Multi-modal knowledge graph is an emerging research hotspot in the field of artificial intelligence in recent years. It builds entities of multiple modalities and their semantic relationships between multimodal entities based on the traditional knowledge graph, which can provide important textual, visual, and audio knowledge. This paper proposes a method for constructing multimodal knowledge graph of electric power based on transfer learning. By establishing a multimodal knowledge graph framework based on model transfer learning, it constructs an entity discovery and relationship extraction model for multimodal data, realizes feature-based alignment of multimodal entities of typical electric power business data elements, and efficiently integrates large-scale multi-source and multimodal knowledge graphs. Finally, it completes the construction of a multimodal knowledge graph for the typical electric power business field and applies this method to marketing, safety monitoring, and other typical electric power businesses.