Online reactive power optimization of active distribution network based on decision tree integration
To achieve voltage control in distribution networks under frequent fluctuations in new energy output,based on parameter planning theory,a mixed integer second-order cone programming model with parameters was es-tablishedwith node load demand and new energy output as uncertainty parameters.To improve the online solving ef-ficiency of reactive power optimization models,the integer variables and tight constraints of the optimal solution of mixed integer cone programming were defined as discrete decision actions,and a decision tree ensemble classifica-tion model was used to learn the mapping relationship between uncertainty parameters and decision actions.The de-cision action dataset was constructed using historical data,and a decision tree model was trained,the optimal deci-sion action was predicted based on the measured parameters,and the integer variables and tight constraints in the mixed integer programming were fixed,so as to improve the solving efficiency of the model.The improved IEEE 33 node system was used for numerical analysis,and compared with traditional mathematical optimization methods,the feasibility and advantages of the proposed method were verified.
active distribution systemvolt/var optimizationparametric planning,distributed new energy,decision tree
万方数据
维普
