Model-Driven Approach to Smart Grid Stability Prediction in Neo4j
Dec 08, 2021
Stability is of utmost importance when it comes to smart grid infrastructures. Dramatic parameter variations and fluctuations can lead to wrong decisions, which could lead to fatal consequences. In this paper, we propose a model-driven methodology for highly automated machine learning approach to smart grid stability prediction. Stability prediction is treated as binary classification problem and implemented relying on Neo4j graph database's Graph Data Science Library (GDS). The proposed framework is evaluated on open, publicly available dataset. According to the achieved results, the predictive model shows better performance in terms of accuracy and execution time compared to other solutions based on deep learning. On the other side, the adoption of model-driven approach is beneficial when it comes to reusability and convenient experimentation compared to manual, non-automated design.