In order to solve the problem of reduced capacity caused by airspace congestion, a capacity optimization method considering the complexity of intersections is proposed. First, it analyzes the pain points and difficulties in actual civil aviation operations, as well as the lack of cross-point modeling, the difficulty of controller load measurement, and the high complexity of solving time in the previous related research on capacity. Second, an abstract method of airspace mathematics is proposed from the three aspects of altitude, intersections and flight operations, and the mathematical description of the airspace is simplified according to the number of intersections of the nodes, eliminating the navigation stations that do not cross, and reducing the airspace. The problem of sparse node matrix after the transportation network is abstracted. Thirdly, it analyzes that the impact of intersections on capacity is mainly in two aspects: flow and number of intersections, and the cost function of intersections is established from these two aspects. Fourth, with the goal of minimizing delays, the traffic balance, sector and route capacity, flow control capacity, and integer non-negative constraints are used to establish a capacity optimization model. Fifth, analyze and point out that air traffic networks with negative tolerances are more prone to delays, and based on the characteristics of network delays, an iterative algorithm considering the back propagation of delays is proposed. Finally, taking the airspace in North China as an example, simulations are carried out from three aspects: the delay time under different flow control levels, the number of affected flights, and the calculation time of the algorithm. The results show that the model and algorithm can reduce delays by 33.58% on average, and reduce delays to the greatest extent through reasonable allocation of diversion, timing and reduction.