Three-phase AC/DC power-flow for balanced/unbalanced microgrids including wind/solar, droop-controlled and electronically-coupled distributed energy resources using radial basis function neural networks

This study presents a novel approach for robust, balanced and unbalanced power-flow analysis of microgrids including wind/solar, droop-controlled and electronically-coupled distributed energy resources. This method is based on using radial basis function neural networks that can be applied to a wide range of non-linear equation sets. Unlike conventional Newton-Raphson, the presented method does not need to calculate partial derivatives and inverse Jacobian matrix and so, has less computation time, can solve all the equation sets for the power grid and distributed energy resources exactly and simultaneously, and has enough robustness with respect to the R/X ratio and load changes. Also, because the power electronic interface provides some degrees of freedom in the steady-state and dynamic models, a new approach is required to solve the non-linear set of the power grid and distributed energy resource equations even with unequal number of equations and variables. The proposed method is a general method applicable to all types of power networks, including radial, meshed, and open-loop, and includes all types of buses, i.e. PQ, photovoltaic and slack buses. This method is tested on different microgrid test systems, and the comparative results validate its efficiency and accuracy.