Research On Dynamic Reactive Power Optimization Of Distribution Network Based On Load Forecasting

As an important part of the power system directly connected to the user,the distribution network is not only an important guarantee for the power supply to the users,but also intuitively reflects the user's requirements for power system safety,economy and power quality.Therefore,it is particularly important to perform reactive power optimization on the distribution network to make it economically safe.First,this paper introduces Locally Linear Embedding(LLE)and Support Vector Machine(SVM).Then the advantages of the two algorithms applied to load forecasting are analyzed by numerical simulation and short-term load characteristics.Based on the advantages of these two algorithms,a new load forecasting model-LSVM model is established.The LSVM model can not only reflect the advantages of the LLE algorithm in dimensionality reduction,that is,it can maintain the topology of the data in the dimension reduction process,and can also reflect the characteristics that the SVM algorithm can find the global optimal value of the objective function.Finally,the LSVM model is validated by using the simulation of the example to prove the feasibility of the model.Secondly,this paper introduces the static reactive power optimization of the distribution network in detail,and applies the genetic algorithm to the reactive power optimization of the distribution network.Aiming at the shortcomings of traditional genetic algorithm for the poor global optimization ability in static reactive power optimization,this paper uses the simulated annealing algorithm to improve the fitness function of traditional genetic algorithm.Through simulation,the improved genetic algorithm can be effective.The ability of the objective function to find the global optimal value is improved,and the convergence speed of the algorithm can be effectively improved to obtain a better optimization effect.Aiming at the problem that the load real-time fluctuation and the number of actions of the equipment can not affect the reactive power optimization of the distribution network,the paper divides the load,and each load segment after segmentation performs static reactive power optimization.Get the device value for each load segment.Then subtract the device values of the same device to get the pre-action table.According to the initial pre-action table and the real-time change of the load in each load segment,each time period of the initial pre-action table is reset,so as to achieve dynamic reactive power optimization of the distribution network.Finally,the dynamic reactive power optimization model of distribution network based on load forecasting is applied to IEEE-30 nodes for verification.The results show that this paper is feasible to solve the dynamic reactive power optimization problem of distribution network.