With the integration of a high percentage of PV into the distribution network

the stochasticity of PV output

computational efficiency

and control flexibility requirements pose a great challenge for centralized reactive power optimization in the distribution network. To this end

a reactive power optimization method is proposed for a high percentage of PV distribution networks

considering the day-ahead random and intraday rolling distribution. A day-ahead random optimization model is established based on the scenario method

which is transformed into a solvable form by combining second-order cone relaxation and the large M method. Further

a real-time rolling distributed reactive power optimization model is established in the intra-day optimization stage. Considering the zoning dynamics

a comprehensive index system including the degree of intra-zone-interzone coupling

the power balance

and the number of nodes in the zone is proposed and solved based on Cuckoo's Search Algorithm to obtain the zoning scheme of the distribution network. The RSMprop-SADMM algorithm is also proposed for parallel distributed computation to improve the convergence speed and adaptability of the algorithm. Finally

the simulation results based on the improved IEEE69 node verify the effectiveness of the method proposed in this paper.