Research On Reactive Power Optimization Of Distribution Network Based On Multi-center Interior Point Method Of Load Forecasting

With the rapid development of economy,the amount of electricity demand by users is increasing,which puts forward higher and higher requirements for the stability and reliability of power system.in order to maintain the stable and reliable operation of the power system,it is necessary to adjust the power system.dynamic reactive power optimization has been paid more and more attention as an effective means.on the other hand,load forecasting of power system is a basis of dynamic reactive power optimization.the accuracy of its prediction is directly related to the effect of dynamic reactive power optimization of power system.Based on this,the reactive power optimization of distribution network is studied in this paper,and the dynamic reactive power optimization method of distribution network based on multi-center internal point method of load forecasting is proposed.Specific studies are as follows:Firstly,this thesis introduces the background and significance of the research,and then introduces the short-term load forecasting technology at home and abroad,static reactive power optimization technology and dynamic reactive power optimization technology at home and abroad.In addition,the application of forecasting algorithm in short-term load forecasting is introduced.Secondly,this thesis introduces the principle of time series algorithm and support vector machine(SVM)algorithm and the relevant applicable conditions.Based on the relevant features of short-term load forecasting,the advantages of two algorithms are adopted,and a hybrid algorithm based on time series algorithm and SVM algorithm is established to predict the short-term load of power system.By using the algorithm to predict the reactive power load,and using the root mean square relative error RMSRE to evaluate the prediction results,through the relevant simulation,and compared with the traditional method,the simulation results show that this method has the characteristics of high prediction accuracy and adaptability.Then,this thesis reactive power optimization based on dynamic reactive power optimization is analyzed in detail.Aiming at the disadvantages of slow convergence and high initial value of multicenter-corrected internal point method,the immune algorithm is introduced into the multicenter-corrected internal point method.Throughthe simulation results,the multi-center-corrected interior point method combined with immune algorithm can get better reactive power optimization effect.Finally,the multi-center-corrected interior-point method combined with immune algorithm is used to calculate the static reactive power optimization of each conforming segment,and the equipment value controlled by each load segment is obtained.The dynamic reactive power optimization strategy is formed by introducing the load coefficient and controlling the time of the reactive power compensation device trigger action by the value of the load coefficient in each time period.By establishing the simulation model of IEEE30 node system,and through simulation analysis,it is verified that this method has the characteristics of validity and reliability.