标题： 基于物理信息的机器学习用于植物基食品微尺度干燥：计算模型和实验见解的系统综述 显示英文标题

标题： Physics-Informed Machine Learning for Microscale Drying of Plant-Based Foods: A Systematic Review of Computational Models and Experimental Insights

摘要： 本综述考察了植物基食品材料（PBFM）干燥过程中微尺度细胞变化的当前研究状况，特别强调计算建模方法。 综述解决了在微尺度研究中对先进计算方法的关键需求。 我们系统分析了PBFM干燥的实验研究，突出它们在捕捉细胞水平现象方面的贡献和局限性，包括在不同干燥条件下数据获取和测量精度的挑战。 对用于微观结构研究的计算模型的发展进行了深入探讨，从传统的数值方法到当代最先进的方法，特别关注它们处理植物细胞材料复杂非线性特性的能力。 特别关注数据驱动模型的出现及其在预测PBFM干燥过程中微尺度细胞行为方面的局限性，特别是数据集获取和模型泛化方面的挑战。 综述深入分析了物理信息机器学习（PIML）框架，探讨了其理论基础、相关领域的当前应用以及将物理原理与神经网络架构结合的独特优势。 通过这一全面评估，我们确定了现有方法中的关键差距，评估了不同建模方法之间的权衡，并提供了改进对PBFM干燥过程中细胞水平转变理解的未来研究方向的见解。 综述最后提出了将实验和计算方法相结合的建议，以推动食品保鲜技术领域的发展。 显示英文摘要

摘要： This review examines the current state of research on microscale cellular changes during the drying of plant-based food materials (PBFM), with particular emphasis on computational modelling approaches. The review addresses the critical need for advanced computational methods in microscale investigations. We systematically analyse experimental studies in PBFM drying, highlighting their contributions and limitations in capturing cellular-level phenomena, including challenges in data acquisition and measurement accuracy under varying drying conditions. The evolution of computational models for microstructural investigations is thoroughly examined, from traditional numerical methods to contemporary state-of-the-art approaches, with specific focus on their ability to handle the complex, nonlinear properties of plant cellular materials. Special attention is given to the emergence of data-driven models and their limitations in predicting microscale cellular behaviour during PBFM drying, particularly addressing challenges in dataset acquisition and model generalization. The review provides an in-depth analysis of Physics-Informed Machine Learning (PIML) frameworks, examining their theoretical foundations, current applications in related fields, and unique advantages in combining physical principles with neural network architectures. Through this comprehensive assessment, we identify critical gaps in existing methodologies, evaluate the trade-offs between different modelling approaches, and provide insights into future research directions for improving our understanding of cellular-level transformations during PBFM drying processes. The review concludes with recommendations for integrating experimental and computational approaches to advance the field of food preservation technology.