A novel permeability index prediction method for blast furnace based on the improved extreme learning machine

The permeability index of blast furnace reflects the operating conditions of the blast furnace and is an important basis for operators to judge whether the blast furnace is going forward. Based on K-ELM, this paper proposes a prediction model of permeability index based on KOS-ELM. Aiming at the shortcomings of KOS-ELM, such as high computational complexity and long learning time, this paper proposes an ALD-KOS-ELMF combining ALD and the sliding window. In the modeling process, taking into account the noise accompanying the blast furnace production data, this paper chooses the wavelet transform to reduce the noise inside the data. The experimental results show that the prediction model of permeability index based on ALD-KOS-ELMF not only improves the computational complexity and learning time, but also has better prediction accuracy and generalization performance than KOS-ELM. The model can meet the actual needs of relevant operators.