关键词: 无线电监测, 异常信号, 构造神经网络, 覆盖聚类

Abstract: In this paper，the constructive neural networks is used to cluster large-scale communication signals，then a time-spectrum correlation method to detect abnormal radio signals is proposed.In the clustering，a method of adding the weight of initial clustering granularity is used to improve the covering clustering algorithm，which can enhance clustering performance.In the detecting of abnormal radio signal，the time and spectrum characters of a signal are related to find whether the signal is abnormal.The experimental results show the feasibility and effectiveness of this method.

Key words: radio monitor, abnormal signal, constructive neural networks, covering clustering