提出了一种基于改进Kruskal算法的无线传感器网络(WSN)故障节点检测方法.该方法首先通过集中式的改进Kruskal最小生成树算法来获取可信的节点集合,之后依据可信节点,采用邻居节点比较算法对传感器节点的感知值进行分布式分析和处理,判定发生故障的传感器节点.同时为了容忍节点的临时故障,引入了时间冗余.仿真结果表明,在节点故障率高达35%时,该方法依然能快速定位故障节点,并且同时保证很高的检测精确度. This paper proposes a new detection method for WSN fault nodes based on improved Kruskal algorithm.The method first obtains the trusted node set by improving the Kruskal minimum spanning tree algorithm and then uses the neighbor nodes The algorithm compares and evaluates the sensor nodes’ perceived values ​​by distributed algorithms to determine the sensor nodes that have failed, and introduces time redundancy in order to tolerate the temporary node failures.The simulation results show that when the node failure rate is as high as 35% The method still locates the failed node quickly and at the same time guarantees a high detection accuracy.