Grid reduction approach for state identification of distribution grids

Against the background of the increasing amount of DG sources and the intention of local utilities to offer system services like voltage control and demand management at the distribution level which is not entirely observed, knowledge of the grid state becomes more and more important. In this paper an entirely new approach on grid state identification is introduced. This new method uses already measured nodal and terminal values as input variables and extensively exploits typical properties of distribution grid topology. In doing so, it is able to more accurately estimate the grid state even without installation of new measurement facilities which is a further advantage. To estimate nodal voltages, the grid is reduced to the observed central busbars and a minimum of auxiliary nodes on remaining tie-lines. Thereby the equation system size is reduced as well and the resulting equation system gets solvable. After the reduced grid state is calculated a recursive algorithm is applied to back-reference to the original grid by re-adding a small number of nodes in each recursion step until the original grid is reconstructed again. The obtained results are more accurate than the ones obtained by previous methods. Furthermore, the grid reduction approach does not need a linearized model and thus approximation errors are independent of the deviation from an arbitrary chosen operation point.