Abstract: In order to improve the peak regulation capacity of the load center at the receiving end and reduce the influence of source-load bilateral uncertainty on the cross-regional consumption and the dispatching economy of new energy, and considering the access of nuclear power and carbon capture power units, a two-stage robust optimization model for multi-source coordination of DC cross-regional interconnected power grid is proposed, which takes into account the source-load uncertainty. Firstly, the peak load regulation model of nuclear power was improved to promote the accuracy of load following. The characteristics of "peak shaving and valley filling" and power rapid adjustment of the carbon capture units were analyzed. Afterwards, the uncertain characteristics of the wind power and load were described by constructing the convex polyhedron uncertain sets. Then, according to the different scheduling characteristics of the power supply types, the start-up and shut-down costs of the units, the operation and peak regulation costs of the nuclear power were taken as the objective functions at the first stage. The generating unit costs, the wind abandonment penalty costs and the carbon trading costs were taken as the objective functions at the second stage. Meanwhile, the column and constraint generation(C & CG) algorithm was introduced to transform the optimization model into a master problem and a sub-problem to be solved iteratively. Finally, the validity and rationality of the model are verified by a cross-regional simulation example. The results show that the proposed coordinated optimization method can improve the new energy absorption capacity of the cross-regional power grid, reduce carbon emissions, and significantly increase the peak regulation margin of the receiving end grid. The scheduling results with different conservatism degrees can be obtained by changing the uncertainty budget parameters to balance the robustness and economy of the system.