使用K-means分群以及基因演算法建置最佳化無線感測網路系統

Abstract

由於無線結構健康監測網路中之感測器本身是用鋰電池來提供運作所需的能量，所以如何穩定而持久的供給運作所需能源是無線結構健康監測網路一個重要研究議題且 需突破的限制，要解決此一問題能藉由研發新的能源擷取機制，或透過有效的節能管理系統，例如控制感測器進入低功耗模式或稱為睡眠模式，當滿足特定條件時(如地震、颱風、或其他人為災害造成結構物震動)則可迅速地喚醒感測器進入工作模式，並即時量取結構物受震反應訊號達到結構健康診斷的目的。 另外，配置適量的感測器於最佳量測點(位置) 亦是有效節能的手段之一。本研究之目的分兩部分，第一部分係在考慮感測器數量及量測資訊量多寡下，使用有效獨立法EFI(Efective Idependence method)進行最佳化感測器布置，第二部分為使用K-means以及基因演算法在傳輸距離以及單群數量限制下進行感測器分群，妥善的安排感測器的分群以及接收器的布置，達到增加無線感測網路監測之品質與穩定度。

Lithium-ion batteries are the conventional energy source for sensing nodes in a wireless sensors network (WSN) in structure health monitoring (SHM). The stability and durability of the energy supply for sensing nodes in a WSN need improvement. Developing a new energy harvesting system or control sensors entering low-power mode or sleep state periodically by duty cycling the radio is common used to reduce energy consumption.In addition, the proper sensor configuration to the optimal measuring point (location) is also one of the effective approaches of energy conservation. The purpose of this study of two parts, the first part is considering the number of sensors and measuring the amount of information, using effective independence method EFI (Efective Idependence method) to optimize the sensor arrangement the second part is using K -means and Genetic Algorithm group the sensors under considering transmission distance and the number of single group limit sensors, grouping and proper arrangements to receive the sensor arrangement, to increase the quality and stability of the wireless sensor network monitoring.