After the earthquake,the communication data usually undergoes abnormal changes. Through the analysis of the communication data abnormality,effective disaster data can be provided,which is a basis for decision-making for earthquake relief. Analyzing abnormal communication data helps to better understand the impact of the earthquake and can provides auxiliary support for following-up command and disaster relief. This paper analyzes the communication data anomalies after the earthquake based on LSTM,which mainly includes:communication data stream preprocessing,anomaly detection model based on LSTM,and data distribution change detection model. The model in this paper is capable of identifying abnormal changes in communication data and provides disaster data for subsequent hazard analysis.