基于RIME-VDSR神经网络的声场超分辨率重建OA

This paper investigates the reconstruction problem of ultrasonic fields in liquids based on a Very Deep Super-Resolution(VDSR)Deep Neural Network.The COMSOL-MATLAB co-simulation method is adopted to simulate the sound fields in the liquid under the radiation of transducers with different positions and different operating frequencies.The simulation data are saved to construct a dataset.It constructs a VDSR Deep Neural Network,integrates the RIME optimization algorithm,and utilizes the dataset to complete the neural network training and testing.The research finds that using the RIME optimization algorithm can improve the reconstruction precision.Furthermore,the paper analyzes the reconstruction of low-resolution sound fields obtained through down-sampling with varying scaling factors.It reveals that reconstruction accuracy gradually decreases as the scaling factors reduce,and the reconstruction accuracy of high-frequency sound field is more sensitive to the scaling factor than that of low-frequency sound field.Finally,the method is compared with the existing sound field reconstruction method.The results show that the reconstruction precision of the proposed method is slightly better than the existing methods for low-frequency sound field,and the advantages of the proposed method are more significant for high-frequency sound field.