法 22
3.1 資料前處理 24
3.1.1 資料篩選及刪除 25
3.1.2 遺漏值填補 25
3.1.2.1 鏈式方程插補法 26
3.2 特徵篩選 27
3.2.1 隨機森林 29
3.2.2 最小絕對壓縮挑選機制 30
3.2.3 多重共線性 33
3.2.3.1 多重共線性檢查 33
3.2.3.2 多重共線性處理方式 34
3.3 資料整合 36
3.3.1 標準化 37
3.4 建構預測模型 37
3.4.1 自迴歸整合移動平均模型 37
3.4.2 長短期記憶類神經網路 43
3.4.3 門控遞迴單元 46
第四章 實證研究 50
4.1 資料蒐集 50
4.2 資料前處理 51
4.2.1 遺漏值填補 52
4.3 特徵篩選 52
4.3.1 隨機森林 53
4.3.2 最小絕對壓縮挑選機制 54
4.3.3 投票法 55
4.3.4 多重共線性檢查 56
4.4 資料整合 56
4.5 建構預測模型 57
4.5.1 自回歸移動平均模型 57
4.5.2 長短期記憶類神經網路 63
4.5.3 GRU 68
4.5.4 小結 72
4.6 實證研究-總量管制計畫 73
4.6.1 條件及假設 73
4.6.2 模型擬合 75
4.6.3 應變程序 76
第五章 結論與未來研究 78
5.1 結論 78
5.2 未來研究 80
參考文獻 81
附錄 85
參考文獻 一、中文文獻
1.Lynn,機器學習的衰頹興盛:從類神經網路到淺層學習,Stockfeel,Available:
https://www.stockfeel.com.tw/%E6%A9%9F%E5%99%A8%E5%AD%B8%E7%BF%92%E7%9A%84%E8%A1%B0%E9%A0%B9%E8%88%88%E7%9B%9B%EF%BC%9A%E5%BE%9E%E9%A1%9E%E7%A5%9E%E7%B6%93%E7%B6%B2%E8%B7%AF%E5%88%B0%E6%B7%BA%E5%B1%A4%E5%AD%B8%E7%BF%92/ (Accessed by June. 3, 2020)
2.Jimmy Kao,速記AI課程-深度學習入門(一),Medium,Available:
https://medium.com/@baubibi/%E9%80%9F%E8%A8%98ai%E8%AA%B2%E7%A8%8B-%E6%B7%B1%E5%BA%A6%E5%AD%B8%E7%BF%92%E5%85%A5%E9%96%80-%E4%B8%80-68e27912ce30 (Accessed by June. 3, 2020)
3.行政院環境保護署,「空品不良季節之空污防制措施」簡報,2019。
4.邱瑞麟,影響石化原料國際行情的關鍵因素探討-以丁二烯為例,國立中山大學高階經營碩士班,碩士論文,2009。
5.李宗翰,運用ARIMA與向量自我廻歸模式探討新竹科學園區半導體產值預測,國立交通大學管理學院管理科學學程,碩士論文,2014。
6.周百建,資料科學與強化學習於石化原物料之價格預測與採購決策,國立成功大學製造資訊與系統研究所,碩士論文,2018。
7.孫維淋,植基於機器學習之分散式阻斷服務攻擊偵測,國立高雄科技大學電子工程系碩士班,碩士論文,2019。
8.高雄市經濟發展局,高雄市產經情勢分析季報107年度第4季,2019。
9.曾俊樺,資料探勘技術於全基因組關聯研究:以漢族雙極症疾患為例,國立成功大學製造資訊與系統研究所,碩士論文,2016。
10.黃婕,基於RNN-CNN集成深度學習模型的PM2.5小時濃度預測,浙江大學,碩士論文,2018。
11.經濟部加工出口區管理處,現有區內事業核准投資家數分類,Available:
https://www.epza.gov.tw/people/formdownload.aspx?pageid=eff51312b9bb0d30 (Accessed by June. 3, 2020)
12.楊宗翰,台灣實質匯率與所得不均關係之探討,國立高雄大學應用經濟學系,碩士論文,2016。
13.詹麗芬,時間序列方法應用於台灣汽車銷售量分析,中華科技大學企業管理系經營管理碩士班,碩士論文,2019。
14.鄭曼婷、邱嘉斌、楊宏隆、陳紀綸,沿海地區大氣懸浮微粒污染來源分析,第十五屆空氣污染控制技術研討會論文集,第733-740頁,1998。
15.鄭哲宇,應用統計及機器學習混合模型以預測台灣PM2.5濃度,輔仁大學管理學院統計資訊系應用統計碩士班,碩士論文,2019。
16.蔡鴻德、徐淑芷,從Lancet污染與健康報告看我國空氣污染防制策略,中華民國環境工程學會電子報,106年第4期,Available:https://www.cienve.org.tw/epaper/106_4/tech3.aspx(Accessed by June. 3, 2020)
17.黎彥彣,基於多方面的情感分析模型之共同處理方法,國立台灣科技大學電機工程系,碩士論文,2019。
18.謝明瑞、周信佑,台灣空污問題與防治政策,財團法人國家政策研究基金會,Available:
https://www.npf.org.tw/2/18414(Accessed by June. 3, 2020)
19.魏國彥,認識細懸浮微粒(PM2.5)科普手冊,行政院環境保護署,台北市,2015。
20.顏宇俊,LASSO一致性的經驗研究,國立東華大學應用數學研究所,碩士論文,2008。

二、英文文獻
1.Alasadi Suad A. and Bhaya Wesam S.. Review of Data Preprocessing Techniques in Data Mining. Journal of Engineering and Applied Sciences. 12: 4102-4107. 2017.
2.Breiman, L. Random forests. Machine learning. 45(1). 5-32. 2001.
3.Contreras, J., Espinola, R., Nogales, F. J., & Conejo, A. J. ARIMA models to predict next-day electricity prices. IEEE transactions on power systems, 18(3), 1014-1020. 2003.
4.Cao, J. J., Chow, J. C., Lee, F. S. C., Watson, J. G. Evolution of PM2.5 Measurements and Standards in the U.S. and Future Perspectives for China. Aerosol and Air Quality Research, 13: 1197–1211. 2013.
5.Cho Kyunghyun, Bart van Merrienboer, Caglar Gulcehre, Dzmitry Bahdanau, Fethi Bougares, Holger Schwenk, Yoshua Bengio. Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation. arXiv:1406.1078. 2014.
6.Chatterjee, S., and Hadi, A. S. Regression analysis by example. John Wiley & Sons. 2015.
7.Chen, Huey-Kuo, Yi-Hsiu Lin, and Chia-Yen Lee. Convex Nonparametric Least Squares and Stochastic Semi-Nonparametric Frontier to Estimate the Shadow Prices of PM2.5 and NOx for Taiwan’s Transportation Modes. Accepted in International Journal of Sustainable Transportation. 2020
8.Fernando H.J, Mammarella M.C., Grandoni G., Fedele P., Di Marco R., Dimitrova R., Hyde P.. Forecasting PM10 in metropolitan areas: Efficacy of neural networks. Environmental Pollution Volume 163. Pages 62-67. 2012.
9.Guo Runxia, Wang Jiaqi, Zhang Na , Dong Jiankang . State prediction for the actuators of civil aircraft based on a fusion framework of relevance vector machine and autoregressive integrated moving average.  Proceedings of the Institution of Mechanical Engineers Part I Journal of Systems and Control Engineering. 2018
10.Hochreiter S. and Schmidhuber J.. Long Short-Term Memory. Neural computation, vol. 9, no. 8, pp. 1735-1780. 1997.
11.Ho K.F., Lee S.C., Chan Chak K., Yu Jimmy C., Chow Judith C., Yao X.H.. Characterization of chemical species in PM2.5 and PM10 aerosols in Hong Kong. Atmospheric Environment 37. p31–39. 2003.
12.Holczer, B. Random Forest Classifier – Machine Learning, Available:https://www.globalsoftwaresupport.com/random-forest-classifier/ (Accessed by June. 3, 2020)
13.Huang Chiou-Jye and Kuo Ping-Huan. A Deep CNN-LSTM Model for Particulate Matter (PM2.5) Forecasting in Smart Cities. Sensors. Vol. 18 Issue 7. p2220. 2018.
14.Jeffrey R. Brook , Tom F. Dann & Richard T. Burnett. The Relationship Among TSP, PM10, PM2.5, and Inorganic Constituents of Atmospheric Participate Matter at Multiple Canadian Locations. Journal of the Air & Waste Management Association, 47:1, 2-19. 1997.
15.Lewis. C. D, Industrial and Business Forecasting Model. London Butterworth. 1982.
16.Luis A.Díaz-Robles, Juan C.Ortega, Joshua S.Fu, Gregory D.Reed, Judith C.Chow, John G.Watson, Juan A.Moncada-Herrera. A hybrid ARIMA and artificial neural networks model to forecast particulate matter in urban areas: The case of Temuco, Chile. Atmospheric Environment Volume 42, Issue 35. Pages 8331-8340. 2008.
17.Lee, Chia-Yen and Peng Zhou. Directional Shadow Price Estimation of CO2, SO2 and NOx in the United States Coal Power Industry 1990-2010. Energy Economics, 51, 493–502. 2015
18.Lee, Chia-Yen, and Ke Wang. Nash Marginal Abatement Cost Estimation of Air Pollutant Emissions Using the Stochastic Semi-Nonparametric Frontier. European Journal of Operational Research, 273 (1), 390-400. 2019
19.MacCallum, R. C., Widaman, K. F., Zhang, S., & Hong, S. Sample size infactor analysis. Psychological methods, 4(1), 84. 1999.
20.Melissa J. Azur, Elizabeth A. Stuart, Constantine Frangakis, Philip J. Leaf. Multiple imputation by chained equations: what is it and how does it work? International Journal of Methods in Psychiatric Research. 20(1): 40–49. 2011.
21.Martinelli, N., Olivieri, O., Girelli, D. Air Particulate matter and cardiovascular disease: a narrative review. European Journal of Internal Medicine 24, 295-302. 2013.
22.Murtaza Roondiwala and Harshal Patel and Shraddha Varma. Predicting stock prices using lstm. IEEE. 2015.
23.Mahajan, S., Liu, H. M., Tsai, T. C., & Chen, L. J. Improving the Accuracy and Efficiency of PM2.5 Forecast Service Using Cluster-Based Hybrid Neural Network Model. IEEE Access, 6, 19193-19204. 2018.
24.Rubin, Donald B. Multiple Imputation for Nonresponse in Surveys. New York: John Wiley & Sons. 1987.
25.Schafer J.L., Graham J.W. Missing data: our view of the state of the art. Psychological Methods. 7. 147–177. DOI: 10.1111/1467‐9574.00218. 2002.
26.Sharma, M. and Maloo, S. Assessment of Ambient Air PM10 and PM2.5 and Characterization of PM10 in the City of Kanpur, India. Atmospheric Environment. 39. 6015-6026. 2005.
27.Simeon Kostadinov, Understanding GRU Networks, Medium, Available:
https://towardsdatascience.com/understanding-gru-networks-2ef37df6c9be (Accessed by June. 3, 2020)
28.Tibshirani, R. Regression shrinkage and selection via the lasso. Journal of the Royal Statistical Society. Series B (Methodological). 267-288. 1996.
29.Tacq, J. J. & Tacq, J. Multivariate analysis techniques in social science research: From problem to analysis. Sage. 1997.
30.Trevor Hastie, Robert Tibshirani, Martin Wainwright. Statistical Learning with Sparsity:The Lasso and Generalizations. Monographs on Statistics and Applied Probability vol. 143.p11.2015
31.World Health Organization. Air quality guidelines for Particulate matter, ozone, nitrogen dioxide and sulfur dioxide. 278-279. 2005.
32.Zhang Chengqi, Zhu Xiaofeng, Zhang Jilian, Qin Yongsong, and Zhang Shichao. GBKII: An Imputation Method for Missing Values. Advances in Knowledge Discovery and Data Mining. PAKDD 2007. 1080-1087. 2007.
論文全文使用權限
  • 同意授權校內瀏覽/列印電子全文服務,於2020-07-27起公開。
  • 同意授權校外瀏覽/列印電子全文服務,於2022-07-08起公開。


  • 如您有疑問,請聯絡圖書館
    聯絡電話:(06)2757575#65773
    聯絡E-mail:etds@email.ncku.edu.tw