Reduced order modeling for load aggregation based on empirical covariance matrices

Load model contributes a lot in power system stability analysis. Induction motors aggregation is one of the most important parts in load modeling. This paper presents a new method for aggregating induction motors based on nonlinear model reduction. Firstly, the induction motors were seen as a nonlinear input-output system with bus voltage as input and equivalent impedance as output. Secondly, nonlinear model reduction technique was applied on the system to calculate the empirical covariance and transformation matrix of the system. Then the reduced-order system was obtained by performing Galerkin projection on the original system. Lastly, an adaptive genetic algorithm was used to get more accurate result. Simulations and analysis were carried out in IEEE 39-bus system. The simulation results show that the new method could attain high accuracy to simulate all the individual induction motors under different operating states and improve the precision of simulation.