The “dual carbon” objectives impose elevated standards for the high-quality advancement of the energy storage sector, necessitating a comprehensive exploration and utilization of knowledge pertinent to this domain. There exists an urgent demand for precise knowledge recommendation services tailored to specific scenarios. This study proposes an innovative knowledge recommendation framework that synergizes large language models with knowledge graphs within the realm of energy storage: it judiciously considers both the accessibility and authority of patents within this field to ascertain data sources. By categorizing entities within the energy storage sector into eight types across two dimensions—metadata and text granularity—the rigorous experiments comparing various machine learning models to identify the most effective model for entity extraction tasks have been conducted. Leveraging entity relationship templates, the triples are generated to facilitate the construction of a knowledge graph specifically relevant to energy storage. Three sophisticated response strategies have been proposed: general question answering, patent recommendations, and knowledge combination suggestions. By harnessing the formidable semantic comprehension and generative prowess of large language models, the approach automates the extraction of pertinent information from the knowledge graph in response to user inquiries, thereby facilitating precise recommendations within specialized domains. Empirical research findings in superconducting energy storage reveal that the innovative approach surpasses traditional LLM-KB methods across four task categories: single-target recommendation, dual-target recommendation, knowledge combination recommendation, and hybrid recommendation. Specifically, compared with conventional LLM-KB approaches, enhancements of 20.47% and 16.57% were recorded in ROUGE-L and BLEU-4 metrics, respectively. This study integrates methodologies from both knowledge graphs and large language models to provide novel insights into accurate recommendations within the energy storage sector, while improving users efficiency and effectiveness in acquiring high-quality information.