Modeling and prediction of DC electric arc furnace based on chaos theory and neural network

In the paper, research of the modeling problem of DC electric arc furnace (EAF) is presented. Based on the chaotic characteristic of the electrical fluctuations in the arc furnace voltage, the modeling and prediction of DC EAF using the phase space reconstruction theory and neural network are studied. First, the largest Lyapunov exponent of arc voltage waveform is calculated to confirm the chaotic behavior of arc furnace. Then a reconstructed phase space is obtained and the delay time and embedding dimension are calculated. Radial basis function neural network is applied to predict the arc voltage of arc furnace based on the calculated embedding dimension. It is showed by analysis that the proposed method can be applied to the modeling problem of DC EAF and shows better result than the previous method. Meanwhile, the prediction results of multi-step can also be used as a theoretical reference for the effective control of arc voltage.