透過您的圖書館登入
IP:18.191.102.112
  • 學位論文

模糊群聚於異常行為的偵測

Fuzzy Clustering in Abnormal Behavior Detection

指導教授 : 許輝煌

摘要


在這篇研究中,我們模擬老人獨自生活,老人一個在家可能有意外發生,希望在意外發生的時候,可以在最短時間內通知其他人。在本實驗利用RFID設計一套系統來收集老人行為資料,在老人的居家環境裡面放上許多RFID Tag,RFID Tag擺的位置是老人平常可能活動的範圍,並且讓老人攜帶具有RFID Reader的PDA,由RFID Reader讀取RFID Tag的訊號,由於每一個RFID Tag距離不同而産生不同的訊號強弱值(RSSI 範圍:0~255),我們希望由這個RFID系統記錄老人平常移動狀況和行為模式,藉由記錄的RSSI值來判斷老人是否有異常行為發生。一開始並不知道何謂正常的行為資料,首先必須記錄老人平常的行為,利用平常收集行為資料用Fuzzy C-Means來做學習,利用Fuzzy C-Means的方法來建立正常行為的模型,建立好正常行為模式後可以判定後來的行為是否屬於異常,在實驗中我們模擬老人可能出現的異常行為當作異常資料,並且定義警報條件將判斷為異常資料點做判斷,使系統可以在正確的時間發出警報,最後利用不同的異常資料的結果證明Fuzzy C-Means的方法判斷異常行為是有效率的,雖然有些行為並沒有有效的判斷出來,但是大部份的異常行為判斷的時間點是好的,最後未來還可以利用其它方法對行為偵測的部份做改進,提升這套系統判斷的準確率。

並列摘要


In our research, we consider the elderly who lives alone. The elderly may involve with an accident. We hope to discover the accident in time. In this paper, we design an intelligent system that uses ten RFID tags deploying in the home environment. The elderly carries a PDA with an RFID reader that receives RSSI values of the RFID tags. We record movement of the elderly by RSSI values every day. First, the intelligent system needs to record movement data for about two weeks. The movement data are trained as normal patterns by Fuzzy C-Means. If a test data sequence is not part of normal movements, it is treat as abnormal. Secondly, we develop an alarm system that is used to determine the right time to send an alarm. Finally, we simulate some possible abnormal behaviors of the elderly. The experimental results show that the intelligent system can detect anomaly as expected.

參考文獻


[16] 許輝煌, 陳建丞, "使用RFID與機器學習技術於獨自在家老人的異常行為偵測," 淡江大學資訊工程學系, 2009
[7]E. Alpaydin, "Introduction to Machine Learning," Chapter7, The MIT Press., 2004
[9] J. C. Bezdek, "Pattern recognition with fuzzy objective function algorithm," New York:Plenum, 1981
[11] D. Wang, D. S. Yeung, E. C. C. Tsang, "Structured One-Class Classification," IEEE Trans. Systems, Man, and Cybernetics, Part B: Cybernetics, Vol. 36, no. 6, pp. 1283 – 1295, 2006
[12] H. Seki, "Fuzzy Inference Based Non-Daily Behavior Pattern Detection for Elderly People Monitoring System," in Proc. Engineering in Medicine and Biology Society, Annual International Conference of the IEEE, EMBC 2009, pp. 6187 – 6192, 2009

延伸閱讀