Application of RBF neural networks and unscented transformation in probabilistic power-flow of microgrids including correlated wind/PV units and plug-in hybrid electric vehicles

Due to the intermittent characteristics of wind and solar distributed energy resources and moreover, uncertainties in the power demand, the conventional power-flow methods could not cope with the active distribution networks and microgrids. Using some statistical methods like Mont Carlo simulation is always a reliable solution. However, it is time-consuming and cannot be applied to the large power systems. In this paper, a novel is proposed for robust probabilistic power-flow in radial and meshed electric power systems including renewable energy resources. The ability of radial basis function artificial neural networks for nonlinear mapping is exploited with an acceptable level of accuracy, and even exact to solve nonlinear equation set of power-flow analysis. This ability improves the speed of the algorithm because unlike conventional methods, the proposed method does not require calculating partial derivatives and inverse Jacobian matrix. The proposed method includes all types of buses, i.e. PQ, PV and Slack buses. The probability density function and cumulative distribution function for some of power system variable are compared with the other probabilistic power-flow methods for different test systems and the results validate its authenticity, robustness, efficiency and accuracy.