Chaotic Prediction of Fluctuant Voltage Time Series in EAF Power Supply Grid

In order to solve the problem of voltage flicker produced by electric arc furnace (EAF) operation, it is a prerequisite to analyze the characteristics of voltage time series acquired from AC EAF power supply grid and to conduct voltage prediction based on chaos theory. In this paper, the voltage time series is firstly reconstructed in phase space to determine optimal delay time and embedding dimension and the maximum Lyapunov exponent (MLE) is computed to testify the chaotic characteristics of voltage fluctuation, which indicate the chaotic approach is feasible for voltage prediction. Then, the voltage is respectively predicted by MLE method and adding-weight one-rank local method. The comparison of the prediction result shows that the two methods both lead to good prediction effect, but the latter one is better in precision. Therefore, the addingweight one-rank local method is more suitable to predict the fluctuant voltage of AC EAF power supply grid. Key-Words: EAF, voltage fluctuation, chaotic prediction, MLE, adding-weight