Integrated Simulation of Blackstart Grid Restoration with Cognitive Operator Modeling Control Implementation

Efficient restoration of electrical grids from a complete shutdown (i.e. ‘blackstart’) is nontrivial and critical to the safety and resiliency of communities. The problem is well studied from an optimization perspective; grids are analyzed and a set of blackstart plans are created so they are ready for an operator to follow them in an effort to restore the system. However, most existing approaches don’t explicitly account for time, and assume a known network state and full observability and control by grid operators which is not always the case. Here, we introduce a technique for the integrated modeling of optimized restoration plans with a cognitive simulation of operator actions. To this end, we built the CogTasks simulation library that allows for the cognitively sound simulation of hierarchical tasking and problem solving to serve as the basis of the grid operator model. We utilize a power flow-informed restoration framework to generate optimal restoration plans that are then enacted by our cognitive agent on a a dynamic representation of the grid. We show that utilizing this cognitive agent introduces extra noise and stability considerations into the implementation of the restoration plan; while the optimization reward function is still reasonably well correlated with the observed restoration time in our tests, this integrated model reveals value to including operator limitations and actions in restoration planning and lends explainability to the optimized schedules.