農業4.0是由工業4.0所延伸出來的農業模式，主要的訴求為應用前瞻的技術提升農業的生產力，包括自動控制、智慧生產、減少人力資源。主要利用有感測、物聯網以及大數據分析技術，從農業的溫室之中蒐集環境、植物資料並且解析，將資料轉換成可用的資訊。目前使用在設備控制的方法有HAVC控制器、PID控制器等等，這些控制器主要是適用於在要求環境能達到穩定的狀態，並不適用在動態環境之中。另外，我們觀察到在溫室之中，環境存在著慣性變動的特性。此特性會造成溫室內環境因子超過使用者所設立的規則邊界，也會讓設備有多餘的開啟時間，造成多餘的能源消耗。因此本論文提出一個分析溫室內部環境的方法，建立時序性的環境預測模組以及設備的效能分析模組，能夠藉由溫室的環境歷史資料預測溫室內部環境走向，並應用設備效能模組，預測設備的關閉時間，減少設備的開啟時間，而環境也能限縮於在使用者設立的規則邊界之中。本方法可使用在非固定的規則控制導向環境之中，並且我們建立了一個小型的溫室，內部由SCADA系統進行設備的自動控制來進行本篇論文的實作。藉由本實驗的結果，可了解本論文所提出的方法是可行且有效的。
透過此篇論文的研究，可以更了解多變數的時間序列模型以及動態模型的建立，未來更可以透過使用不同的多變數時間序列模型來改善預測準確度，能將誤差減至最低，使能源的使用量達到最小值。

Agriculture 4.0 is the new concept extended from Industry 4.0. Agriculture 4.0 aims to enhance agricultural productivity by automatic control, smart production solution and reducing usage of human resources. In order to reduce the energy consumption in automatic control system, a novel automatic device controlling module which employs environment tendency prediction and device efficiency analysis is proposed in this thesis. We collect the biological sensing data, meteorological data and geographic data from the greenhouse to construct time series environment forecasting model for environment tendency prediction. Moreover, the energy consumption data of devices are also collected to construct the device efficiency model as the reference model for the best control timing selecting. The environment time series forecasting model can predict the change of environmental factors. Further, it also can analyze the tendency of environment change. With the prediction result of environment change tendency, we can apply the device efficiency model to determine the time to turn off devices for eliminating the energy consumption.
The purpose of this thesis is that using environment prediction result to infer the environment tendency and further determining the time to turn off device through the device efficiency. In our experiment, a simulation greenhouse is employed as the testing target. Moreover, there is a commercial SCADA system is used as the automatic control system. The experiment result shows our approach not only innovation but also practicable.

[COA 15] Council of Agriculture, “PROductivity 4.0”, 2015
[MAT 13] Fernando Mateo, Juan Jose Carrasco, Abderrahim Sellami, Monica Millan-Giraldo, Manuel Dominguez, Emilio Soria-Olivas “Machine learning methods to forecast temperature in buildings”, 2013 Expert Systems with Applications Volume 40, Issue 4, March 2013, 1061-1068
[KUS 13] Andrew Kusiak, Yaohui Zeng, Guanglin Xu, ”Minimizing energy consumption of an air handling unit with a computational intelligence approach”, 2013 Energy and Buildings 60, 355-363
[ZHA 12] Hai-xiang Zhao, Frederic Magoules, “A review on the prediction of building energy consumption”, 2013 Renewable and Sustainable Energy Reviews 16, 3586-3592
[NAM 14] Abbas Namdar, Hamid Berenji, “A comparison of computational intelligence techniques for energy time series forecasting”, 2014 IEEE International Conference on Fuzzy Systems, 1648-1653
[DOU 09] A.I. Dounis, C. Caraiscos, “Advanced control systems engineering for energy and comfort management”, 2009 Renewable and Sustainable Energy Reviews 13, 1246–1261
[KUS 12] Andrew Kusiak, Guanglin Xu, “Modeling and optimization of HVAC systems using a dynamic neural network”, 2012 Energy 42 241-250
[BAB 12] Narendra Babu, B. Eswara Reddy, “Predictive Data Mining on Average Global Temperature using variants of ARIMA models” 2012 IEEE- International Conference On Advances in Engineering, Science and Management, 256-260
[LIN 12] Ondrej Linda, Dumidu Wijayasekara, Milos Manic, Craig Rieger, “Computational Intelligence based Anomaly Detection for Building Energy Management Systems”, 2012 5th International Symposium on Resilient Control Systems (ISRCS)
[GRU 09] J.K. Gruber, J.L. Guzm’an, F. Rodr’ıguez, C. Bordons, M. Berenguel, “Nonlinear Model Predictive Control of Greenhouse Temperature Using a Volterra Model”, 2009 Proceedings of the European Control Conference, 1299-1304
[XIE 09] Hang Xie, Hao Tang, Yu-He Liao, “Time Series Prediction Based On NARX Neural Networks: An Advanced Approach”, 2009 on Machine Learning and Cybernetics International Conference