Research On Multi-time-scale Reactive Power Optimization Method Of Power Grid With Photovoltaic Power Generation

The integration of photovoltaics into the power grid increases the complexity of the grid reactive power optimization problem.Aiming at the above problems,this paper proposes a new multi-time scale reactive power optimization method.Recently,a time division model was established to solve the reactive power output of discrete reactive power compensation equipment,and a day and night differential reactive power allocation strategy was established to solve the reactive p ower output of continuous reactive power compensation equipment.First,a day-ahead optimization method is proposed to divide time periods according to the grid voltage.Determine the dominant node of the power grid according to the sensitivity matrix;solve the voltage prediction value of the dominant node;divide the time period corresponding to the voltage prediction value based on the statistical data of variance and standard deviation;use the particle swarm optimization algorithm to solve the optimal output of the reactive power source.Then,a simplified cosφ(P,U) control is proposed.This paper analyzes the constant Q control,constant cosφ control,cosφ(P) control,Q(U)control and cosφ(P,U)control five methods respectively.On the basis of five methods,a cosφ(P,U)simplified control is proposed.In order to meet the grid voltage’s demand for reactive power,it does not violate the restriction of the active output of the photovoltaic array on the reactive output.Finally,this paper proposes a diurnal differential reactive power allocation strategy.During the day,the photovoltaic system outputs reactive power according to the simplified cosφ(P,U) control.At night,the photovoltaic inverter and SVG output reactive power in proportion to the reactive power margin.Based on the model predictive control theory,an intraday optimization model is established that considers the photovoltaic diurnal differential reactive power allocation strategy.Using the Deep Q-learning(DQN)strategy in reinforcement learning,the neural network model is trained by data-driven,and it can quickly respond to the power flow fluctuation in the power grid.Calculate the optimal reactive power out put to ensure the safety of the power grid and improve the robustness of the power grid.The IEEE33 numerical example simulation verification shows that the proposed method is effective.