Fast on-line voltage security monitoring using synchronized phasor measurements and decision trees

This paper presents the results of a study for an online voltage security monitoring system in AEP. It is designed to work inside the control center with phasor measurement units (PMUs) providing the synchronized real-time measurements. Classification type decision tree (DT) models are utilized to predict voltage security status. Fast and direct measurement from PMUs combined with decision trees quickly assesses voltage security, leaving more time for corrective or preventive actions. Two attributes are considered in the decision tree models: generator VArs and angular difference. Both are observed to be good indicators of voltage security status. The paper focuses on the angle difference attributes because bus voltage angle is directly measured by the PMUs. A stressed power system is characterized by widening angular separation of bus voltage angles as it moves towards voltage insecurity. DTs exploit the complex nonlinear relationship between voltage security status and generator VArs/angular difference in terms of hierarchical rules extracted from a large number of offline loadflow simulations. Several DT models were analyzed using the existing PMUs, and in the future when additional PMUs are incrementally installed. A PMU placement technique is discussed to determine the critical locations of new PMUs. The voltage security monitoring system was tested on a 360-bus subsystem inside AEP.