Research On Stochastic Power Flow Algorithm Considering The Uncertainty Of Photovoltaic Source And Branch Fault

Photovoltaic power generation has many advantages of cleaning and abundant reserves,and it has been widely used with the gradual depletion of fossil energy, but the intermittent and volatility of the photovoltaic source will have an impact on the power grid safety and stability. This thesis focuses on the research of stochastic power flow algorithm considering the uncertainty of photovoltaic source and branch fault.Studying the power system uncertainties deeply, this thesis establish the light intensity probability model obeying Beta distribution in the short term, PV output probability model,use the compensation method to analog branch fault and establish its probability model, and establish linear AC flow model.On the basis of existing algorithm, considering the uncertainty of photovoltaic source and branch fault, we use a method of combined cumulant and series expansion to calculate stochastic flow, this algorithm avoid complex convolution calculations by using the additive of combined cumulant, it reduces the amount of calculation, and only runs deterministic flow for one time. To improve the rapidity of stochastic flow considering branch fault, using the compensation method based on full probability theory, regarding the grid topology has not changed, the random branch fault is equivalent to node injection compensation power obeying the 0-1 distribution, It avoids the repeated calculation from changes of grid topology.Applying this algorithm to the IEEE14 nodes power system containing large-scale PV source and calculate its stochastic flow, we can measure the light intensity data and obtain Beta distribution shape parameters of PV model based on the LNUT New Energy Technology Laboratory 5kW grid-connected photovoltaic systems, the results of simulation show that:(1) The algorithm of this thesis can effectively obtain the distribution function of node status and branch flow, even its over-limit probability. And analysis results shows, the fluctuation degree of unknown variable is inversely proportional to the distance between PV source and itself; flow over-limit probability is correlated with the load capacity of PV source access point, and with the increase of PV source permeability, the over-limit probability will arise integrally.(2) The fitting precision of three kinds of series expansion is different on the non-normally distributed random variable, the precision of Edgeworth and Gram-Charlier series is similar, the precision of Cornish-Fisher series is better and it can effectively eliminate negative probability than the formers.(3) Comparison shows this algorithm is faster than the Monte Carlo method by 20-25 times, and it improves the rapidity in the premise of guaranteeing its accuracy.