大语言模型赋能气候适应型电力系统的挑战与前景展望OA

[Objective]Recent advances in large language models(LLMs)have demonstrated breakthroughs in semantic understanding and reasoning-based generation,providing new technological pathways for multi-source information integration,complex scenario decision-making,and cross-agent coordination.However,a systematic understanding of the mechanisms and application frameworks of LLMs in climate-adaptive power systems remains lacking in both academia and industry.Against this background,the application potential and core issues of LLMs in enhancing the climate adaptability of power systems are systematically reviewed.[Methods]First,starting from the multi-scale impacts of climate change on power systems,the key capability constraints of power systems in information integration,decision generation,and coordinated interaction are analyzed.Second,a unified analytical framework of"information understanding-decision generation-coordinated interaction"is established,and the typical methodological paradigms of LLMs in semantic understanding,reasoning-based generation,and multi-agent coordination are systematically summarized.Furthermore,key challenges of LLMs in terms of output reliability,consistency with physical constraints,real-time responsiveness,data security,and adaptability to non-stationary environments are analyzed based on engineering application requirements.Finally,future research directions and implementation pathways for LLM-empowered power systems are envisioned to address system evolution requirements under climate uncertainty.[Conclusion]Facing the challenges of climate change,the deep empowerment of LLMs will drive power systems from passive response toward proactive adaptation,advancing toward a new paradigm of coordinated development characterized by security,low-carbon development,and high resilience.