Real-time Parameter Estimation of Dynamic Power Systems using Multiple Observers

Abstract

In this paper we describe a method suitable for real-time estimation of parameters of differential algebraic equation (DAE) dynamical models, generally used to model power systems. The method uses multiple observers that run in parallel, processing the same measured data from the process under consideration. The outputs from these observers are then used in a secondary least-squares estimation (LSE) process in order to identify the unknown parameters. We will demonstrate this parameter estimation using multiple observers (PEMO) on a single machine infinite bus (SMIB) example (where the generator inertia, prime mover torque and damping of the generator are unknown). This method can also be used as a fault detection, isolation and identification (FDI) filter that tracks parameter changes that can be indicative of such events as line outages, load and generation changes in a power system. We illustrate this concept on a nine-bus example.