A robust state estimator for medium voltage distribution networks

Abstract

A closed-loop robust distribution state estimator was investigated. An approach that is suitable for medium voltage distribution networks which are either under-determined with limited real-time measurements or over-determined but with delayed information from smart meters was developed. The state estimator was designed to be robust against the effect of measurement errors, the type, location and accuracy of measurements, as well as temporary failure of the smart metering communication system. A machine learning function provides reliable input information to a robust state estimation algorithm. The output of the state estimator is then fed back to the machine learning function creating a closed-loop information flow which improves the performance of the state estimator. Test results and analysis on a 33-node system are provided.