Research On The Power Forecasting And Optimal Placement For Distributed Photovoltaic

With the emphasis on energy,ecology,climate and other issues of attention in international society,to greatly develop renewable energy becomes the important measure for dealing with these problems.As the renewable energy,solar energy develops rapidly in recent years;meanwhile,since distributed photovoltaic is the object in next development of photovoltaic industry,based on the characteristics of flexible operational mode,saving energy consumption and high energy utilization rate,it is the significant component of distribution link of smart power grid.Due to the intermittency and instability of photovoltaic output power,the structure as well as operation mode of power distribution network will change along with the increase of permeability of distributed Photovoltaic in power distribution network.In order to reduce its influence on power distribution network,it is necessary to predict the short-term power of distributed Photovoltaic and optimize the location as well as capacity of distributed Photovoltaic in power distribution network.For this reason,this paper established the short-term power prediction model of distributed photovoltaic,predicts the collected samples and implements error analysis on result;with respect to the optimal configuration of distributed photovoltaic,multi-objective optimal model is constructed;by means of multi-objective differential evolution algorithm,the optimal configuration of distributed photovoltaic in standard power distribution network is carried out.The major research contents are as follows:(1)The composition and classification of photovoltaic power generation system is researched.Through the simulation of theoretical model and real data,from the perspective of influencing photovoltaic power output,the photovoltaic power characteristic is researched,and the input of photovoltaic prediction model is determined.In the viewpoints of system load flow,power supply reliability and power quality of power distribution network,consumption of power distribution network,the influence of distributed photovoltaic grid connected on power distribution network is analyzed;the optimal objects as well as constraint conditions of multi-objective optimal model are selected.(2)The basic principle of wavelet transform is researched;after applying wavelet transform pretreatment to the collected photovoltaic power,the sequence components of power are obtained so as to establish prediction models respectively for prediction.The principle and structure of ESN neural network are analyzed to determine the training method of ESN network.According to the collected photovoltaic data,the structural parameters,input and output of prediction model are determined,which are used to set up WT+ESN photovoltaic power prediction model.(3)In matlab platform,the samples in sunny day and cloudy day are selected;the four prediction models of ESN,BP,WT+ESN and WT+BP are utilized to predict the power generation of distributed photovoltaic;three kinds of error indicators are applied to evaluate prediction result.The simulation result shows that,compared with other three models,the prediction curve in WT+ESN is more stable and the variation trend is more closer to active curve,whose evaluation index is the optimal in of three errors and verifies the effectiveness as well as superiority of WT+ESN prediction model.(4)The multi-objective optimal model containing cost and maintenance expenses of distributed photovoltaic,active power consumption of powerdistribution network,voltage stability index VSI and various constraint conditions is established.Meanwhile,the mathematical description of multi-objective optimal model is introduced.The principle and algorithm process of differential evolution algorithm are researched;in view that,traditional differential evolution algorithm excessively depends on experience control parameters,the self-adaptive strategy is integrated into the algorithm;multi-objective differential evolution algorithm—MOSADE is proposed combined with Pareto dominance concept.(5)Aimed at the standard power distribution network in IEEE-33 node,the research of distributed photovoltaic optimal configuration is carried out in matlab platform.The simulation result shows that,in the aspect of optimal network consumption,compared with DE and LDWPSO,SADE maintains population diversity in earlier stage of algorithm and improves convergence rate in later stage of algorithm;with respect to global optimum,MOSADE is adopted to implement effective configuration on DPV in group 1-10.The result indicates that,the reasonable configuration of DPV can enhance voltage level of power distribution network,reduce active power network consumption,and increase electricity generation income.Therefore,it verifies multi-objective optimal model,and rationality as well as effectiveness of MOSADE algorithm.