Abstract: To overcome the bottlenecks of high labor intensity, high operating costs, and low intelligence levels in the manual separating process of the raw coal refuse discharging system, Jiangzhuang Coal Mine implemented a full-process intelligent reconstruction of its “shallow-depth heavy medium bath & manual reverse separating” process (in operation since 2014). This was achieved through the coupling and integration of control technologies: “50～200 mm intelligent dry separating, >200 mm LIDAR separating, and >50 mm total refuse crushing.” The study utilizes a three-stage roller screen classification as the starting point. Raw coal blocks of 50～200 mm are directed into an X-ray & vision dual-mode intelligent dry separating system, while blocks >200 mm are identified by 3D LIDAR and diverted at high speed via guide plates. All refuse >50 mm is reduced to <80 mm by a double-toothed roll crusher and transported via an enclosed pipe belt conveyor. Simultaneously, a centralized control platform was embedded to integrate data from seven stages—feeding, screening, crushing, separating, transportation, dust extraction, and video monitoring—into the dispatch center, enabling “one-click start-stop, remote linkage, and fault self-diagnosis.” Production practice demonstrates that the labor efficiency of the raw coal system surged from 320.51 t/shift to 757.57 t/shift, with an annual cost saving of 4.1706 million RMB. Manual separating and mine-car refuse discharge have been eliminated, resulting in zero minor injuries or major accidents for two consecutive years. The system features wet-dry compatibility, allowing a switch to the backup shallow-depth system within 5 minutes to ensure continuous production. This practice establishes an integrated “classification-identification-crushing-centralized control” refuse discharging process. It constructs a demonstration model for aging mines in their resource-depletion stage to reduce manpower through technology, ensure safety via intelligence, and cut costs through intensification, providing a replicable and scalable low-cost intelligent upgrade path for similar coking coal preparation plants.