Capsizing accidents are regarded as marine accidents with a high rate of casualties per accident. Approximately 89% of all such accidents involve small ships (vessels with gross tonnage of less than 10 tons). Stability calculations are critical for assessing the risk of capsizing incidents and evaluating a ship’s seaworthiness. Despite the high frequency of capsizing accidents involving them, small ships are generally exempt from adhering to stability regulations, thus remaining systemically exposed to the risk of capsizing. Moreover, the absence of essential design documents complicates direct ship stability calculations. This study utilizes hull form feature data—obtained from the general arrangement of small ships—as input for a deep learning model. The model is structured as a multilayer neural network and aims to infer hydrostatic curves, which are required data for stability calculations.

Capsizing accidents are regarded as marine accidents with a high rate of casualties per accident. Approximately 89% of all such accidents occur in small ships (vessels with gross tonnage less than 10 tons). Stability calculations are critical for assessing the risk of capsizing incidents and evaluating a ship's seaworthiness. Despite the high frequency of capsizing accidents involving small ships, they are generally exempt from adhering to stability regulations, thus remaining systemically exposed to the risk of capsizing. Moreover, the absence of essential design documents complicates direct ship stability calculations. This study utilizes hull form feature data—obtained from the general arrangement of small ships—as input for a deep learning model. The model is structured as a multilayer neural network and aims to infer hydrostatic curves, which is required data for stability calculations.

Cylindrical bodies generate flow-induced noise when exposed to external flows, which can be predicted numerically using Computational Fluid Dynamics (CFD) combined with the Ffowcs Williams–Hawkings (FW–H) Equation. Accurate prediction, however, requires turbulence models such as Detached Eddy Simulation (DES) with fine spatial resolution and small time steps, in addition to time-dependent surface pressure data and receiver arrangements. These requirements greatly increase computational costs and limit the applicability of such methods during the design stage. To address this challenge, a Deep Neural Network (DNN) model was developed to predict flow-induced noise around a cylinder. Training data were generated from CFD cases using cylinder geometry and inflow velocity as design variables, with multiple receivers arranged in a polar coordinate system. Acoustic signals were computed using Farassat’s Formulation 1A, the time-domain surface solution of the FW–H Equation. The DNN was trained with design variables, receiver coordinates, and octave-band center frequencies as inputs, while the Sound Pressure Level (SPL) served as the output. Model performance was evaluated using the adjusted coefficient of determination (${\mathrm{R}}_{\mathrm{adj}}^2$) and the root mean squared error (RMSE). In addition, interpolation capability was tested by varying receiver spacing to examine robustness under sparse data conditions. The results confirm that the proposed framework provides accurate and computationally efficient predictions suitable for early-stage design.