Acoustic benchmark validation of GRASP ASW search plans

Abstract

A genetic algorithm is used in non-homogeneous and anisotropic environments to nearly optimize sonar search tracks. The optimization metric is maximum cumulative detection probability for a specified sonar (passive or active) against a target with specified characteristics (acoustic and tactical) during a fixed time period. This application for search planning is named GRASP, for Genetic Range-dependent Algorithm for Search Planning. A validation of GRASP solutions in various ocean environments is shown under benchmark conditions, i.e., fairly simple synthetic environments and a simple target distribution. Directional, range-dependent sonar performance (signal excess) is estimated from parabolic equation calculations of transmission loss. The search tracks produced by the genetic algorithm are generally intuitive; they usually remain in high detection areas. When track solutions are counter-intuitive, we explain unexpected behavior (e.g., zigzag turns, tracks offset from symmetric features, and occasional departures from high detection areas) in terms of details in the acoustic field.