Inverse Problem Theory and Methods for Model Parameter Estimation
Inverse Problem Theory and Methods for Model Parameter Estimation
| Hi-index | 0.00 |
Geoacoustic parameters inverted from reverberation vertical correlation (RVC) are often directly used to predict the acoustic transmission loss (ATL) in shallow water. However, little work has been applied to the problem of quantifying uncertainty in predicted ATL produced by geoacoustic parameters uncertainties. In this paper, a posterior predictive probability analysis method (PPPAM) is first employed to evaluate the effects of geoacoustic parameters uncertainties inverted from RVC data on both coherent and incoherent ATL predictions. Where, the geoacoustic parameters uncertainties are characterized by their posterior probability distributions (PPD). And then the uncertainties of ATL prediction are analyzed quantitatively based on the posterior predictive probability distributions of ATL, which are the function of the PPDs of geoacoustic parameters and can be estimated using a Markov Chain Monte Carlo sampling method. Finally, the Yellow Sea Reverberation experimental results illustrate the PPPAM and show that: (1) in the range from 1km to 5 km, the mean values of 90% posterior credibility intervals (PCI) of coherent and incoherent ATL in frequency range of 500∼800Hz exceed 6dB and 3dB, respectively; (2) the coherent ATL are more difficult to predict near the positions of destructive interference of the normal modes. These results derived in this paper are helpful to evaluate and improve the detection and localization performance of sonar system.