New monitoring approach for distribution systems

This paper introduces a new data-driven bottom-up monitoring approach for distribution systems. Unlike model-based techniques, which require a given number of measurement inputs for their state estimation equations, this approach uses an artificial neural network to directly estimate the voltages. Thus, the process does not use state estimation equations and is flexible with regard to the number of required measurements. Depending on the available measurements, the estimation accuracy may vary but there are no convergence issues. Furthermore, rather than performing voltage estimations for the entire system in a single step, this approach uses a hierarchical, bottom-up structure to build up the overall picture. More precisely, the estimations performed at the MV/LV substations are communicated to the upper-level HV/MV substation, contributing to more accurate voltage estimation at MV level. The estimation process is computationally simple and can be executed on low-cost hardware, as demonstrated in this work. In our test, BeagleBone Black was used for implementing the developed algorithm. Preliminary results are presented which show representative estimations in an LV feeder.