In view of the randomness and controllability of wind power and energy storage, reactive power coordinated optimization is difficult. A scene analysis method based on time domain probability is proposed, using CURE clustering method to reduce the total scene number, and calculate the time domain probability of each scene at the same time. A coordinated reactive power optimization method which considering source, network, load, storage in active network is proposed. The active power optimization model based on the minimum power consumption of main online shopping is used to solve energy storage and interruptible load active power. On this basis, a coordinated reactive power optimization model is established. By using the multi-objective particle swarm algorithm, a 10 kV distribution line in a prefecture level city is finally taken as an example to verify the feasibility and robustness of the proposed method.