Reasonable use of the flexible regulation characteristics of distributed power sources and adjustable loads is the key to realize active control of active distribution network, but the randomness and fluctuation of distributed power sources among them affect the reliability of real-time dispatching. To realize real-time optimal dispatching of active distribution network, an optimization strategy for real-time dispatching of active distribution network is proposed based on model prediction control and opportunity constraint. Firstly, the small time scale scheduling system of the active distribution network is analyzed. Based on this, the optimal economic dispatch is used as the optimization target, and the model predictive control combines the system future state sensing with real-time state feedback to perform rolling optimization of real-time dispatch and minimize the influence of distribution network uncertainty. Opportunity constraints are used in the rolling optimization to further reduce the influence of random fluctuations of distributed energy and adjustable loads, thus high reliability and high economy of active distribution network real-time dispatching are realized. Finally, the applicability and superiority of the proposed strategy are verified by a comprehensive operation example.