Modeling of DC electric arc furnace using chaos theory and neural network

In the steel industry, numerical modeling of electric arc furnaces (EAFs) is an important method to improve the power quality. However, the complicated nature of EAFs makes this process rather difficult. In this study, the complex behavior of an EAF is analyzed using chaos theory and neural network. According to the embedding theorem, if the embedding dimension and delay time are chosen properly, the state can be reconstructed without a change in the dynamical properties. In particular, after proper selection of the embedding dimension and delay time, the state is reconstructed in the form of delay coordinates. The reconstructed state can be used to perform one-step prediction, which involves finding an appropriate mapping function from the state to time series values. Because a neural network is a good choice for this problem, several neural networks were tested and a multi-layer perceptron was selected here. With such a network, we can develop models of arc voltage, current, and resistance, with high accuracy.