Analytics and Optimization Techniques on Feeder Identification in Smart Grids

One of the problems faced by electric power distribution system operators is to know with certainty the actual location of all their assets in order to manage properly the grid and provide the best service to their customers. In this work, we present a procedure for the identification of low voltage feeders or distribution lines in smart grids that is based on the mathematical formulation of the problem as an optimization model. In particular, we define the model with 0-1 variables (as many as meters to be identified in the different feeders) and with as many restrictions as the number of points in time that are considered. Given the large size of the problem in practice, the use of conventional optimization software becomes unfeasible. Based on this approach, and making use of the linear relaxation of the problem, some analytics over the coefficients (i.e., meter loads) and the special structure of the problem itself, we have developed an iterative procedure that allows us to recover the entire solution of the initial model in an efficient way. We have carried out a computational experience on a set of anonymized real data, obtaining results that support the efficiency of the proposed procedure.