Abstract: To address the challenges in coal preparation plants, including the difficulty of integrating multi-source heterogeneous data, the low efficiency of traditional manual data management, and significant data silos, a production data integration system was developed based on Python and Selenium to enhance production management efficiency and data application value. The system adopts a four-layer architecture encompassing data acquisition, processing, storage, and application. It collects both directly displayed webpage data and data displayed within ActiveX controls through automated login, web scraping, and screenshot recognition. The acquired data then undergoes time-series processing, parsing, deduplication, and OCR (Optical Character Recognition) identification to achieve structured storage. This system implements functionalities such as segmented production data comparison, data alerts and anomaly alarms, visual analysis of time-series production data, and cross-system automatic data pushing. This technical solution has successfully achieved the integration and preliminary analytical application of heterogeneous industrial data across different systems, effectively resolving the problem of data silos inherent in traditional production management systems. It provides a feasible path for data integration in legacy information systems and serves as a practical example for the coal preparation industry in deep data mining, collaborative application of industrial knowledge, and digital transformation.