With the growing scale of new energy,new energy power contribute often exhibit a strong correlation,traditional random flow algorithm for strong correlation random variables was considered less.A method of probabilistic power flow algorithm based on semi-variable and Gram-Charlier series expansion was proposed in this paper.According node voltage and branch current expectations and sensitivity matrix,with wind power output,load changes,forced outages and generator fault lines and other uncertainties considered,load and conventional generators,wind turbine output and each node injection power of each order half invariant were calculated.Probability density function and probability distribution function were obtained by Gram-Charlier series expansion.IEEE-30 node test shows that the algorithm can reflect the uncertainty of large-scale new energy accessing to the system,and probability density function can be simplified to the semi-invariant algebra.Greatly reduce the computation time and it has a good convergence.