Fracture networks, fluid flow and heat extraction within fractures constitute pivotal aspects of enhanced geothermal system advancement. Conventional hydraulic fracturing in dry hot rock reservoirs typically requires high breakdown pressure and only produces a single major fracture morphology. Thus, it is imperative to explore better fracturing methods and consider more reasonable coupling mechanisms to improve the prediction efficiency. Cyclic fracturing using liquid nitrogen instead of water can generate more complex fracture networks and improve the fracturing performance. The simulation of fluid flow and heat transfer processes in the fracture network is crucial for an enhanced geothermal system, which requires a more comprehensive coupled thermo-hydro-mechanical-chemical model for matching, especially the characterization of coupling mechanism between the chemical and mechanical field. Based on the results of field engineering, laboratory experiments and numerical simulation, the optimum engineering scheme can be obtained by a multi-objective optimization and decision-making method. Furthermore, combining it with the deep-learning-based proxy model to achieve dynamic optimization with time is a meaningful future research direction.