A Deep Learning Model via Long Short Term Memory for Voltage Sag Location in Sparsely Monitored System

Voltage sag has already been recognized as a critical power quality issue in power system. In fact, not only economic loss but also social impact has been produced due to voltage sag. And hence, voltage sag location is of great importance to taking effective measures, evaluating power quality level, dividing responsibility and constructing harmonious power supply and consumption environment. And hence, a deep learning method via Long Short Term Memory for voltage sag location in power system, which is sparsely monitored is presented. In detail, for the presented model, the input is measured voltage through limited sensors in a sparsely monitored power system, and meanwhile, the output is the detailed line in the whole network. In this study, the data is collected via Matlab software and the algorithm is conducted through TensorFlow tool. The test results through IEEE 30-bus system illustrate that the accuracy of voltage sag location can be achieved with high accuracy.