Short-circuit current (SCC) over-limit is becoming increasingly common, seriously troubling the development of power system operation methods. Currently, limiting the SCC is based on the predicted day-ahead worst-case scenario, ensuring safety at the cost of the economy. The decision is reasonable when the start-up mode is more certain. However, with the large-scale connection of renewable energy sources, the uncertainty of their output leads to the worst-case scenario may be a very small probability scenario far from the normal one, and the decision to SCC limiting measures based on this will seriously affect the economics. With the rise of artificial intelligence technology, it has become possible to predict and limit SCC urgently in real-time. Therefore, it is possible to develop more economical and safe flow limiting strategies by handling high probability scenarios in advance and urgently handling low probability scenarios without any omissions within the day. To sum up, this paper proposes a day-ahead and intra-day two-stage SCC limiting measures optimization framework based on scenario probability. Based on the principle of the lowest combined cost expectation of day-ahead and intra-day, we divide day-ahead and intra-day treatment scenarios and formulate day-ahead current limiting measures. For day-ahead emergency current limiting, we propose a real-time prediction method of SCC based on the combined data-driven and model-driven methods to discover and treat very small probability over-limit SCC scenarios. The results show that the proposed two-stage optimization framework has better feasibility, security, and economy.