Sequential source localization using unscented particle filter under circumstances of highly nonlinear time-evolving sound speed profiles

摘要

Ocean dynamic processes such as sea waves, rainfall, and internal waves lead to modeling mismatches and influence the performance of source localization of matched-field processor (MFP). Based on empirical orthogonal functions (EOFs) and the state-space model, which describe the evolution characteristics of sound speed profile (SSP), source range and depth are estimated via the acoustic array data simulated by the measured SSPs and prior seabed acoustic properties. In consideration of the characteristics of nonlinear systems and non-Gaussian distributions in the underwater acoustic channel, an algorithm for the unscented particle filter (UPF) is implemented for the tracking of moving source under the circumstances of time-evolving sound speed profiles. The validity and practicality are demonstrated by the simulation, which indicate that the proposed scheme enables the continuous tracking of the moving source.