The intelligent mining sector suffers from a lack of big data support, and the application of foundational AI platforms and algorithms is limited. The main reason is the absence of an industrial internet architecture cloud platform tailored for the coal industry, which restricts resource integration. This leads to high failure rates of equipment on comprehensive mining workfaces, along with high maintenance and labor costs. By integrating cloud-native technology, edge computing, big data platforms, and artificial intelligence platforms, a flexible, modular industrial internet architecture remote maintenance platform has been developed to meet these challenges. The platform employs graph and document large model technologies and uses the Bert-BiLSTM-CRF model for knowledge extraction and knowledge graph construction, significantly improving the accuracy of knowledge extraction. The successful development of the platform has narrowed the gap with other energy sectors' industrial internet applications and laid a foundation for the realization of intelligent and unmanned coal mining in the coal industry.