FUZZY CONTROL OF STATE ESTIMATION ROBUSTNESS

This paper reports the main results from the application of Fuzzy Inference Systems - FIS - to tackle the problem of selecting the most adequate set of weights to use in State Estimation algorithms directed to distribution networks. These networks have distinctive characteristics regarding transmission ones turning the migration of software packages from EMS to DMS systems not immediate. In previous papers, the authors described a Fuzzy State Estimation algorithm that has the flexibility to adequately address these problems. However, this algorithm requires fine tuning of several weights. The authors solved this problem by training a FIS system using a set of rules derived from a large number of exercises run for small networks. This approach, as it will be illustrated in the paper, proved to be very successful in the sense that the FIS displayed a remarkable capacity of generalization since it extremely improved the convergence pattern of the Fuzzy State Estimation algorithm when analyzing larger networks, differently from the ones used to build the training set.