建立肉雞飼養管理之決策支援資訊系統

白肉雞具有生長快速、飼料轉換率以及育成率佳的優勢，且其價格僅為有色肉雞之一半，於2015年產值達到187億元，為家禽產業中第三大之項目。隨著台灣加入WTO，並於2005年開放國外雞肉進口後，台灣的白肉雞產業受到進口低價雞肉的衝擊，導致整體利潤降低，而近年來大陸與東南亞地區的飼養技術逐漸提升更是一大威脅，如何在面對進口雞肉競爭時，降低飼養成本並提升飼養效率，是養雞業者應努力之方向。目前針對白肉雞所作的研究，多是針對其環境與營養條件之影響進行探討，然而許多養雞業者仍是依循經驗法則飼養，缺乏較明確的飼養管理模式。因此，本研究嘗試建立一套以網站為基礎的決策支援管理系統，其中包含生長模式以探討環境和營養對肉雞生長之影響，以方便養雞業者能夠取得較佳的飼養管理策略，使肉雞飼養的體重提升且成本降低，並於場內蒐集環境與生長資料，提供雞場監控之功能，以提升管理之效益。本研究於合作之養雞場進行實驗，以計畫合作單位中興大學架設之無線環境感測網路以及本研究自行開發之肉雞自動秤重系統進行資料蒐集。秤重系統之驗證結果顯示，平均相對誤差MRE能夠維持在6% 以內，表示系統量測重量之準確度在可接受之範圍。另外，本研究透過蒐集肉雞生長模式之相關文獻資料，以反應曲面法建立生長模式，接著以實驗雞場蒐集之資料對模式進行修正與驗證。驗證結果顯示，修正後之雞隻增重模式之判定係數R2達到0.82，而相對標準驗證誤差RSEV為17.89%，顯示其具有不錯之預測能力。修正後之飼料轉換率模式的判定係數R2有顯著提升，但仍偏低為0.42，而相對標準驗證誤差RSEV則為21.14%，有待日後持續修正以進一步改善其預測能力。決策建議部分，針對溫度與飼料配方進行建議，由增重模式取得最佳飼養溫度以期達到最大增重，並由最低成本飼料配方運算方式取得最佳飼料配方，以期在滿足肉雞營養需求的情況下達到最低飼料成本。本研究之決策建議結果以飼養手冊之建議值驗證其合理性，在趨勢上有一致性。最後，本研究整合上述功能，建置成系統網站以及網頁程式作為使用者及飼養經營者介面，使用者透過網路便能使用此系統，並取得溫度與飼料配方之管理建議，以更有系統的方法進行雞場管理。而網站平台之決策支援管理系統，具有可遠端使用、多使用者、跨平台、即時更新與方便維護之優點，對於未來持續開發或是應用上皆較為方便。

關鍵字

白肉雞；自動秤重系統；生長模式；環境與營養管理；決策支援系統

並列摘要

The production of broiler is the third largest item in the poultry industry in Taiwan, with the advantages of fast growth, good conversion rate and high survival rate, which made the value of production achieve 187 billion in 2015. Since Taiwan joined WTO and opened for imported chicken in 2005, the broiler industry has suffered from the impact of low-price imported chicken meat, which caused the profit reduced. Besides, the improved broiler breeding techniques in China and Southeast Asia in recent years has also posted a great threat. As a result, how to further reduce the feeding cost and raise the breeding efficiency is an important issue for the broiler farmers. Most of the researches about broiler have been conducted to study the influence of the environmental and nutritional conditions during the growth. However, due to the lack of reliable growth models, many broiler farmers make the management decisions mainly based on their experience. Therefore, this research aims to develop a web-based decision support system (DSS) for broiler management, and established growth models to investigate the influence of the environmental and nutritional conditions on the growth. In addition, environmental and growth information are collected from the broiler farm to provide the farm monitoring function. This research will help the broiler farmers in obtaining better breeding strategies, to improve the yield, to reduce the cost of the broiler, and to enhance the breeding efficiency. This research was conducted in a cooperated broiler farm, and data were collected by wireless environmental sensing network, provided by the project cooperative team National Chung Hsing University, and self-developed broiler automatic weighing system. The validation result of the automatic weighing system showed that the mean average error (MRE) was within 6%, which indicated that the accuracy of the system was acceptable. For the growth model building, related studies were collected to prepare the data and the response surface method (RSM) was applied to build the growth models. The data collected from the broiler farm was used to calibrate and validate the built models. Regarding the weight gain (WG) model, validation result showed that the coefficient of determination (R2) of calibrated WG model could achieve 0.82 and the relative standard error of validation (RSEV) is 17.78%, which demonstrated the good prediction performance of WG model. Regarding the feed conversion rate (FCR) model, R2 could be significantly improved after calibration, but R2 was still low as 0.42, and RSEV is 21.14%, which meant that further calibration is required to improve the prediction performance. The decision making of breeding strategies was focused on the management of temperature and feed formulation. The optimal breeding temperature could be obtained base on the WG model to achieve maximum gain, and the optimal feed formulation could be calculated by least cost feed formulation (LCFF) method so as to simultaneously meet the nutritional needs of broilers and reach the lowest feed cost. In final, the accomplished functions mentioned above were integrated to develop a website of web-based DSS for broiler management, and webpage was served as the user interface for farmers to obtain management suggestions, with a view to manage the farm more systematically and efficiently. Besides, in the form of web-based DSS, the system possesses advantages of remote use, multi-users, cross-platform, instant update and easy maintenance, which is more convenient for development and application.