A Data-Driven Method for Improving Voltage Quality of Large-Scale Distributed PV in Distribution Network

With the high-proportioned distributed PV supply connected to the distribution network, the voltage quality problem is particularly prominent. Firstly, a reactive voltage optimization model based on distributed PV reactive voltage regulation capability is established to improve the voltage quality by using PV reactive voltage regulation capability. At the same time, considering that the VVO model is nonlinear non-convex optimization model, the direct solution of the heuristic algorithm takes a long time, which is not conducive to the online VVO. In this regard, the SA algorithm is applied to obtain the distributed PV optimal output data set, and then the XGBoost model is trained according to the distributed PV optimal output data set, and the VVO of the distribution network is realized from the data-driven perspective, which significantly improves the solving speed of the distribution network optimization model. Finally, a typical distribution network example of IEEE 33 nodes is used to analyze the rationality of the proposed method.