A New Genetic Algorithm for the Reduction of Dynamic Power System Models

The development of new power grid components and solutions requires dynamic grid simulations, including vast transmission grids. These simulations are often prohibitively computationally expensive. For this, model reduction is a common solution. On the other hand, geographic information becomes increasingly important in grid simulation with an increasing share of renewable energy sources. However, this information has only been considered in reduction methods for static simulations so far. Therefore, we introduce a new parameter optimization method based on a new genetic algorithm that enables dynamic simulations for models previously reduced with such static topology-preserving reduction methods. The new method optimizes the selection and parameters of synchronous machine controls to approximate the original system behavior. We evaluate the new method using a standard IEEE benchmark model and demonstrate its applicability with a real-world transmission grid model with more than 300 nodes.