在追求清洁、绿色与可持续发展的今天，水溶液及其水合物在海洋热能转换、绿色城市建设和碳捕获、利用与封存等关键技术中扮演着核心角色。热导率作为其中基本的物理性质，对这些领域的发展至关重要。然而，不同水溶液和笼状水合物的热导率变化复杂，基础数据不足，且对其物理机制的理解和传热行为的预测充满挑战。本研究采用分子动力学(MD)模拟，旨在深入探究这些水衍生体系的热导性能，以期推动科学认知并促进环境可持续性。首先，采用经典MD方法，模拟了不同盐水溶液的热导随溶质质量分数增加的变化，成功再现了实验测定的变化趋势。同时，研究表明，在单氢键位点的盐溶液体系中，热导率随着盐质量分数的增加而下降，源于单原子离子破坏了水分子间的有序结构，导致氢键数量和寿命不断减少，降低了热传导效率。而在多氢键位点的含氧酸根盐溶液体系中，热导率则表现出非单调变化特性。这种现象可以归因于非单调的氢键网络变化与含氧酸根稳定的溶剂化结构。其次，针对可反应的溶剂化的质子体系，开发了适用于盐酸的机器学习势，并运用MD模拟计算了不同质量分数盐酸溶液的热导率。结果在数值和趋势上均与实验测定数据高度一致，展示了方法的准确性。通过深入探讨热导率降低的趋势，氯离子之间的更紧密的局部结构，导致水的结构更加无序以及氢键数量和寿命下降。以及溶剂化质子频繁和更慢的质子迁移。最终导致热导降低。最后，针对笼状水合物模拟中定量精度不足和忽略原子核量子效应的问题，开发了适用于甲烷和二氧化碳笼状水合物的机器学习势，并通过路径积分MD模拟了这两种水合物在低温区间的热导率。模拟结果成功捕捉到实验中相同温度区间的热导变化趋势。结果显示，水分子的笼结构在热传导中起主导作用，尽管水分子和客体分子之间的相互作用贡献较小，但对导热性的影响不可忽视。此外，本文对甲烷分子进行粗粒化处理。研究表明，客体分子的形状、质量和旋转动力学是影响其导热行为的关键因素。总结而言，本研究通过MD模拟和深度学习技术，成功模拟不同水衍生体系热导率的变化规律，并理清了不同溶质和客体分子对水、冰体系传热性质的影响。这些发现可为能源转换、存储材料的设计与优化提供指导，并为海底可燃冰开采以及海洋环境等问题的模拟提供基本理化数据。

In the contemporary quest for clean, green and sustainable development, aqueous solutions and their hydrates are pivotal in advanced technologies such as ocean thermal energy conversion, sustainable urban construction, and carbon capture, utilization and storage. The thermal conductivity of these substances, as a fundamental physical property, is essential for the progression of these domains. Nevertheless, the variability in the thermal conductivity among different aqueous solutions and caged hydrates is intricate, with a paucity of fundamental data, and the comprehension of their physical mechanisms and prediction of heat-transfer behavior remain formidable challenges. Using molecular dynamics (MD) simulations, this study seeks to elucidate the thermal conductivity of these water-derived systems, thereby contributing to technological advancement and environmental sustainability.Initially, using the classical MD method, the thermal conductivity of salt solutions is simulated as a function of the increasing solute mass fraction, successfully replicating the experimentally observed trends. Concurrently, the study elucidates that in salt solutions of monatomic anions with a single H-bond site, the thermal conductivity decreases with increasing salt mass fraction. This reduction is attributed to monatomic anions that disrupt the ordered structure of water molecules, thereby continuously decreasing the number and lifetime of H-Bonds, which in turn decreases the heat conduction efficiency. In contrast, the thermal conductivity in aqueous solutions of oxyanions salts with multiple H-bond sites exhibits nonmonotonic behavior. This phenomenon can be attributed to the nonmonotonic variation in the network of H-Bonds and the stable solvation structure of oxyacids.Subsequently, a machine learning potential is formulated for the hydrochloric acid system for the solvated proton system, which is a reactive system. The thermal conductivity of hydrochloric acid solutions with varying mass fractions is computed via MD simulations. The findings exhibit remarkable concordance with the experimental data in both magnitude and trend, underscoring the precision of the DP model. An in-depth analysis of the decreasing thermal conductivity trend reveals that as the concentration of chloride ions increases, closer local structure between chloride ions leads to a more disordered structure of water, a reduction in the number and lifetime of H-Bonds, and more frequent and slower proton migration of solvated protons. These factors ultimately result in a reduction in the thermal conductivity.Finally, to address the challenges of inadequate quantitative precision in the simulation of clathrate hydrates and the neglect of quantum nuclear effects at low temperatures, a DP for methane and carbon dioxide clathrate hydrates has been developed. The thermal conductivity of these hydrates at low temperatures is simulated using path-integral MD. The simulation results successfully replicate the experimental trends in thermal conductivity within the same temperature range. The findings indicate that the cage structure of the water molecules predominantly governs heat conduction. Although the interaction between water molecules and guest molecules is minimal, its impact on thermal conductivity remains significant. In addition, methane molecules are modeled with coarse granulation. The results elucidate that the shape, mass, and rotational dynamics of guest molecules are critical determinants of their thermal conductivity behavior.In summary, MD simulations and deep learning methodologies are employed in this study to effectively model thermal conductance variation principles in various water-derived systems. Furthermore, the impact of solute and guest molecules on the thermal transfer properties of aqueous and ice systems is elucidated. These insights offer valuable guidance for the design and optimization of energy conversion and storage materials and provide essential physicochemical data for the simulation of seabed combustible ice extraction and marine environmental studies.