中文摘要：

能量是无线传感器网络应用中最宝贵的资源，由于节点能量消耗完毕死亡而导致网络瘫痪是一种典型故障。针对此故障形式提出了一种基于簇内数据聚类算法的故障检测技术，该方法利用无线传感器网络在按照地理位置进行分簇的基础上，通过采用数据聚类的方法，在能量不受限的汇节点处进行簇内数据再聚类，然后设置阈值进行故障检测。通过仿真实验证明，合理选择检测阈值，该方法在保持基于历史与邻居数据的节点自检测方法较高准确率的基础上，极大地减小了能量消耗，且明显降低了故障检测误警率。

英文摘要：

Energy consumption is vital in wireless sensor networks （WSNs） since node death is a typical fault caused by energy depletion. Aiming at this fault, a fault detection technique based on clustering （FDTC） is proposed in this paper. Employing carefully-chosen threshold, the clustering algorithm is deployed on the sink node with unconstrained energy consumption on the basis of dividing the WSNs into clusters according to their geographical position. Simulation experiment results demonstrate that by choosing appropriate threshold, this algorithm greatly reduces energy consumption and false alarm probability while still maintains high detection accuracy.