學術產出-學位論文
文章檢視/開啟
書目匯出
-
題名 空氣品質感測網路的時間空間關聯模型
Spatial-Temporal Correlation Modeling of Air Monitoring Sensor Network作者 蔡政憲
Tsai, Zheng-Xian貢獻者 沈錳坤
Shan, Man-Kwan
蔡政憲
Tsai, Zheng-Xian關鍵詞 空氣品質估測
缺值
感測器網路
低成本裝置日期 2020 上傳時間 2-九月-2020 12:16:12 (UTC+8) 摘要 近年來，台灣的空氣汙染越來越嚴重，甚至已經開始影響到人的健康，因此針對空氣品質的監測和分析也就越來越重要。隨著無線感測網路技術的進步與發展，低成本微型感測器被採用並建構成大規模高密度的空氣品質監測網絡。但是低成本微型感測器在數據的穩定性上，容易產生大量的缺值。因此缺值問題對於大規模的低成本感測器網絡非常重要。本論文針對微型感測器的缺值問題，研究由歷史資料中學習感測器之間的時間及空間關係的關聯模型。進而運用關聯模型，由鄰居感測器來估測感測器的空氣品質，藉此填補目標感測器的缺值。此外，我們也提出改進的方法，以提升估測的效果。我們考量風力對於空汙擴散的影響，我們提出三種不同的分群策略，將PM2.5時間序列的資料分群，分別訓練關聯模型。實驗顯示我們所提出的關聯模型有顯著的估測效果。而我們所提出的分群策略有明顯的效果改進，平均絕對誤差(MAE)約3.2。
In recent years, air pollution has become more and more serious in Taiwan. It is important to monitor and analyze air quality. With the development of wireless sensing network technology, low-cost sensors have been adopted to build the large-scale high-density air quality monitoring network. However, low-cost air quality sensors are suffered from the missing value problem. Estimation of missing values for low cost air quality sensors is essential for air quality monitoring network.This thesis targets at the machine learning approaches for estimation of missing values of low cost sensors. We investigate the correlation model that discovers the spatial-temporal relationship among sensors from historical data. The correlation model is utilized to estimate the air quality of the target sensor by corresponding neighbor sensors. Moreover, we also propose approaches to improve the effectiveness of the estimation algorithm. We consider the impact of wind on the diffusion of air pollution, and propose three different clustering strategies to group the PM2.5 time series and train the correlation model for each group individually. Experiments show that the proposed correlation model performs well and the proposed clustering strategy leads to prominent performance improvement. The mean absolute error (MAE) is as low as 3.2.參考文獻 [1] C. K. Chou et al., Seasonal Variations and Spatial Distribution of Carbonaceous Aerosols in Taiwan, Atmospheric Chemistry & Physics Discussions, Vol. 10, 7079-7113, 2010.[2] L. J. Chen, Y. H. Ho, H. H. Hsieh, S. T. Huang, H. C. Lee, and S. Mahajan, ADF: an Anomaly Detection Framework for Large-scale PM2.5 Sensing Systems, IEEE IoT Journal, Vol. 5, Issue. 2, 559 - 570, 2018.[3] C.H. Hsu and F.Y. Cheng, Classification of Weather Patterns to Study the Influence of Meteorological Characteristics on PM2.5 Concentrations in Yunlin County, Taiwan, Atmospheric Environment, Vol. 144,397-408, 2016.[4] H. D. He, M. Li, W. L. Wang, Z. Y. Wang, Y. Xue, Prediction of PM2. 5 Concentration Based on the Similarity in Air Quality Monitoring Network, Building and Environment, Vol. 137, 11 - 17, 2018.[5] Health Effects Institute, State of Global Air 2018: A Special Report on Global Exposure to Air Pollution and Its Disease Burden, Health Effects Institute, MA[6] C. R. Lin and M. S. Chen, On the Optimal Clustering of Sequential Data, IEEE International Conference on Data Mining, 2002.[7] Y. Lin et al., Mining Public Datasets for Modeling Intra-City PM2.5 Concentrations at a Fine Spatial Resolution, ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems, 2017.[8] Y. Lin et al., Exploiting Spatiotemporal Patterns for Accurate Air Quality Forecasting using Deep Learning, ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems, 2018.[9] R. Manne, Analysis of Two Partial-Least-Squares Algorithms for Multivariate Calibration, Chemometrics Intelligent Laboratory Systems, Vol. 2, No. 1, 1987.[10] S. Mahajan, H. M. Liu, T. C. Tsai and L. J. Chen, Improving the Accuracy and Efficiency of PM2.5 Forecast Service Using Cluster-Based Hybrid Neural Network Model, IEEE Access, Vol. 6, 19193 - 19204, 2018.[11] R. Rosipal and N. Krämer, Overview and Recent Advances in Partial Least Squares, Lecture Notes in Computer Science, Vol 3940, 34-51, 2006.[12] P. W. Soh, J. W. Chang, J. W. Huang, Adaptive Deep Learning-Based Air Quality Prediction Model Using the Most Relevant Spatial-Temporal Relations, IEEE Access, Vol. 6, 38186 - 38199, 2018.[13] A. P. K. Tai et al., Meteorological Modes of Variability for Fine Particulate Matter (PM2.5) Air Quality in the United States: Implications for PM2.5 Sensitivity to Climate Change, AGU Fall Meeting Abstracts, 2011.[14] Y. T. Tsai, Y. R. Zeng, and Y. S. Chang, Air Pollution Forecasting using RNN with LSTM, IEEE 16th International Conference on Dependable, Autonomic and Secure Computing, 2018[15] D. M. Westervelt et al., Quantifying PM2.5-Meteorology Sensitivities in a Global Climate Model. Atmospheric Environment, Vol. 142, page 43-56, 2016.[16] Z. Wong and Z. Long, PM2.5 Prediction Based on Neural Network, International Conference on Intelligent Computation Technology and Automation (ICICTA), 2018.[17] J. Wang and S. Ogawa, Effects of Meteorological Conditions on PM2.5 Concentrations in Nagasaki, Japan. International journal of environmental research and public health, Vol. 12, No.8, 2015.[18] B. Yang and Q. Chen, PM2.5 Concentration Estimation Based on Image Quality Assessment, ACPR, 2017.[19] L. Yan, Y. Wu, L. Yan, and M. Zhou, Encoder-Decoder Model for Forecast of PM2.5 Concentration per Hour, International Cognitive Cities Conference, 2018.[20] H. Zhu and X. Lu, The Prediction of PM2.5 Value Based on ARMA and Improved BP Neural Network Model, International Conference on Intelligent Networking and Collaborative Systems, 2016.[21] M. A. Zaytar and C. E. Amrani, Sequence to Sequence Weather Forecasting with Long Short-Term Memory Recurrent Neural Networks, International Journal of Computer Applications, Vol. 143. No. 11, 2016.[22] Y. Zheng et al., A Cloud-Based Knowledge Discovery System for Monitoring Fine-Grained Air Quality, Microsoft Research Technical Report, 2014.[23] Y. Zhang et al., A Predictive Data Feature Exploration-Based Air Quality Prediction Approach. IEEE Access, Vol 7, 2019. 描述 碩士
國立政治大學
資訊科學系
107753034資料來源 http://thesis.lib.nccu.edu.tw/record/#G0107753034 資料類型 thesis dc.contributor.advisor 沈錳坤 zh_TW dc.contributor.advisor Shan, Man-Kwan en_US dc.contributor.author (作者) 蔡政憲 zh_TW dc.contributor.author (作者) Tsai, Zheng-Xian en_US dc.creator (作者) 蔡政憲 zh_TW dc.creator (作者) Tsai, Zheng-Xian en_US dc.date (日期) 2020 en_US dc.date.accessioned 2-九月-2020 12:16:12 (UTC+8) - dc.date.available 2-九月-2020 12:16:12 (UTC+8) - dc.date.issued (上傳時間) 2-九月-2020 12:16:12 (UTC+8) - dc.identifier (其他 識別碼) G0107753034 en_US dc.identifier.uri (URI) http://nccur.lib.nccu.edu.tw/handle/140.119/131635 - dc.description (描述) 碩士 zh_TW dc.description (描述) 國立政治大學 zh_TW dc.description (描述) 資訊科學系 zh_TW dc.description (描述) 107753034 zh_TW dc.description.abstract (摘要) 近年來，台灣的空氣汙染越來越嚴重，甚至已經開始影響到人的健康，因此針對空氣品質的監測和分析也就越來越重要。隨著無線感測網路技術的進步與發展，低成本微型感測器被採用並建構成大規模高密度的空氣品質監測網絡。但是低成本微型感測器在數據的穩定性上，容易產生大量的缺值。因此缺值問題對於大規模的低成本感測器網絡非常重要。本論文針對微型感測器的缺值問題，研究由歷史資料中學習感測器之間的時間及空間關係的關聯模型。進而運用關聯模型，由鄰居感測器來估測感測器的空氣品質，藉此填補目標感測器的缺值。此外，我們也提出改進的方法，以提升估測的效果。我們考量風力對於空汙擴散的影響，我們提出三種不同的分群策略，將PM2.5時間序列的資料分群，分別訓練關聯模型。實驗顯示我們所提出的關聯模型有顯著的估測效果。而我們所提出的分群策略有明顯的效果改進，平均絕對誤差(MAE)約3.2。 zh_TW dc.description.abstract (摘要) In recent years, air pollution has become more and more serious in Taiwan. It is important to monitor and analyze air quality. With the development of wireless sensing network technology, low-cost sensors have been adopted to build the large-scale high-density air quality monitoring network. However, low-cost air quality sensors are suffered from the missing value problem. Estimation of missing values for low cost air quality sensors is essential for air quality monitoring network.This thesis targets at the machine learning approaches for estimation of missing values of low cost sensors. We investigate the correlation model that discovers the spatial-temporal relationship among sensors from historical data. The correlation model is utilized to estimate the air quality of the target sensor by corresponding neighbor sensors. Moreover, we also propose approaches to improve the effectiveness of the estimation algorithm. We consider the impact of wind on the diffusion of air pollution, and propose three different clustering strategies to group the PM2.5 time series and train the correlation model for each group individually. Experiments show that the proposed correlation model performs well and the proposed clustering strategy leads to prominent performance improvement. The mean absolute error (MAE) is as low as 3.2. en_US dc.description.tableofcontents 致謝 i摘要 iiAbstract iii表目錄 vi圖目錄 vii第一章 緒論 11-1. 研究背景 11-2. 研究動機與目的 3第二章 相關研究 52-1 Prediction 52-2 Anomaly Detection 62-3 Estimation 7第三章 研究方法 93-1 Neighbor Identification 103-2 Correlation Model 143-2-1 Partial Least Square 153-2-2 Partial Least Square Regression 163-2-3 Air Concentration Estimation 173-3 Segment Model 183-3-1 Trend-Based Approach 193-3-2 Wind-Based Approach 213-3-3 Season-Based Approach 23第四章 實驗設計 264-1. 實驗說明 264-2. 資料集介紹 264-3. 評估方式 284-4. Baseline方法 284-5. 實驗結果：Plain Model 304-6. 實驗結果：Segment Model 364-6-1 Trend-Based Approach 374-6-2 Wind-Based Approach 404-6-3 Season-Based Approach 414-6-4 方法比較 46第五章 結論 47參考文獻 49 zh_TW dc.format.extent 3828449 bytes - dc.format.mimetype application/pdf - dc.source.uri (資料來源) http://thesis.lib.nccu.edu.tw/record/#G0107753034 en_US dc.subject (關鍵詞) 空氣品質估測 zh_TW dc.subject (關鍵詞) 缺值 zh_TW dc.subject (關鍵詞) 感測器網路 zh_TW dc.subject (關鍵詞) 低成本裝置 zh_TW dc.title (題名) 空氣品質感測網路的時間空間關聯模型 zh_TW dc.title (題名) Spatial-Temporal Correlation Modeling of Air Monitoring Sensor Network en_US dc.type (資料類型) thesis en_US dc.relation.reference (參考文獻) [1] C. K. Chou et al., Seasonal Variations and Spatial Distribution of Carbonaceous Aerosols in Taiwan, Atmospheric Chemistry & Physics Discussions, Vol. 10, 7079-7113, 2010.[2] L. J. Chen, Y. H. Ho, H. H. Hsieh, S. T. Huang, H. C. Lee, and S. Mahajan, ADF: an Anomaly Detection Framework for Large-scale PM2.5 Sensing Systems, IEEE IoT Journal, Vol. 5, Issue. 2, 559 - 570, 2018.[3] C.H. Hsu and F.Y. Cheng, Classification of Weather Patterns to Study the Influence of Meteorological Characteristics on PM2.5 Concentrations in Yunlin County, Taiwan, Atmospheric Environment, Vol. 144,397-408, 2016.[4] H. D. He, M. Li, W. L. Wang, Z. Y. Wang, Y. Xue, Prediction of PM2. 5 Concentration Based on the Similarity in Air Quality Monitoring Network, Building and Environment, Vol. 137, 11 - 17, 2018.[5] Health Effects Institute, State of Global Air 2018: A Special Report on Global Exposure to Air Pollution and Its Disease Burden, Health Effects Institute, MA[6] C. R. Lin and M. S. Chen, On the Optimal Clustering of Sequential Data, IEEE International Conference on Data Mining, 2002.[7] Y. Lin et al., Mining Public Datasets for Modeling Intra-City PM2.5 Concentrations at a Fine Spatial Resolution, ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems, 2017.[8] Y. Lin et al., Exploiting Spatiotemporal Patterns for Accurate Air Quality Forecasting using Deep Learning, ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems, 2018.[9] R. Manne, Analysis of Two Partial-Least-Squares Algorithms for Multivariate Calibration, Chemometrics Intelligent Laboratory Systems, Vol. 2, No. 1, 1987.[10] S. Mahajan, H. M. Liu, T. C. Tsai and L. J. Chen, Improving the Accuracy and Efficiency of PM2.5 Forecast Service Using Cluster-Based Hybrid Neural Network Model, IEEE Access, Vol. 6, 19193 - 19204, 2018.[11] R. Rosipal and N. Krämer, Overview and Recent Advances in Partial Least Squares, Lecture Notes in Computer Science, Vol 3940, 34-51, 2006.[12] P. W. Soh, J. W. Chang, J. W. Huang, Adaptive Deep Learning-Based Air Quality Prediction Model Using the Most Relevant Spatial-Temporal Relations, IEEE Access, Vol. 6, 38186 - 38199, 2018.[13] A. P. K. Tai et al., Meteorological Modes of Variability for Fine Particulate Matter (PM2.5) Air Quality in the United States: Implications for PM2.5 Sensitivity to Climate Change, AGU Fall Meeting Abstracts, 2011.[14] Y. T. Tsai, Y. R. Zeng, and Y. S. Chang, Air Pollution Forecasting using RNN with LSTM, IEEE 16th International Conference on Dependable, Autonomic and Secure Computing, 2018[15] D. M. Westervelt et al., Quantifying PM2.5-Meteorology Sensitivities in a Global Climate Model. Atmospheric Environment, Vol. 142, page 43-56, 2016.[16] Z. Wong and Z. Long, PM2.5 Prediction Based on Neural Network, International Conference on Intelligent Computation Technology and Automation (ICICTA), 2018.[17] J. Wang and S. Ogawa, Effects of Meteorological Conditions on PM2.5 Concentrations in Nagasaki, Japan. International journal of environmental research and public health, Vol. 12, No.8, 2015.[18] B. Yang and Q. Chen, PM2.5 Concentration Estimation Based on Image Quality Assessment, ACPR, 2017.[19] L. Yan, Y. Wu, L. Yan, and M. Zhou, Encoder-Decoder Model for Forecast of PM2.5 Concentration per Hour, International Cognitive Cities Conference, 2018.[20] H. Zhu and X. Lu, The Prediction of PM2.5 Value Based on ARMA and Improved BP Neural Network Model, International Conference on Intelligent Networking and Collaborative Systems, 2016.[21] M. A. Zaytar and C. E. Amrani, Sequence to Sequence Weather Forecasting with Long Short-Term Memory Recurrent Neural Networks, International Journal of Computer Applications, Vol. 143. No. 11, 2016.[22] Y. Zheng et al., A Cloud-Based Knowledge Discovery System for Monitoring Fine-Grained Air Quality, Microsoft Research Technical Report, 2014.[23] Y. Zhang et al., A Predictive Data Feature Exploration-Based Air Quality Prediction Approach. IEEE Access, Vol 7, 2019. zh_TW dc.identifier.doi (DOI) 10.6814/NCCU202001442 en_US
