隨著數位化與科技化時代的來臨，現代農業對作物的質與量有更嚴格的標準，如何使作物的品質與產量更好是一項重要的議題。溫室因隔絕外部氣候干擾、可創造出適宜作物生長的條件而越來越受到重視。為維持重要的環境因子穩定的在適合作物生長的範圍內變化，一般溫室環控系統的做法是透過該環境因子之數值大於上界時抑制、小於下界時拉升的機制進行控制，但此一做法會造成每次抑制或拉升的控制動作出現時，環境因子之數值有一小段時間會落在適合作物生長的範圍之外，常此以往，對作物本身是一種傷害。因此，本篇論文提出一個針對重要環境因子的數值預測及相關裝置開關時機的決定方法，經由多變數迴歸的概念訓練出適合溫室內部不同環境狀態的數值預測模型。透過分析各種可能的數值變化情況，在每次環控系統收集到最新資料時，推斷當下應該對該環境因子進行抑制或拉升，並透過環控系統對相關裝置進行控制以達成調整之目的。透過本論文的模擬溫室實驗，能確定該環境因子的數值準確地被控制在在適合作物生長的範圍之內，不僅創造更穩定的栽培環境，也幫助使用者在不需更換精確度更高的環控裝置的前提下，就能使溫室環境控制的結果更為精確、更貼近使用者的需求。

With the coming of digital and technological age, there are more strict criteria for quality and quantity of crops in modern agriculture. It’s an important issue that how to improve quality and productivity of crops. Greenhouse is paid more attentions because it can isolate inner environment from outside climate and create a suitable growth condition for crops. To maintain value of an important environmental factor in certain range which is suitable for cultivated crops, there is a common greenhouse control mechanism: Environmental control system suppresses the environmental factor when value of the environmental factor exceeds predefined upper bound value, and it raise the environmental factor when value of the environmental factor lower than predefined lower bound value. But there this mechanism will make value of the environmental factor out of ideal range for a short period of time when these related switches are turned on or turned off. If this condition occurs frequently, it’s harmful to crops. Hence, this thesis proposes a method which predicts value of the environmental factor and decides turn-on/turn-off timing of related switches. Our system uses multiple regression analysis to generate value prediction models which can suit different environmental conditions in greenhouse. Our system analyzes all conditions that value of the environmental factor may vary, and deduces whether to suppress or raise the environmental factor when environmental control system receives the real-time data. Then our system make environmental control system to control related switches to conduct its control instruction. With the simulation greenhouse experiments, it can be sure that value of the environmental factor can be controlled within the ideal range which is suitable for crops development. Our system is not only helpful to create a more stable cultivation environment, but also make users can get a more precise control consequence without changing original devices.

[ALT76] Palo Alto, “A Space-Economical Suffix Tree Construction Algorithm,” Association for Computing Machinery, P262-272, 1976
[CHA04] 張碧芳, “溫度對作物的影響及熱休克蛋白質的功能,” 植物重要防疫檢疫病害診斷鑑定技術研習會專刊(三), P119-156, 2004
[CHA15] 張嘉滿, “仙履蘭花期調節技術之研究,” 臺南區農業改良場研究彙報第 65 號, P10-19, 2015
[DON15] Guozhu Dong, “Pattern-Aided Regression Modeling and Prediction Model Analysis,” IEEE Transactions on Knowledge and Data Engineering, P2452-2465, 2015
[JIA12] 蔣士棋, “應用工研院量測技術，林全茂提高蘭花生產良率,” 工業技術與資訊月刊8月號, P32-34, 2012
[KAN08] BeomJin Kang, “A Study on the Greenhouse Auto Control System based on Wireless Sensor Network,” International Conference on Security Technology, P41-44, 2008
[KHA10] F. Khazaei, “Study the Correlation, Regression and Path Coefficient Analysis in Sweet Corn under Different Levels of Plant Density and Nitrogen Rate,” Journal of Agricultural and Biological Science, P15-19, 2010
[LEW82] Lewis, C.D, “Industrial and business forecasting methods: A radical guide to exponential smoothing and curve fitting,” Butterworth Scientifie, P305-307, 1982
[LIN07] J. Lin, “Experiencing SAX: A novel symbolic representation of time series,” Data Mining Knowledge Discovery, P107–144, 2007
[SAL08] Raquel Salazar, “A Neural Network Model to Control Greenhouse Environment,” International Conference on Artificial Intelligence, P311-318, 2008
[TAT07] Giridhar Tatavarty, “Discovery of Temporal Dependencies between Frequent Patterns in Multivariate Time Series,” IEEE Symposium on Computational Intelligence and Data Mining, P688-696, 2007
[WAN07] Guojun Wang, “Polynomial Regression for Data Gathering in Environmental Monitoring Applications,” IEEE Conference on Global Telecommunications, P1307 – 1311, 2007
[YAN11] Yang Bin, “Modeling and Control for Greenhouse Temperature System Based on Hybrid Automata,” Proceedings of the 30th Chinese Control Conference, P1627-1631, 2011
[ZHU14] Dennis E.H. Zhuang, “Discovery of Temporal Associations in Multivariate Time Series,” IEEE Transactions on Knowledge and Data Engineering, P2969-2982, 2014