Preliminary Systematic Modeling and Dynamic Optimization of Power System Stability

Abstract

Dynamic stability is a primary concern in power systems. Small disturbances without safe control can develop into widespread blackouts. In 2015, 3571 recorded outages in the US affected 13 million people (United States Annual Report 2015). However, today’s regulating and controlling methods mostly aim at local optimization but lack a systematic optimization framework and dynamic interaction analysis, which may not perform well and cannot trace and control cascading events in a real complex system. Machine learning cannot deal with various unanticipated cascading events. In this paper, we integrate all primary operating conditions and regulating methods of power systems in a novel systematic model. Then we innovatively apply state transitions between operating conditions to describe the dynamic complexity of power systems. Our work supports the feasibility of adaptive model-based machine learning and hybrid Human-AI electrical power management system.