<p>This study presents a new coal mining technique that uses top-cut unloading to induce the fragmentation and expansion of collapsed ore. Stress compensation and control are achieved by using a constant-resistance, large deformation anchor cable. The research employs laboratory physical model tests and an artificial neural network (ANN) time-series prediction model to monitor, analyze, and predict the movement and crack development of the overlying rock layers in tunnels. The results show that the support characteristics of constant-resistance, large-deformation anchor cables, combined with the filling effect of collapsed ore fragmentation and expansion, automatically form tunnel supports. This new mining method significantly reduces the displacement of the overlying rock layers and minimizes surface subsidence. Furthermore, the ANN time-series prediction model accurately forecasts displacement at multiple monitoring points in the overlying rock mass and the axial force of constant-resistance, large deformation anchors. This innovative mining method provides valuable insights for ecological protection and exploration of underground works.</p>