基于Resnet和FrFT的水下多目标定位OA北大核心CSTPCD

Aiming at the problem that the accuracy of underwater multi-target localization depends on the sampling interval,a multi-target localization algorithm(MLRF)combining the deep residual network(ResNet)and the fractional fourier transform(FrFT)was proposed.Firstly,an simulation model of underwater multi-target localization was built.Secondly,a network suitable for multi-target localization was trained using ResNet.Thirdly,FrFT was performed on the echoes to obtain their spectral maps,and ResNet was used to identify the spectral peaks and calculate the number of targets.Finally,a multi-target localization algorithm was designed to distinguish different target data and estimate the position and velocity of each target.Simulations were conducted to compare the effects of signal-to-noise ratio(SNR)and the number of anchor nodes on localization.The results show that the root mean square error of localization of the MLRF algorithm decreases with the increase of SNR and the number of anchor nodes,and is below 2 m when SNR equals-5.The approach is less affected by noise and can effectively address the problem of underwater multi-target localization as well as contribute to the study of underwater multi-target tracking.