Research on the Positioning and Casualty Prediction Model of Mobile Phone Signaling Data in Disaster Warning Based on Deep Learning

A kind of disaster alarm model is proposed, which can realize the location of people and forecast the disaster by means of mobile telephone signal. First, this paper preprocesses the mobile phone signaling data, and uses data cleaning, denoising and feature extraction techniques to extract time series and spatial information from it. Then, LSTM and CNN are combined to model the spatiotemporal data, and the time series features and spatial features in the signaling data are processed respectively. The model can locate the distribution of personnel in real time after the disaster occurs, and predict the number of casualties by combining environmental factors (such as building structure, geographic information, etc.). The experimental simulation results show that the positioning accuracy of the model based on deep learning reaches 89.7%, which is 13.5% higher than that of the traditional method; in terms of casualty prediction accuracy, the prediction error of the model is 15.3%, which is significantly higher than the 30.2% of the traditional linear regression method. Further experiments also show that the system can effectively optimize the evacuation process and improve the efficiency of post-disaster rescue. It is verified by comparing with actual disaster data.