Abstract: Due to the high dimensions, strong nonlinearity, and high randomness of the "dual high" power system, existing modeling, and stability analysis methods are limiting in scale, difficult to solve, and low in accuracy. This paper proposes an analytical framework for "dual high" power systems that includes nonlinear reduced-order modeling and domain of attraction optimization calculation estimation. First, the block-decoupling-order-reduced model of grid-connected DC microgrid is established based on the quadratic error model. And then, the system is completely transformed into the form of first-order quadratic differential equations. Secondly, an optimized calculation method is presented to contrast the optimal estimated region of attraction (OEROA), where the Kronecker product property is utilized in conjunction with the idea of constructing an optimization model using auxiliary variables based on Lyapunov's theory. Finally, a microgrid system is taken as an example under distributed photovoltaic cloud cover and continuous power load disturbance. Compared with Lyapunov's method based on the LaSalle theorem, and the T-S fuzzy model method, it is proven that the estimated attraction domain constructed by the proposed method has a lower conservatism, and the proposed analysis framework is valid.