Applying Deep Learning Model to Predict Smart Grid Stability

Abstract

Smart grid is an advanced concept of power systems which harmonizes electricity and communication in systems networks. It provides information for the producers, operators, and the consumers in real time. There is an extreme demand to efficiently conduct the power supplied to the consumption domains such as households, organizations, industries, and smart cities. In this respect, a smart grid with a stable system is being required to supply the dynamic power requirements. Predicting smart grid stability is still challenging due to the many factors which affect the stability of grid, one of these factors is customer and producer participation because identifying the participation can lead to the stability of smart grid. In this work, we propose a deep learning model based on Densely Connected Convolutional Network and Residual network structure to detect the stability of smart grids. The results of the proposed model are compared to other popular classifier models used in different studies. Those models are Support Vector Machine, Logistic Regression, Decision Tree, Random Forest, Gradient Boosted Trees, Multilayer neural network, Gated Recurrent Units, Recurrent Neural Networks and Long Short-Term Memory and the proposed model outperforms the other models.