農業4.0之紫式決策架構與肉雞重量預測提升聰明生產之實證研究

UNISON Decision Framework to empower Agriculture 4.0 and An empirical study for the broiler weight prediction for smart production

阮有山(Huu-Son Nguyen)

指導教授 ： 簡禎富

國立清華大學/工學院/工業工程與工程管理學系/碩士(2022年)

摘要

全球對農產品的需求增加，預計十年後增長 15%，且於2029年肉類需求量將提升約 12%。其中，肉雞產品需求預計將在2050年提升一倍。為符合市場需求，肉雞飼養策略和飼養週齡即為挑戰。肉雞屠宰體重準確性為肉雞採食計劃過程中的關鍵問題之一，且為肉雞供應鏈資金比重佔據最高的一環。為解決肉雞供應鏈問題，本研究建立紫式決策架構（UNISON FRAMEWORK）並結合人工智慧與數學模型，用於環境參數進行肉雞屠宰體重預測用於精準預測家畜養殖。本研究架構為提前7天的滾動預測肉雞屠宰體重機制，用以輔助肉雞屠宰過程中的決策，以優化肉雞供應鏈。一項實證研究於台灣的肉雞養殖場進行，以驗證本研究架構的有效性。結果指出本研究架構能提供有效與較低成本的飼養方案，以解決肉雞體重的長期滾動預測問題，同時防止因超餵、異常屠宰體重、運營費用和客戶關係而造成的損失。

關鍵字

農業 4.0 ； 肉雞屠宰體重預測 ； 精準畜牧業 ； 肉雞採食計劃 ； 決策分析 ； 肉雞供應鏈

並列摘要

Global demand for agricultural products has been increasing and is projected to rise by 15% over the next decade, and meat consumption is predicted globally to grow about 12% by 2029. In particular, the global demand for broiler chicken products is expected to double by 2050. The high demand for broiler chicken is a great opportunity for the development of the broiler farming industry, and also is a big challenge to improve broiler rearing efficiency and productivity to meet the huge demand from customers. The accuracy of the weight prediction of broilers is the major issue leading to serious problems for the broiler harvesting planning process, and costs a lot of money in the whole broiler supply chain. To address realistic needs, this study aims to develop a UNISON decision framework that integrated artificial intelligence methodology and mathematical growth function for future broiler weight prediction based on the environmental parameters for precision livestock farming. The framework provides the prediction mechanism for the long rolling-forecast period of 7 days ahead to support decisions in the broiler harvesting production processes for broiler supply chain optimization. An empirical study was conducted on broiler farms in Taiwan to demonstrate the validity of this framework. The results have shown that the proposed framework is an effective and low-cost solution to address the long-term rolling-forecast problem for broiler weight prediction and prevent profit loss from overfeeding, abnormal weight, operation expenses, and losing key customers.

並列關鍵字

Agriculture 4.0 ； Broiler weight prediction ； Precision livestock farming ； Broiler harvesting planning ； Decision analysis ； Broiler supply chain

參考文獻

Rumelhart, D. E., Hinton, G. E., & Williams, R. J. (1986). Learning representations by back-propagating errors. Nature, 323(6088), 533-536. doi:10.1038/323533a0

Amraei, S., Mehdizadeh, S. A., & Sallary, S. (2017). Application of computer vision and support vector regression for weight prediction of live broiler chicken. Engineering in agriculture, environment and food, 10(4), 266-271.

Astill, J., Dara, R. A., Fraser, E. D. G., Roberts, B., & Sharif, S. (2020). Smart poultry management: Smart sensors, big data, and the internet of things. Computers and Electronics in Agriculture, 170. doi:10.1016/j.compag.2020.105291

Cândido, M., Tinôco, I., Albino, L., Freitas, L., Santos, T., Cecon, P., & Gates, R. S. (2020). Effects of heat stress on pullet cloacal and body temperature. Poultry Science, 99(5), 2469-2477.

Chen, Y.-J., & Chien, C.-F. (2018). An empirical study of demand forecasting of non-volatile memory for smart production of semiconductor manufacturing. International Journal of Production Research, 56(13), 4629-4643.