Linear and nonlinear modeling and prediction of harmonic signals.

摘要

Harmonic and interharmonic signals arise due to nonlinear power electronic devices and cause various problems of quality control in electric power systems. For solving these problems, controllers are needed for negative feedback of the estimate of the harmonic or interharmonic content. This involves on-line calculation of the power signal parameters including voltage and current magnitude, frequency and phase angle in the presence of measurement noise.; This work, based on the observation made by previous researchers, regarding the random nature of such signals, presents seven kinds of harmonic signal models. These stochastic models consider the sinusoidal signal amplitude, radial frequency and phase angle with measurement noises in many different ways and show the outcomes by computer simulation.; Based on these linear and nonlinear harmonic signal models, estimators are designed to estimate their one-step-ahead predicted values. The focus points of this dissertation are the estimators designed for some novel nonlinear harmonic models. The performance of these estimators is given in a comparative way so the reader can draw conclusions.; This work also shows the models' and estimators' robustness by use of the estimators designed based on one signal model to track the signal generated by another harmonic signal model. Also included are some simulations involving estimator performance based on real data from Ft. Davis, TX, electric distribution system. Simulation results show that these harmonic signal models and the signal predictors designed based on them perform adequately to be used in an on-line fashion in active filtering of harmonic and interharmonic signals.