Identification of ice loads on ship structure using a hybrid regularization strategy.

Area of Science:

• Naval Architecture and Marine Engineering
• Applied Mechanics
• Signal Processing

Background:

• Accurate ice load identification is vital for polar vessel structural safety.
• Existing regularization methods struggle to balance impact peaks and background load reconstruction.
• Ill-posed problems in ice load identification are sensitive to limited data and noise.

Purpose of the Study:

• To develop an improved ice load identification method for polar vessels.
• To address the limitations of current regularization strategies in capturing both transient and background ice loads.
• To enhance the physical plausibility and accuracy of ice load identification.

Main Methods:

• Formulated an inverse problem model using Green's kernel functions for dynamic mapping.
• Introduced a hybrid regularization model combining L1 and L2 norms.
• Employed a coordinate descent algorithm for optimization and Bayesian Information Criterion (BIC) for parameter selection.

Main Results:

• The hybrid regularization strategy accurately identifies peak ice loads and reconstructs low-amplitude background loads.
• The method demonstrates enhanced accuracy, mitigates signal truncation, and shows balanced performance.
• Validated through high-fidelity numerical simulations and scaled model experiments, showing robustness in multi-source impact scenarios.

Conclusions:

• The proposed hybrid regularization method offers an effective solution for ice load identification in polar vessels.
• The approach improves accuracy and robustness, crucial for ensuring the structural integrity of ships operating in icy conditions.
• This method provides a more physically plausible and reliable means of assessing ice loads.