Research On Control Strategy Of Power System Reactive Power Based On Model Predictive Control

China's new energy base has the characteristics of "large-scale wind power and photovoltaic power generation access to sending power grid".The gap of its output leads to the voltage fluctuation problem far beyond the traditional power grid,which brings great pressure to the reactive voltage operation of the power system.At the same time,the development of distributed generation and micro-grid on the demand side makes the source-side and the load-side diversity increase,the uncertainty increases,and the operation scenario becomes more complicated.How to coordinate various reactive control devices to achieve the reactive power and voltage finely control is an urgent problem to be solved.At present,the online reactive power optimization of the power grid is mostly carried out by using the real-time data acquired by the EMS system.It belongs to the post-correction control,and the control strategy has poor timeliness,which is difficult to cope with the current situation of the grid operation with increasing uncertainty of the source-side and the load-side.Aiming at the above problems,this paper adopts the model predictive control theory,combined with the operating characteristics of the reactive power control equipment in the grid,and designs a reactive power optimization control architecture based on model predictive control.In order to solve the difficult problem of nonlinear prediction in bus reactive load prediction,a prediction method based on dual-input long short-term memory neural network is proposed.This method can be based on the depth-time characteristics of load active and reactive power sequence data mining and represents the nonlinear relationship.The example shows that the proposed method can accurately predict the bus reactive load,and its prediction accuracy is better than the time series and the general long short-term memory neural network prediction model.Based on the bus reactive load prediction model and the coordination control requirements of reactive devices under multi-time scales,this paper proposed a reactive power control strategy based on model predictive control.This method transforms the nonlinear constraint approximation in the reactive power control model into the second-order cone linear constraint condition based on the second-order cone programming,and establishes the linear equivalent model of the discrete reactive power adjustment device and its action times.The reactive power and voltage finely control of the power grid under multiple time scales is realized through the planned control,the intra-day real-time control and the feedback correction link.The example shows that the method can effectively reduce the network loss,reduce the number of reactive devices,and improve the economics of power grid operation.