Application Of Photovoltaic Probabilistic Model In Probabilistic Load Flow Calculation

With continuous changes of power system operating environment and rapid development of new energy for electricity,the randomness of system is constantly enhanced,especially the large-scale access of distributed power generation represented by photovoltaic power to power grid,making state of power system more complex and changeable,which will bring great challenges to planning and analysis of power system operation and security assessment.As a comprehensive calculation method,Probabilistic load flow(PLF)takes into account a variety of random factors,and the probability distribution characteristics of system state variables can be comprehensively analyzed,thus more accurately revealing weak links and potential dangers of system operation.Therefore,under background of new energy generation,probabilistic power flow is more suitable for safety and stability analysis,economic operation and reliability analysis of power system.In this paper,based on photovoltaic generation,probabilistic power flow calculation method and its application in power system are studied.The rapid development of photovoltaic power generation intensifies correlation and uncertainty of power system operation.Accurate modeling of photovoltaic output is a prerequisite to ensure accuracy of probabilistic power flow calculation.In view of this,this paper proposes a method to reconstruct probability density function of photovoltaic output using cubic spline function,which can avoid error caused by assumption of selecting known parameter distribution in modeling.Secondly,on the basis of Nataf transform,the calculation method of correlation coefficient is improved,and calculation efficiency is improved and combines with LHS to obtain the correlated sample.The semi-invariant method of piecewise linearization is introduced to reduce calculation error,and correlation sample is decomposed by Cholesky and sensitivity matrix is corrected.Finally,probability density function of output variable is obtained by using Cornish-Fisher series.Algorithm not only takes into account correlation of photovoltaic force,but also reduces error caused by single point linearization of traditional semi-invariant method.Compared with that of Monte Carlo simulation method(MCSM),the computation time is greatly shortened.The accuracy and practicability of proposed method are verified by testing IEEE 34 bus system and actual power grid in Huining County,Gansu Province.In order to more accurately analyze the influence of time-dependence oncorrelation of photovoltaic output,the dynamic probabilistic power flow calculation method considering correlation is proposed in this paper.In the aspect of photovoltaic output modeling,photovoltaic output is classified according to time period,and edge probability density of photovoltaic output is constructed by using non-parametric kernel density estimation.Then,a single or the hybrid copula function is selected according to characteristics of photovoltaic output in different periods.The joint distribution of photovoltaic output with correlation is constructed.Secondly,hybrid copula dependent probabilistic sequence operation is combined with linearization of AC power flow equation to obtain node voltage and power flow distribution.Finally,IEEE34 node is selected to verify algorithm,and simulation results show that:A more accurate probability density function can be established by selecting the appropriate copula function and classifying photovoltaic outputs according to time period.A reasonable serialization step can get the same calculation precision as the Monte Carlo method(MCSM),and the calculation time length can be shortened.