The continuous increment of permeability of distributed photovoltaic(PV) power generation in distribution network leads to reduction of transmission power in feeders

which causes serious overvoltage conditions at the nodes and large network loss. Therefore

an intelligent voltage regulation strategy for photovoltaic power distribution network with PV output power short-term prediction is proposed. Firstly

a deep neural learning network(DNN) is applied to predict photovoltaic output power after the anomaly detection and similar day clustering. Secondly

in large-scale photovoltaic power generation

a controllable photovoltaic installation and grid connection point is selected based on the advanced electrical distance. Finally

the new NSGA-III is used to regulate the active and reactive power output of the controllable photovoltaic equipment to adjust the voltage of the distribution node and reduce the network loss. A regional distribution network in Anhui province is analyzed. The experimental result demonstrates that the proposed voltage regulation strategy is able to solve the problem of voltage over-limit finely and reduce the total network loss.