Abstract Ice loads are important environmental loads that can influence the structural safety of ships during navigation in ice-covered waters. The identification of ice loads on ship hulls is the core of ice load monitoring. In this study, a new ice load identification model based on Green kernel and regularization methods is established. First, the forward model for ice load identification is developed through the discretised convolution integral of ice loads. Next, three commonly used regularization methods, including Tikhonov, truncated singular value decomposition, and least square QR-factorization (LSQR) are adopted to reduce solution errors. The LSQR method is thereafter selected as the optimal regularization operator, and its regular property is proved by numerical cases with ice-induced strains that contain noise. Finally, two load identification cases are conducted on an experimental rig to evaluate the feasibility of the model in ice load identification. The identified loads can determine the signal features of applied loads in the time domain with good accuracy. This identification model provides new insights for full-scale ice load monitoring.

中文翻译：

---

用绿核和正则化方法识别冰船壳结构的冰载荷

摘要 冰载荷是重要的环境载荷，会影响船舶在结冰水域航行时的结构安全。船体冰载荷的识别是冰载荷监测的核心。本研究建立了一种基于格林核和正则化方法的新的冰负荷识别模型。首先，冰载荷识别的正向模型是通过冰载荷的离散卷积积分开发的。接下来，采用三种常用的正则化方法，包括 Tikhonov、截断奇异值分解和最小二乘 QR 分解 (LSQR)，以减少求解误差。此后选择LSQR方法作为最佳正则化算子，并通过包含噪声的冰致应变的数值案例证明了其规则性。最后，在实验台上进行了两个载荷识别案例，以评估模型在冰载荷识别中的可行性。识别出的载荷可以准确地确定时域中施加载荷的信号特征。这种识别模型为全面冰负荷监测提供了新的见解。