Then, the time delay neural network is exploited to model the long term feature representation and constructs the regression model. The eigenvectors associated with the modal signal space are extracted as the input feature. The first adopts the classical two-stage scheme, in which feature extraction and DNN analysis are independent steps. Two methods are proposed to estimate the range and depth of a broadband source through different neural network architectures. This paper applies DNNs to source localization in a shallow water environment. This method is evaluated using the SWARM95 experimental environment to address the real-world issues.ĭeep neural networks (DNNs) are advantageous for representing complex nonlinear relationships. As an application, one can use the SIW-induced striations to monitor the positions (and wavefronts) of nonlinear internal waves on a continuous time basis. Theory and simulations are developed in this paper. The striation is caused predominantly by mode coupling (as opposed to mode interference), which changes the mode amplitudes as the SIWs move. In this case, the striation slope is related to the source-to-SIW range, and not the source–receiver range. When nonlinear solitary internal waves (SIWs) are present in the propagation path, one finds a different phenomenon: the depth-integrated energy will display striations with respect to time and frequency. Plotted against range and frequency, the spectrogram displays striated bands of constant acoustic intensity levels as explained by the waveguide-invariant theory, the slope of which can be used to estimate the source range by the waveguide-invariant parameter “beta.” Given a vertical array of receivers, one finds, however, no frequency striation in the depth-integrated acoustic energy, as the total energy corrected for geometrical spreading remains “conserved” except for mode attenuations. Modal interferences are evident in the spectrogram of broadband signals propagated over distances and received on a single receiver. This method can also be used to analyze the value of β for any other kind of SSP.

An analytic criterion of the sign of β is developed. The results show that β can be positive and increase continually from around 1 to +∞ or negative from -∞ to about -3. In this paper, by dividing the SSP into the non-refractive part and the refractive part, a clear explanation of how the value of β is affected by the thermocline is presented. When a thermocline exists, β for refractive modes (with a turning point) is no longer constant and can even change its sign. It was known that β has some 'canonical' values for waveguides with simple sound speed profiles (SSP). A reasonable prediction of the value of β is often necessary in such applications. It has been found that β may be useful in many applications in underwater acoustics. On the basis of the principle of stationary phase, Chuprov proposed a scalar parameter-the waveguide invariant β, to interpret the dispersive properties of underwater acoustic waveguide. The resulting range estimators demonstrate better peak-to-sidelobe ratios than a simple incoherent average. An appendix demonstrates how these sidelobe properties can be exploited when combining ambiguity surfaces through use of gradient and Radon transform information. Sidelobe trajectories also exist in the range-depth plane, but they contain no information about the correct source depth. These interference trajectories are demonstrated to exist in simulations, broadband source tows, and a type A blue whale vocalization. As the ambiguity surface of the Bartlett matched-field processor has a physical interpretation in terms of a time-reversed acoustic field, with the sidelobes analogous to local interference maxima, these invariant concepts can be reformulated for application to MFP. This behavior can be explained by the theory of waveguide invariants, which predict the local behavior of interference maxima/minima of acoustic intensity in the frequency-range plane. In isovelocity or bottom-interacting environments the sidelobe trajectories are straight lines that converge to the true range at zero frequency, while environments with upward-refracting sound-speed profiles have trajectories that asymptotically converge as the frequency approaches infinity.

This sidelobe shift can be viewed as a continuous trajectory in a range-frequency plane at a fixed depth, where the trajectories converge to the correct source range for a perfectly matched surface. Ambiguity surface sidelobes generated by the Bartlett matched-field processor (MFP) shift location with frequency.

Soundwaves bartlett