在農業方面，許多專業決定都依靠農民的經驗，且這些關鍵的判斷往往難以用簡單的數值去量化。但是在農業的進步當中，人力已漸漸被機器所取代，農民從密集勞動者轉變成為指令決策者，從而擴大種植規模或是提升品質。農業物聯網系統成為新農業的趨勢，資料監控與自動化系統幫助了許多農民，除此之外，機器學習也被廣泛的運用在農業領域，最為常見的就是灌溉。如何有效利用農業物聯網系統與機器學習來改善傳統農業即是本篇論文的主要目的。
本篇論文提供了基於農業物聯網系統的集成學習灌溉模型。農業物聯網系統提供了完善的資料收集，資料監控的功能，並提供了網頁及手機應用程式以利農民使用。在機器學習的方式當中，我們選擇集成學習。集成學習灌溉模型藉由環境數據、天氣數據以及過往的灌溉控制紀錄來學習農民的經驗，最後將模型與物聯網系統整合。考量到農業物聯網系統收集過往資料的時間長短，我們建置了單棚模型與多棚模型。單棚模型僅為特定的溫室服務，而多棚模型則為農場當中其餘溫室服務。
我們也在農場進行模型的實驗，單棚模型在一次栽種週期當中普遍的均方誤差為10秒左右，而多棚模型在一次栽種週期當中普遍的均方誤差為15秒左右。此結果顯示我們在精準灌溉上已達到了極高的標準，不僅能優化農民灌溉的方式，也能減少農民的人力需求。

In agriculture, many decisions rely on the experiences of farmers, and these decisions are often difficult to quantify with simple numerical values. However, manpower has gradually been replaced by machines in the progress of agriculture. The farmers change from the laborers to decision makers for expanding planting scale or improving quality. The agricultural Internet of Things (IoT) system has become the trend of new agriculture. The data monitoring and automation control system have helped many farmers. In addition, machine learning is widely used in the agriculture. The irrigation is the most common. The main purpose of this thesis is to effectively use agricultural IoT systems and machine learning to improve traditional agriculture.
This thesis provides an ensemble learning irrigation model based on the agricultural IoT system. The IoT system provides the data collection and data monitoring functions, and also supports the interfaces for the convenience of farmers. The ensemble learning irrigation models are trained via environmental data, weather data and past irrigation control records from agricultural IoT system. At last, the models are embedded in the agricultural IoT system. Considering the different data collection periods, we build single-shed models and multi-sheds models. The former serves only specific greenhouse, while the latter serves the remaining greenhouses on the farm.
We experiment the models on the farm. The general mean square error of models during a planting cycle is about 10 to 15 seconds. The result shows that we achieve the goals in precision irrigation, which can not only optimize the irrigation methods, but also reduce the manpower requirements.

Chapter 1 Introduction 1
Chapter 2 Related Works 4
2.1 LoRa, LoRaWAN, and LoRa P2P 4
2.1.1 LoRa 5
2.1.2 LoRaWAN 5
2.1.3 LoRa P2P 7
2.2 Agricultural IoT System 8
2.3 Ensemble learning 9
2.3.1 Stacked Generalization (Stacking) 10
2.3.2 Boosting 11
Chapter 3 System Design and Implementation 13
3.1 System Architecture 13
3.1.1 Master and Slave 14
3.1.2 Control Center 16
3.1.3 API Server 16
3.1.4 Forecast System Design for Small-Scale Model 16
3.1.5 Forecast System Design for Large-Scale Model 17
3.2 System Implementation 18
3.2.1 Practical Deployment 19
3.2.2 Data Collecting 20
3.2.3 Data Monitoring 22
Chapter 4 Model Design 26
4.1 Data Preprocessing 26
4.2 Simple Algorithms Experiments 31
4.2.1 Regression Algorithms 31
4.2.2 Classification Algorithms 32
4.3 Final Model Architecture 34
4.3.1 Adaptive Boosting 34
4.3.2 Stacked Generalization 35
Chapter 5 Experiments and Results 38
5.1 Single-Shed Model 39
5.2 Multi-Sheds Model 40
Chapter 6 Conclusion and Future Works 45
References 47

[1] “M2M”, https://en.wikipedia.org/wiki/Machine_to_machine, accessed: 2020-7-15
[2] “Wireless Technologies,” https://iot-fpms.fandom.com/wiki/LPWAN_networks
[3] “LoRa,” http://www.semtech.com/technology/lora, accessed: 2020-7-15
[4] “Semtek,” https://www.semtech.com/technology/lora/what-is-lora, accessed: 2020-7-15
[5] “LoRaWAN,” http://www.lora-alliance.org/, accessed: 2020-7-15
[6] LoRa Alliance, “LoRaWAN specification,” Jul. 2016, Available: https://lora-alliance.org/ accessed: 2020-7-15
[7] “LoRaWAN Architecture,” https://lora-alliance.org/about-lorawan
[8] “TDM,” https://en.wikipedia.org/wiki/TDM, accessed: 2020-7-15
[9] Y. Chang, T. Huang and N. Huang, "A Machine Learning Based Smart Irrigation System with LoRa P2P Networks," in 2019 20th Asia-Pacific Network Operations and Management Symposium (APNOMS), Matsue, Japan, 2019, pp. 1-4
[10] L. Dan, C. Xin, H. Chongwei and J. Liangliang, "Intelligent Agriculture Greenhouse Environment Monitoring System Based on IOT Technology," in 2015 International Conference on Intelligent Transportation, Big Data and Smart City, pp. 487-490, Dec 2015.
[11] S. B. Saraf and D. H. Gawali, "IoT based smart irrigation monitoring and controlling system," in 2017 2nd IEEE International Conference on Recent Trends in Electronics, Information & Communication Technology (RTEICT), pp. 815-819, May 2017.
[12] “ZigBee,” http://www.zigbee.org/, accessed: 2020-7-15
[13] “LPWAN," https://en.wikipedia.org/wiki/LPWAN, accessed: 2020-7-15
[14] D. Davcev, K. Mitreski, S. Trajkovic, V. Nikolovski and N. Koteli, "IoT agriculture system based on LoRaWAN," in 2018 14th IEEE International Workshop on Factory Communication Systems (WFCS), pp. 1-4, June 2018.
[15] D. Ilie-Ablachim, G. C. Pătru, I. Florea and D. Rosner, "Monitoring device for culture substrate growth parameters for precision agriculture: Acronym: MoniSen," in 2016 15th RoEduNet Conference: Networking in Education and Research, pp. 1-7, Sep 2016.
[16] L. Jing and Y. Wei, "Intelligent Agriculture System Based on LoRa and Qt Technology," in 2019 Chinese Control and Decision Conference (CCDC), Nanchang, China, 2019, pp. 4755-4760c
[17] G. Singh, D. Sharma, A. Goap, S. Sehgal, A. K. Shukla and S. Kumar, "Machine Learning based soil moisture prediction for Internet of Things based Smart Irrigation System," in 2019 5th International Conference on Signal Processing, Computing and Control (ISPCC), Solan, India, 2019, pp. 175-180
[18] “Multiple linear regression,” https://www.investopedia.com/terms/m/mlr.asp, accessed: 2020-7-15
[19] “Gradient Boosting Regression Trees,” https://www.datarobot.com/blog/gradient-boosted-regression-trees/, accessed: 2020-7-15
[20] “Random forest,” https://en.wikipedia.org/wiki/Random_forest, accessed: 2020-7-15
[21] “Ensemble learning,” https://cs.nju.edu.cn/zhouzh/zhouzh.files/publication/springerEBR09.pdf, accessed: 2020-7-15
[22] R. Polikar, "Ensemble based systems in decision making," in IEEE Circuits and Systems Magazine, vol. 6, no. 3, pp. 21-45, Third Quarter 2006
[23] R. Maclin, D. Opitz, "Popular Ensemble Methods: An Empirical Study," in Journal Of Artificial Intelligence Research, Volume 11, pages 169-198, 1999
[24] Rei Sonobe, Yuki Yamaya, Hiroshi Tani, Xiufeng Wang, Nobuyuki Kobayashi, Kan-ichiro Mochizuki, "Crop classification from Sentinel-2-derived vegetation indices using ensemble learning." in Journal of Applied Remote Sensing 12 (2018)
[25] J. Treboux and D. Genoud, "High Precision Agriculture: An Application Of Improved Machine-Learning Algorithms," in 2019 6th Swiss Conference on Data Science (SDS), Bern, Switzerland, 2019, pp. 103-108
[26] Chih-Wei Hsu, Chih-Chung Chang, Chih-Jen Lin (2003), "A Practical Guide to Support Vector Classification, " Department of Computer Science, National Taiwan University
[27] “Bagging, Boosting and Stacking,” https://towardsdatascience.com/ensemble-methods-bagging-boosting-and-stacking-c9214a10a205, accessed: 2020-7-15
[28] Ho, T., "Random Decision Forests", in Proceedings of the Third International Conference on Document Analysis and Recognition, pp. 278-282, 1995.
[29] “Kaggle,” https://www.kaggle.com/, accessed: 2020-7-15
[30] “Titanic dataset,” https://www.kaggle.com/c/titanic, accessed: 2020-7-15
[31] “Introduction to Ensembling,” https://www.kaggle.com/arthurtok/introduction-to-ensembling-stacking-in-python, accessed: 2020-7-15
[32] “Otto Group Product Classification Challenge,” https://www.kaggle.com/c/otto-group-product-classification-challenge/discussion/14335, accessed: 2020-7-15
[33] “AdaBoost,” http://www.inf.fu-berlin.de/inst/ag-ki/adaboost4.pdf, accessed: 2020-7-15
[34] “Raspberry Pi3,” https://www.raspberrypi.org/products/raspberry-pi-3-model-b, accessed: 2020-7-15
[35] “MQTT,” http://mqtt.org, accessed: 2020-7-15
[36] “RS-485,” https://en.wikipedia.org/wiki/RS-485, accessed: 2020-7-15
[37] “RESTful,” https://en.wikipedia.org/wiki/Representational_state_transfer, accessed: 2020-7-15
[38] “Richards equation,” https://en.wikipedia.org/wiki/Richards_equation, accessed: 2020-7-15
[39] H. Bonab and F. Can, "Less Is More: A Comprehensive Framework for the Number of Components of Ensemble Classifiers," in IEEE Transactions on Neural Networks and Learning Systems, vol. 30, no. 9, pp. 2735-2745, Sept. 2019