A Prediction Model for the Equivalent Parameters of an Acoustic Transducer Based on DPSD and LSTM Neural Network

This article proposes a novel model for predicting the equivalent parameters of underwater acoustic transducers. To address this challenging task, the model utilizes Digital Phase-Sensitive Detection(DPSD) on the detection signal to obtain the amplitude and phase difference. Subsequently, the model applies Long Short-Term Memory(LSTM) and Gated Recurrent Unit(GRU) neural networks for classification and regression predictions of the equivalent models and parameters of the transducer. Through simulation experiments using MATLAB, the experimental results prove that the LSTM prediction model has higher prediction accuracy and robustness than the GRU prediction model. This provides a new method for modeling equivalent parameters of underwater acoustic transducers.