請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/69516
標題: | 開發物聯網系統以即時監測音頻並評估蜂群狀態 Developing an IoT-based Audio Monitoring System for Evaluation of Honey Bee Colony Conditions in Real-time |
作者: | Sheng-Hao Chen 陳聖皓 |
指導教授: | 江昭皚(Joe-Air Jiang) |
關鍵字: | 音頻監測系統,蜜蜂,健康指標, Audio monitoring system,Honey bee,Colony health indicator, |
出版年 : | 2020 |
學位: | 碩士 |
摘要: | 目前,人類食物中約有三分之一來自開花植物,而其中80%的植物需要蜜蜂授粉。然而,蜜蜂和養蜂業正在遭受全球性危機了解蜂箱狀況是許多研究的目標,開發新的監測系統 以改善蜂巢管理使其更為有效。因此,在這項研究中,建立了一個蜂巢音頻監視系統來記錄蜜蜂的聲音。此外,音頻監控系統會在收集音頻數據後立即對資料進行分析以了解目前蜂群的狀態,這表示本系統能夠即時的回傳蜜蜂狀態。以往的文獻中已證實蜂群聲音在異常狀態將與正常狀態下有很大不同,並且針對失王、分蜂等問題多有研究,但是,它們大多數只聚焦於蜜蜂的單個異常狀態而並非針對多種蜜蜂異常現象進行研究。因此,在這項研究中,我們提出了一種結合機器學習系統的音頻監控系統,用於數據分析和分類模型的建立。除了能夠同時對三種不同的蜜蜂異常進行分類外,包括失王,病毒和農藥。該系統具有許多功能例如實時分類,遠程控制和分類模型的自動更新。為了使養蜂人或用戶更輕鬆地管理蜂群,實時音頻監視系統可以準確,快速地檢測當前蜂箱的狀態。隨著數據收集的增加,系統本身的分類模型將不斷得到改進和糾正。 Currently, about one-third of human food comes from flowering plants, and 80% of the plants need honey bees for pollination. However, bees and beekeeping are suffering a global crisis. Constant information on beehives’ conditions would be a key to study new diseases like colony collapse disorder and to develop new beekeeping tools to improve the hive management and make it more efficient. Therefore, in this study, a beehive audio monitoring system is established to record the sound of bees. Furthermore, the audio monitoring system will classify the condition of the beehives as soon as it collects the audio data, which means that the system is able to return the real-time honey bee conditions. Previous studies on bee sound have confirmed that the bee sound in an abnormal state will be very different from the normal state. However, most of them only focused on a single abnormal state of bees. Therefore, in this study, we propose an audio monitoring system combined with a machine learning system for data analysis and classification model establishment. In addition to being able to classify three different bee abnormalities at the same time, including queenless, virus, and pesticides. The system has many functions, such as real-time classification, remote control, and automatic updating of the classification model. To allow beekeepers or users to manage bee colonies more easily, the real-time audio monitoring system can accurately and quickly detect the status of the current beehive. The classification model of the system itself is continuously improved and corrected as the data collection increases. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/69516 |
DOI: | 10.6342/NTU202003948 |
全文授權: | 有償授權 |
顯示於系所單位: | 生物機電工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
U0001-1808202012185000.pdf 目前未授權公開取用 | 4.88 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。