2025_programme: Phase-Feature-Based Deep Learning Method for Target Depth Discrimination in Shallow-Water Environments

Back


  • Day: June 17, Tuesday
      Location / Time: D. CHLOE at 15:50-16:10
  • Last minutes changes: -
  • Session: 06. Enhancing underwater acoustic sensing through machine learning
    Organiser(s): Cheng Chi
    Chairperson(s): Peng Xiao
  • Lecture: Phase-Feature-Based Deep Learning Method for Target Depth Discrimination in Shallow-Water Environments
    Paper ID: 2297
    Author(s): Qianyan Li, Wenbo Wang, QUNYAN Ren, Li Ma
    Presenter: Qianyan Li
    Abstract: Research on active target discrimination in shallow-water environments is a critical application in underwater acoustics. Theoretical analysis and numerical simulations suggest that the phase of the ratio derived from complex pressure echoes as recorded by two hydrophones (same-depth, different ranges) shows a certain correlation with target depth [1]. A phase of complex pressure ratio (PCPR) matching method has been proposed to discriminate target depths, whose performance is greatly dependent on thermocline gradients of sound-speed profile and signal-to-noise ratio (SNR). To mitigate these dependencies while preserving discrimination accuracy, a deep learning framework is proposed for target depth discrimination based on a modified residual neural network (ResNet). The model to discriminate target depth is pre-trained on extensive data from a typical shallow-water environment with negative-thermocline sound-velocity profile. The input of the model is the feature matrices of PCPR calculated from the output of KrakenC[2]. Numerical simulation results demonstrate that the proposed method achieves higher accuracy than the PCPR matching method, especially under varying thermocline gradients and low SNR conditions. Preliminary results obtained from experimental data suggest that the proposed model is effective.\n\n[1] Li Qianyan, Wang Wenbo et al. Resolution of Surface and Underwater Targets Based on a Horizontal Array in a Shallow Sea Thermocline Environment. ACTA ACUSTICA. (Received in December 2024 but not yet published)\n[2] HLS Research. Acoustics Toolbox. Available online: http://oalib.hlsresearch.com/AcousticsToolbox/ (accessed on14 October2023).\n
  • Corresponding author: Ms Qianyan Li
    Affiliation: Key Laboratory of Underwater Acoustic Environment, Institute of Acoustics, Chinese Academy of Sciences Beijing
    Country: China

Back