Seismic phase recognition model with low SNR based on U-net

Aiming at the problem of low recognition accuracy and high missed detection rate of seismic phase recognition of low signal-to-noise ratio seismic signals, a new seismic phase recognition model UBAN (U-net-Bidirectional Gated Recurrent Unit-Attention Network) is designed based on U-net neural network framework, combined with Bi-GRU bidirectional gated recurrent unit and Attention attention mechanism. In this model, Bi-GRU bidirectional gated recurrent unit and Attention attention mechanism are added between the U-net coding layer and the decoding layer. Bi-GRU is suitable for processing long time series signals,and attention mechanism is used to pay attention to the time series characteristics of seismic phases, ignoring the advantages of useless features such as noise and peaks, improving the network ‘s ability to perceive the arrival time characteristics and improving the recognition accuracy. The seismic data of Stanford University are used to train and test the designed UBAN phase recognition model. The experimental results show that the UBAN phase recognition model shows good performance in the phase recognition of low signal-to-noise ratio seismic signals, which provides a new idea for the research of low signal-to-noise ratio seismic signal phase recognition method.