An Enhanced Transmission Operating Guide Creation Framework Using Machine Learning Techniques

摘要

Transmission Operating Guide (TOG) is a part of constraint management in power system real-time operations and short-term planning. TOGs contain instructions on interface limit values to be used as transmission constraints for specific operating conditions. Current TOG creation process is manual and time-consuming. It heavily relies on engineer's subjective judgment. This could lead to overlooking of some factors that are critical to interface limits as well as their inaccurate estimation, which might jeopardize the system security or underutilize available transfer capabilities. To address these issues, an enhanced TOG creation framework using machine learning techniques is proposed. The advantages of the proposed framework are demonstrated on two datasets of the ISO New England System. Simulation results show that (1) the proposed ensemble feature selection approach can effectively identify power system features that significantly impact interface limits, and (2) decision tree enables accurate prediction of interface limits while maintaining a simple representation of TOGs.