Neural-network-based load modeling and its use in voltage stability analysis

Voltage stability analysis is very important for predicting potential voltage instability. Load modeling plays a key role in voltage stability assessment. In the literature, most available approaches to the voltage stability problem are either static or quasistatic, which do not take load dynamics into account. First, this paper presents a survey of those approaches, makes a comparison between them, and points out the possible consequences of not considering load dynamics, which at worst can be a complete voltage collapse. Based on this observation, modeling of load dynamics is considered in this paper, and neural networks including recurrent neural networks are applied for load modeling. Furthermore, this paper presents the strategies for the first time to incorporate the neural-network-based load model into static and dynamic voltage stability analysis. The computation of the relevant sensitivity is carried out for the neural-network-based load model, and the results are used in the popular modal analysis. The proposed methods are tested on both the IEEE 14-bus system and real data.