Abstract: Implementing large language models in power systems is becoming increasingly prevalent with pervasive development. This paper presents a preliminary investigation into the applications of large language models in power systems, analyzing their technical advantages and practical value in areas such as prompt engineering, multimodal processing, and model fine-tuning. Through experiments, the study validates the potential and capabilities of large language models in supporting auxiliary operations within power systems and examines their performance across various tasks. The results demonstrate that the large language model can address power system issues such as power prediction, optimal power flow calculation, and specialized knowledge question and answer through the designed prompt engineering framework and model fine-tuning. This provides electrical engineers with an initial reference for utilizing large language models to solve power system problems while laying a solid foundation for broader future applications.