Moving Platform Ocean Acoustic Inversion With State-Space Models:Theory, Methods And System

Ocean acoustic inversion is a technique used to infer ocean environmental parame-ters. It has the advantage of infering the state of the interior of the survey domain from measurements along the periphery; it can thus provide large-temporal-spatial-scale and range-averaged estimations of ocean environmental parameters with relatively few instru-ments. However, traditional ocean acoustic inversion methods with moored nodes face the tradeoff among system complexity, coverage and resolution. This is because with limited moored nodes, the number of measurement equations are also limited; under certain cov-erage requirement, increasing the resolution means increasing the unknown parameters, which may cause the problem into an underdetermined inverse problem. Introduction of numerous inexpensive moving nodes, such as Autonomous Underwater Vehicles (AUVs) and Underwater Gliders, increases the number of measurement equations due to their mo-bility capability, which provides a new thinking for solving the above problem. This thesis focuses on the theory and system development of using the AUV as a moving source for ocean acoustic inversion.Modeling of the acoustic field generated by a moving acoustic source is a fundamental problem of ocean acoustic inversion with moving platforms. Due to the waveguide effect, different modes have different Doppler shifts, i.e., each modal component propagates at a distinct frequency. To incorporate the Doppler effects for better matching the measured data, the acoustic field is derived based on the waveguide Doppler and normal mode theory. The matched-field inversion approach is adopted to inverse the environmental parameters and source parameters (range, depth, and velocity) with the derived acoustic field as the forward acoustic model. In practice, since the moving platform will be up and down around the specific trajectory, the acoustic field model is also derived for the case that the trajectory is mismatched. The inversion performance analyses are implemented under that model.For range-dependent environment parameters inversion, the number of unknown pa- rameters may be more than the number of measurement equations, which will cause the problem into an underdetermined inverse problem. Since the measured data will be up-dated when the moving platform is at different positions, the measurement equations are increased. The thesis then reformulates the matched-field ocean acoustic inversion prob-lem into a state-space model to track the range-dependent environmental parameters and moving source parameters along the AUV path with the constantly updated measurement equations. The forward acoustic model used here is derived based on the range-dependent waveguide Doppler model. Performances of several sequential filters, including extended Kalman, unscented Kalman, and particle filters, are compared. Different modes will be ex-citated or the ray paths will be different when the source is at different depths, which will influence the inversion performance. In this thesis, a path-planning problem is then formu-lated to find the optimal AUV depths that minimize the mean square error of the sound speed profile along its path.An AUV platform is developed for verification of the ocean acoustic inversion concep-t, which has a modular mechanical, electronic and software design allowing for a simple integration of payload sensors selected for different applications. The modular software development of the platform is based on the MOOS-IvP architecture which has a star-like topology. Two pool and two lake experiments have been performed to test the functions of sensor payloads, navigation, and autopilot of the platform.One lake and one sea experiments have also been conducted to demonstrate the feasi-bility of using the AUV as a moving source for ocean acoustic inversion. Radiated noises of the AUV were received on a vertical line array and processed with the theory proposed in this thesis; estimates of the environmental and source parameters agree well with the on-site measurements.