Data-driven approach for power flow direction estimation in presence of renewable energy

Knowledge of power flow directions in electrical grids is crucial for maintaining grid stability, efficiency, and effective resource utilization. This paper suggests a technique based on data analysis to estimate power flow directions in a power grid by utilizing Deep Neural Networks (DNNs). The DNN model is developed using an extensive dataset obtained from power flow simulations carried out on the IEEE 14-bus system, which includes real-time load profile information and solar generation data. The model leverages the system's active and reactive power injections as input to accurately forecast power flow directions across each transmission line. Notably, Gaussian noise is introduced into the dataset to enhance its robustne ss and address potential measurement errors. This method could enhance the grid monitoring and control capabilities, particularly in situations where real-time data acquisition might be compromised.