Real-Time Optimal Control of Large Scale Power Systems with Application to Improving Transient Stability Response
MetadataShow full item record
Electric power systems are large nonlinear dynamical systems and therefore require unique control designs compared to the linear systems engineers often encounter. A special characteristic of the power system is structural change as a control option, with end-point stability the objective. However, these structural changes are not always treated rigorously when control systems are designed today. This research investigates the theory of control through structural changes and shows the need for an optimality based approach. A control framework is then created that considers both state trajectory and control action performance measures. Recovery to a stable operating point is through rapid and minimal control actions, which simultaneously are selected to constrain the dynamic response of relevant state variables along the path towards stability. The approach is well suited for responding to high order contingencies. These contingencies, while rare, are difficult to mitigate with existing controls, and can lead to cascading system collapse.Subsequently, several realizations are researched, designed, analyzed, and validated. The first is a solution that places few constraints on implementation complexity. It is model-based and predictive for control selection. Then, an improvement is developed that decreases computational requirements by reducing the space of admissible controls. A method to select this space that exhibits minimal impact on system disturbance response is invented. Finally, the optimality constraint is relaxed, but in a manner that maintains much of its original benefits. No model is required. Each of these instances demonstrate robustness to parameter errors, robustness to control actuation failures, robustness to modeling errors, and a significant improvement in the disturbance size the controller can tolerate while driving the system to an acceptable solution, compared to present approaches. Performance trade-offs between optimality of the selected controls and computational demands of the controllers are investigated. Experimental validation of results is demonstrated with the IEEE 39-bus test system.