Since wind power has strong stochastic volatility

the accommodation of lager-scale wind power requires thermal power to provide auxiliary services. However

thermal power units have high capacity but low flexibility in China

which severely restricts the accommodation of wind power. This paper develops a long-term generation scheduling model of power systems considering the flexibility reformation of thermal power units. Since this model is a large-scale optimization model with multiple coupling time dimensions

the Benders decomposition algorithm is used to decompose the original model into an investment master problem with regard to flexibility reformation

and a set of subproblems with regard to the operation in each scheduling time period under different reformation schemes. The efficient computation of large-scale optimization model with multiple time dimensions is then realized. The numerical example shows that the proposed model can improve the economy of power systems and prompt the accommodation of high-proportion wind power.