As the global energy system transitions toward cleaner and low-carbon solutions, the thermal power industry faces dual challenges of improving efficiency and environmental protection. Industrial Internet technology, through integrating IoT, big data, and AI, provides crucial support for optimizing production processes, enabling intelligent equipment maintenance, and enhancing energy efficiency in thermal power plants. This paper focuses on the architecture design of industrial Internet platforms for thermal power plants, proposing a three-tier architecture model comprising edge layer, platform layer, and application layer. The edge layer integrates IoT protocol adaptation with edge computing technology to standardize access for multi-source heterogeneous devices and perform real-time data preprocessing, effectively addressing issues such as fragmented protocols and high latency in traditional thermal power plant equipment. The platform layer constructs a data middle platform and business middle platform using containerized microservices architecture, combining cloud computing and big data technologies to form a highly concurrent and available industrial PaaS platform that supports massive data storage, analysis, and service-oriented encapsulation. The application layer develops core modules including equipment health management, combustion optimization, and energy consumption analysis through digital twin technology and AI algorithms, establishing an intelligent decision support system that covers the entire production process.