A Transparent Data-Driven Method for Stability-Constrained Load Restoration Considering Multi-Phase Load Dynamics

Load dynamics significantly complex the system stability behavior during the load restoration process after an outage event. It is highly challenging to model the load dynamics and solve the stability-constrained load restoration problem. In this paper, the load dynamic model considering the inrush phase and enduring phase of the cold load pickup effects are firstly modeled. Then, the load restoration optimization model is formulated with stability constraints, ramping constraints, and system operation models. A transparent data-driven method based on optimal decision trees is developed to address all the short-term stability criteria (including frequency, voltage, and transient stability) considering the inrush phase of load dynamics. A set of accurate, linear, and tractable stability constraints are extracted offline and implemented for fast online decision-making. Numerical studies based on the New-England testing system are conducted to validate the proposed method.