Abstract: Due to the harsh environment of coal mine heading face，the intelligent control in roadway or at remote sur⁃ face depends on the interactive collaboration of operators，equipment，roadway，and surrounding in cyber⁃physical space. The digital twin of heading face is important for integrating virtual excavation and physical excavation. Consid⁃ ering the demands of the virtual⁃real interactive control，a digital twin model of intelligent heading face is construc⁃ ted to achieve a high⁃fidelity description for the collaborative operation process. It is done by analyzing the static char⁃ acteristics，dynamic characteristics，behavior rules，and the interaction of heterogeneous elements in the excavation face，which consists of operator agent，equipment agent，the agents of the roadway and its surrounding. The control state discretization and event⁃driven mechanism are introduced to construct the discrete logic model of the equipment agent，and the five modeling steps of potential event monitoring，event extraction，state discretization，state evolution dy⁃ namics modeling， and event⁃driven state jumping are elaborated. The state transition mechanism of single⁃agent and multi⁃agent in the parallel operation process is explored to realize the integration of the geo⁃metric⁃physical model and discrete logical model of the heading face，which is a high⁃fidelity virtual model of the heading face in digital space. Furthermore，the parallel control theory is introduced to construct a parallel intelligent control architecture of the heading face based on its digital twin and the physical system. This architecture deconstructs the digital twin into the description subsystem，prediction subsystem，and guidance subsystem for integrating digital twins with the nu⁃ merical computational models of visual positioning，collision detection，and excavation control. The safety boundary of optimal control is determined by virtual simulation technology to predict the trend of state changes in the heading face and assist physical excavation control， which ultimately achieves the requirements of parallel intelligent control between the digital twin and the physical entity. Taking the trajectory control of the roadheader’s cutting head as an example，the visual⁃feedback⁃based physical control logic is compared with the digital⁃physical⁃fusion⁃based parallel intelligent control logic. Moreover，the mechanism of intelligent remote control state transition for the shape⁃cutting of heading face is formed， which improves the ability to control all elements at the heading face， and the pro⁃ posed method provides a new implementation path for the intelligent control in roadway or at remote surface.