A Novel Centralized Load Shedding Approach to Assess Short-Term Voltage Stability: A Model-Free Using Time Series Forecasting

In the wake of ever-growing integration of dynamic loads and power electronics-interfaced devices, power systems are becoming more complex with growing nonlinear behaviors, and preserving these systems' dynamic stability is a crucial challenge more than ever. Under voltage load shedding (UVLS) is one of the protection schemes adopted to enhance stability of the system at the events jeopardizing the system voltage. Despite underway efforts to design UVLS using analytical approaches/models, it is barely possible to estimate their various parameters with sufficient accuracy. Also, due to steady changes in characteristics of elements, especially loads, the estimated model at a time is not legitimate long lasting. Eluding these difficulties, this article presents a novel data-driven UVLS scheme with a centralized design to enhance short-term voltage stability during fault-induced delayed voltage recovery (FIDVR) events. The proposed methodology utilizes time series forecasting in conjunction with the long short-term memory neural network. In order to quantize FIDVR severity, a novel dynamic index called short-term voltage index (STVI) is introduced for motor-load buses. STVI is a function of both voltage magnitude and load angle. The proposed approach is successfully verified on an actual power system model.