An improved joint normal transform (JNT) method is proposed to model the dependence factors in power systems in stochastic power flow computation. First

the procedure for the traditional JNT method is described and the cause of correlation structure remaining unchanged is analyzed with reference to the properties of rank correlations when JNT method is utilized in dependence modelling. Then

the most important part is applying orthogonal transformation to the traditional JNT method according to the characteristic that the JNT method is based on normal distribution. This improved sampling method is combined with Monte Carlo method to calculate the stochastic power flow. Finally

a calculation example is designed to verify the feasibility of the proposed improved JNT method and stochastic power flow algorithm by adjusting the parameters of an IEEE 30-bus network. This work is supported by National Natural Science Foundation of China (No.51377021)

National Key Technologies R&D Program (No. 2013BAA01B00) and State Grid Corporation of China (No. SGTJ0000KXJS1400087).