Substation Equipment Fault Diagnosis Based On A Multimodal Large Model

This paper introduces an intelligent fault diagnosis framework grounded in a multimodal large model, designed to meet the intricate fault detection needs of substation equipment. This paper incorporates the LLaVA model, LoRA fine-tuning methodology, and Rag knowledge base-augmented retrieval, offering precise fault identification for substation equipment alongside comprehensive cause analysis and practical handling suggestions. Experimental comparisons validate the effectiveness of Rag knowledge base enhancement and LoRA fine-tuning in enhancing diagnostic accuracy. This paper underscores the potential of multimodal large models in industrial smart maintenance, offering a technical foundation for the intelligent advancement of substation equipment fault diagnosis systems.