The energy crisis and environmental pollution are worsening. Therefore, water-based hydraulic fluids, i.e., aqueous ethylene glycol-based, fire-retardant hydraulic fluid concentrates (HFCs), are becoming increasingly common. However, seawater intrusion inevitably occurs under marine conditions, generating hazards, such as corrosion and friction, within hydraulic system components, pipelines, and materials. Moreover, HFCs have several drawbacks, including low viscosity, inadequate lubrication, and high corrosivity. Therefore, the tribological characteristics and corrosivity of HFCs must be improved and reduced, respectively. This can be achieved using additives. Herein, we summarize the fundamental characteristics of HFCs and their modifications for use in the marine environment, focusing on the optimal water–ethylene glycol proportion and its influence on the physicochemical, lubricating, and tribological properties of this HFC under varying conditions. We discuss the latest progress on the effects of seawater on the tribological corrosion of HFCs and the reduction in corrosivity in the presence of different additives. Finally, we highlight challenges and propose future research to improve performance in the marine environment.

Machine learning (ML) is transforming material research and development (R&D), driving a fundamental shift from experience-driven approaches to data-driven frameworks. This review systematically highlights the transformative breakthroughs brought by machine learning throughout the entire process of intelligent material innovation. And it provides a comprehensive full chain analysis, from atomic scale design to macroscopic applications, emphasizing multi-scale modeling that combines physical mechanisms with data-driven methods, running through all stages of material innovation. In the design phase, ML promotes performance-oriented structural optimization through inverse design systems and generative models. For synthesis and processing, closed-loop autonomous systems and green controllable synthesis strategies significantly improve efficiency and sustainability. In terms of advanced representation, ML-powered techniques can help proactively tackle key challenges of complex structures. Performance prediction models enable precise correlations between material properties and extreme properties (such as auxiliary structures) by revealing catalytic descriptors and decoding biological interface mechanisms. Ultimately, these ML-driven advancements are unlocking practical applications in key fields, such as energy, biomedicine, environmental remediation, and structural engineering. This article aims to provide a comprehensive technological roadmap for the next generation of smart material development by integrating cross scale insights and autonomous strategies, and to outline future directions for this rapidly developing paradigm.