Abstract: To address industry pain points such as insufficient identification of bunker blockage patterns, high false alarm rates, and a lack of proactive control capabilities, an intelligent coal bunker blockage clearing and management system was developed based on an integrated “Perception—Decision—Execution—Optimization” control strategy. By leveraging technologies including distributed intelligent flow promotion, multi-source data fusion, edge computing, and Long Short-Term Memory (LSTM) networks for intelligent diagnosis, a four-layer architecture—comprising the perception, platform, execution, and application layers—was constructed. Research was conducted on refined modeling, intelligent diagnosis of blockage trends, and closed-loop clearing strategy optimization, with subsequent engineering application at the Panji Coal Preparation Plant. The results demonstrate that the system achieves 10 ms-level synchronous timing of multi-source data and can preemptively identify blockage hazards such as “bridging” and “rat-holing.” Compressed air consumption was reduced by 30% to 50%. Application data reveals that the loading efficiency for a single train reached 4.33 times the pre-implementation rate, manual entry into bunkers for clearing was eliminated, and annual transportation costs were reduced by approximately 210000 yuan. This technological framework breaks through the limitations of traditional systems—namely single-source perception and passive disposal—transitioning coal bunker management from reactive emergency response to proactive prevention. It balances production efficiency, intrinsic safety, and energy conservation, providing critical technical support for the intelligent upgrading of coal preparation plants.