Evaluation of short-term stress stability in real-time using PMU measurements and artificial intelligence : towards a self-healing smart grid

This article presents a new methodology for real-time evaluation of short- term voltage stability (ETCP) using Phasor Measurement Units (PMU), intelligence techniques, artificial intelligence, and data mining. This stability assessment methodology using measurements of PMU-WAMS consists of the real-time application of an intelligent machine designed and trained offline in a three-step process. First, a database is generated with the result of dynamic simulations using Monte Carlo simulation. In the second step, it classifies the working status of the system with a systematic procedure that provides the first constraint, which quickly collapses with the estimate of the maximum Lyapunov exponent and, subsequently, the dynamic performance of the stress amplitude in the calculation of dynamic indices. In the third step, data processing is performed to extract a symbolic representation of the temporary series multi-variable, which is used to train an intelligent machine based on the algorithm random forest learning. The proposed methodology is validated in the New England system of 39 bars and the performance of the methodology to classify real-time ETCP is verified, showing that I was able to anticipate voltage stability problems, reduce temp and perform control actions such as give you advice that prevents or treats the problem.