自組織映射圖網路應用於東方果實蠅監測網自動檢測與早期警報系統系統

An Automatic Diagnosis and Warning Scheme for the Ecological Monitoring System for the Bactrocera dorsalis (Hendel) using Self-Organizing Map

廖敏勝(Min-Sheng Liao)

指導教授 ： 廖國基

共同指導教授 ： 江昭皚

國立臺灣大學/生物資源暨農學院/生物產業機電工程學研究所/碩士(2011年)

https://doi.org/10.6342/NTU.2011.01020

摘要

近年來，無線感測器網路技術發展迅速，已被廣泛地應用在各種商業與工業的應用中，儼然成為無線感知技術的重要基礎技術之一。搭配適當的後端監控技術，可使無線感測器網路被用來檢測特殊事件，例如：異常的高(低)溫、濕度，後端監控系統可針對特定情況作相關的後續環控補償或緊急通報作業。但是，此類特殊事件的產生也有可能是因為感測器校正錯誤或感測器故障所造成。有鑑於此，為避免系統誤報或誤判的情況發生，吾人須建立一套自主式的檢測機制，以預防類似的錯誤發生。本論文以東方果實蠅(Bactrocera dorsalis)生態監測網為例，設計一套即時監測系統，在害蟲族群數量激增時，會即時地將警報訊息傳出；另外，當感測器讀數發生錯誤時，亦可正確地將事件歸類為故障事件，並通知系統管理員進行維修。
東方果實蠅為臺灣地區水果作物最主要的害蟲之一，過去傳統的害蟲族群量調查方式是以人力為主。但受制於人力成本有限，且以人工調查的方式，無法同時取得大量的環境參數資料，以至於無法即時反映當下蟲害發生時的環境因素，而造成無法作適當的蟲害控制與管理。為了取代過去藉由人工調查的記錄方式，本研究結合GSM與無線感測器網路之技術，發展一套測量耕地環境參數且自動化的即時監測系統。自主式檢測機制則是採用自組織映射圖網路來作為判斷各項讀數是否有特殊事件的發生。本研究將實現三種主要檢測事件類型：感測器是否正常運作中、感測器是否監測到東方果實蠅出沒的現象，以及感測器是否發生故障。
本研究已於民國九十九年七月二日與九月八日，分別設置兩套植基於無線感測器網路技術之東方果實蠅生態監測系統於宜蘭縣大湖地區的果園之中。此系統可監測該果園受到東方果實蠅危害的程度，並利用自組織映射圖網路建立四種季節的分類模型，並將感測讀數分類到本研究所設定的三種主要檢測事件類型。實驗結果顯示四種季節的分類模型，對檢測事件類型的判定效果極佳，顯能有助於系統鑑別監測資料是否異常或發生故障，已達農業自動化之效。

關鍵字

東方果實蠅 ； 無線感測器網路 ； 自組織映射圖 ； 農業自動化

並列摘要

Recently, wireless sensor networks (WSNs) technologies have been rapidly developed. WSNs have been widely utilized in a variety of commercial and industrial applications. If a back-end monitoring technology accompany with WSNs, it can be used to detect the specific events of WSNs. For example, with unusually high or low temperature and humidity, the back-end monitoring system can aim at the specific events and prepare for the follow-up operations of environmental control and emergency notification. However, the causation of the specific events may also be resulted from sensor calibration error or sensor failure. In order to avoid the false positives of the monitoring system, it must establish a mechanism of autonomous detection to prevent similar situations. This work instanced an oriental fruit fly (Bactrocera dorsalis) ecological monitoring network, and it designed a real-time monitoring system. This system can send warning messages to the correspondents when the pest surged. In addition, when a sensor reading error occurs, this system can accurately classified as a fault event, and notify the correspondents to conduct system maintenance.
The oriental fruit fly is the major pest that attacks fruit in Taiwan. In the past, the monitor techniques mostly depended on manual measurement. Due to limited budgets on manpower, manual measurements cannot acquire much environmental data at the same time, thereby losing the immediateness of subsequent data analysis, so it is almost impossible to execute appropriate pest control in the right time at the right place. In order to replace previous manual measurements, this work combined GSM technologies with WSN technologies to develop an automated real-time monitoring system which can measure environmental parameters for cultivated land. The mechanism of autonomous detection used self-organizing map to detect the parameters of specific events. This work achieves three primary goals: 1) the sensors operate normally; 2) the sensors detect the infestation of the oriental fruit; and 3) the system detects unusual sensor readings.
Two monitoring systems of the oriental fruit fly have been actually deployed in two orange orchards at Yuanshan, Yilan, on July 2 and Sept. 8, 2010, respectively. The systems can monitor the oriental fruit fly in the orchards and use self-organizing map to establish classification models for four seasons. The models will classify the readings based on three primary goals set by this work. The experimental results presented that the efficiency of classification models is excellent, and it can help the monitoring system identify whether an error in the monitoring data occurs to achieve agricultural automation.

並列關鍵字

Bactrocera dorsalis ； wireless sensor network ； self-organizing map ； agricultural automation

參考文獻

林冠璋。2007。使用無線感測器網路之自動化害蟲生態監測系統研製。碩士論文。

Akyildiz, I. F. and E. P. Stuntebeck. 2006. Wireless underground sensor networks: research challenges. Ad Hoc Networks. 4(6): 669-686.

Barhak, J. and A. Fischer. 2001. Adaptive reconstruction of freeform objects with 3D SOM neural network grids. Proceedings of Ninth Pacific Conference on Computer Graphics and Applications. pp. 97–105.

Baronti, P., P. Pillai, V. W. C. Chook, S. Chessa, A. Gotta, and Y. F. Hu. 2007. Wireless sensor networks: A survey on the state of the art and the 802.15.4 and ZigBee standards. Computer Communications. 30(7): 1655–1695.

Bezedk, J. C., 1981. Pattern Recognition with Fuzzy Objective Function Algorithm. Plenum Press, New York.