<p>As intelligent coal mining technology advances, achieving efficient, safe, and intelligent mining in complex environments remains a critical industry goal. Traditional control systems struggle with limited perception, poor adaptability, and low equipment synergy, especially in deep mining with high ground stress and complex geology. To overcome these limitations, this paper proposes a human-machine-environment (man-machine-ring) collaborative control architecture and a dynamic reliability optimization method based on the Mine-Hong system. The proposed system integrates multi-dimensional perception, information fusion, and intelligent control to improve dynamic reliability and collaborative performance in mining operations. Key innovations include a high-precision positioning system combining coordinate transformation and multi-source fusion, which enhances perception accuracy in dust-obscured environments. Additionally, deep learning-based image enhancement and target detection algorithms are developed for better recognition of coal-rock boundaries. Dynamic monitoring is achieved by fusing inertial navigation, UWB positioning, and 3D laser scanning, enabling real-time tracking of spatial deformation, pressure-relief zones, and equipment status. A dynamic reliability assessment model is constructed to optimize coordinated control among hydraulic supports, shearers, and conveyors. Experimental results, conducted across varying mining distances and conditions, validate the effectiveness of the method. The system attained real-time surveillance, accurate recognition of hazard areas, and enhanced dynamic dependability. In comparison to traditional methods, the suggested technique showed improved sensing precision, quicker response times, and increased reliability at the system level. This study offers a practical, smart approach for the future of deep intelligent mining operations, tackling issues related to safety, efficiency, and autonomous management.</p>