Dendritic Net Driven Quadratic Day-Ahead Voltage Control for Power System With Distributed Generation

Abstract

To quickly suppress the rapid voltage fluctuations caused by distributed generations (DGs), multi-time scale voltage control (MTSVC) has been proposed recently to achieve coordinated control of discrete and continuous devices. As one part of MTSVC, day-ahead voltage control (DAVC) mainly formulates the next-day 24-hour control strategy for discrete devices. However, due to the non-convexity of power flow (PF) constraints, uncertainty of DGs, mixed-integer nature of control variables, and daily action constraint of discrete devices, DAVC is indeed a large-scale mixed-integer non-convex nonlinear and stochastic optimization problem. This article proposes Dendritic Net (DN) driven quadratic DAVC, which adopts DN to simplify the modeling and computation of DAVC significantly. In the discrete variable relaxation stage, with DN's approximation property inherited from Taylor expansion, the PF constraints and robust optimization constraints are simplified into quadratic ones, which transforms DAVC into an easily-solved quadratic programming problem. Combined with Gurobi solver being able to solve any quadratic programming, the optimality and robustness of relaxation stage strategy can be guaranteed. In the discretization stage, based on the optimal relaxed strategy, a discretization model is established to quickly complete the formulation of final DAVC strategy. Test results based on modified IEEE 30-bus and 118-bus systems prove the correctness and rapidity of the proposed method.