Research on Anomaly Detection Algorithm for Meteorological Radar Echo Extrapolation Based on Deep Learning

This project intends to use deep learning technology to combine CNN with LSTM to establish a rapid forecast and anomaly detection method for weather radar echo signals, providing a new idea for radar echo extrapolation detection. CNN are used to analyze the spatial characteristics of target echoes, and LSTM are used to analyze their time series. Then an adaptive anomaly detection method based on multi-scale features is proposed to improve the sensitivity of the model. The performance of the proposed method under various meteorological conditions is evaluated using measured meteorological radar data. Experiments have shown that the radar echo extrapolation algorithm based on deep learning has higher accuracy than traditional algorithms, especially for the detection of extreme weather events. Through the analysis of specific data, it is found that the accuracy, recall rate, F1 value, etc. of this method have reached more than 95%, and the false alarm rate has been effectively reduced. The research results of this project will provide a new idea and method for the extrapolation and anomaly detection of real-time echo data in weather forecasting.