Deterministic And Robust Dynamic Reactive Power Optimization Model And Their Two-stage Decoupling Algorithm

Dynamic reactive power optimization is a very important method in maintaining voltage stability,reducing energy loss and improving economy.The mathematical characteristics of multi-period,large-scale,strong coupling,nonlinearity and integer variables make the efficient solution of the traditional dynamic reactive power optimization model a difficult problem.As intermittent loads and distributed power sources s are increasingly connected to the grid,it brings great challenges to the reactive power control of the grid.How to deal with the challenge and realize the correct decision-making of transformers and capacitors to ensure the economic and safe operation of the power grid is the research focus.Based on the above problems,this paper has done some studies as follows:To solve dynamic reactive power optimization problem,this paper proposes a two-stage algorithm based on decoupling interior point method(IPM)and mixed integer programming(MIP)with variable neighborhood search.In the first stage,the discrete variables are relaxed first,and using the sigmoid function processes the absolute value function to realize the high-precision continuity of the original model.Then,adopting the idea of decoupling interior point method(IPM)realizes the efficient solution of the continuous model interior point method by using the blocked diagonal and band-edge matrix of the KKT matrix of IPM.In the second stage,first the original problem is transformed into a mixed integer linear programming(MILP)problem based on continuous solution in the first stage by adopting the idea of high-precision linearization.And then using variable neighborhood search strategy addresses MILP,which greatly improves the solution efficiency of the linearized model.In order to verify whether the algorithm proposed in this paper can efficiently solve the dynamic reactive power optimization problem,a simulation analysis of a 26-node system in a certain area is carried out.In view of the challenges that intermittent loads and distributed power sources bring to grid reactive power control,this paper establishes a robust dynamic reactive power optimization extreme scenario model with the minimum energy loss of the predicted scenario as the objective function and proposes a two-stage algorithm based on scene decoupling and time period decoupling.Using extreme scenario method deals with the randomness of the load in this model.Minimum energy loss throughout the day of the predicted scene is considered as the objective function of the model.Consider power grid security constraints and generator reactive output upper and lower limits as independent constraints for each scenario,and discrete variables upper and lower limits and the number of operations of discrete devices throughout the day as the coupling constraints among scenarios condition.To solve this model,this paper proposes a two-stage algorithm based on scenario decoupling and period decoupling.In the first stage,the original model is first processed continuously.and then separable structure continuous model of robust dynamic reactive power optimization based on extreme scenario method is constrained by constructing a linear coupling constraints for transformer taps and capacitor among each scene,and adding discrete variable upper and lower limits constraints and discrete device action times constraints to the independent constraints of each scene.Then using ADMM solves the separable structure continuous model.Finally using decoupling IPM solves each sub-problem of ADMM.In the second stage,the original model is linearized based on the continuous solution of ADMM method,which is solved by mixed integer linear programming with variable neighborhood search strategy.In order to verify whether the model and algorithm proposed in this paper can provide correct control decisions for the randomness of the source-load,a simulation analysis of a 26-node system in a certain area is carried out.This paper is funded by the National Key Research and Development Program "Wind/Photovoltaic Power Forecasting Technology and Application to Promote the Consumption of Renewable Energy(2018YFB0904200)" and the supporting technology project of the State Grid Co.,Ltd.(same name,SGLNDKOOKJJS1800266).