Abstract: The exploitation of deep coal resources serves as a critical guarantee for China’s energy security. However, as mining depths increase, dynamic disasters induced by the evolution of overlying strata structure are becoming increasingly prominent, emerging as one of the core challenges that constrain safe and efficient mining operations. This paper systematically reviews the research progress on the multi-scale evolutionary characteristics, cross-scale coupling mechanisms, typical catastrophic modes, and associated prevention and control technologies of deep mining overlying strata structures, following the main thread of structural evolution–disaster mechanism-active prevention and control-intelligent prevention and control. Furthermore, a novel prevention and control paradigm oriented towards intelligent decision-making is proposed. Firstly, from the dual perspectives of the engineering system scale and the engineering structure scale, the spatiotemporal evolution law of overlying strata structures characterized by “high-low level linkage and static-dynamic load coupling” is revealed. The cross-scale coupling mechanism, dominated by the fracture of high-level key strata at the engineering system scale and responding to the damage zone at the engineering structure scale, is elucidated. Furthermore, a dynamic analysis framework encompassing three levels: “engineering system, engineering structure, and material damage” is established. Secondly, the mechanical essence of structural instability-induced disasters is summarized as a chain process of “energy driving-damage accumulation-instability criterion”. Three typical catastrophic modes, namely the dynamic load triggering type, the static load mutation type, and the composite progressive type, along with their mutual transformation laws, are analyzed. The limitations of existing single-scale, purely mechanical instability criteria in characterizing cross-scale coupling and multi-field effects are pointed out. Based on this foundation, the active prevention and control technology system, centered on source pressure relief, macro-structure regulation, and strengthening of the surrounding rock bearing structure, is systematically summarized. This system encompasses key technologies such as active weakening of overlying strata structures, optimization of mining processes, local stress field regulation, and high-prestress energy-absorbing support, while identifying critical bottlenecks in terms of mechanism transparency, multi-field sensing integration, and regulation synergy. Furthermore, addressing the complex conditions of deep mining, a new paradigm of integrated intelligent prevention and control: “perception-analysis-decision-control” is proposed, with a focus on technical pathways including transparent perception based on multi-source information fusion, dynamic deduction using physics-informed digital twins, and adaptive regulation through deep reinforcement learning. Taking the deep 130607 working face of the Maiduoshan Coal Mine as the engineering background, a collaborative prevention and control system integrating directional hydraulic fracturing of high-level overburden, large-diameter borehole pressure relief in coal seams, and multi-source data fusion-based intelligent early warning is constructed. Engineering practice has validated the effectiveness of this intelligent prevention and control system. This paper aims to provide a systematic theoretical reference and technical guidance for the precise prevention and intelligent decision-making of dynamic disasters in deep mining overlying strata structures.