UBC Theses and Dissertations
Parameter estimation applied to power system models subject to random disturbances Muller, Hans
For control of an electric generator, it is desirable to approximate the remote part of the power system by a low order model, whose parameters can be estimated from measurements. The proposed model for the remote system consists of a large synchronous machine and a random varying load to account for the dynamic behaviour and for the small fluctuations that are always present. The maximum likelihood algorithm is a good method for parameter estimation of a noisy system. It can even be used for estimation from measurements alone, without applied disturbance, which could be of great advantage in a power system. The algorithm is derived in a form suitable for efficient digital processing of sampled measurements from a linear continuous system with physical parameters. The performance of the algorithm is tested by computer simulation of a simplified model for several cases of deterministic and/or stochastic input. It is demonstrated, that good estimates can be found from the output of a disturbed system when no intentional input is applied.
Item Citations and Data