Cold Rolling Mill Energy Consumption Prediction Using Machine Learning.

Even though energy consumption has a significant impact on the operational cost of tandem cold mills (TCM) of steel strips, not enough attention has been given to this important consumable throughout the years. Machine learning techniques are becoming extremely common in the steel industry due to the high level of automation of the segment and the large databases available. This work proposes a complete system capable of handling input data, training a machine learning algorithm, predicting the energy consumption of a TCM, and evaluating results. A performance comparison of Artificial Neural Networks (ANN) and Random Forest (RF) algorithms with an existing statistical model concludes that the RF outperforms the other two on a product-to-product base and on a monthly base. Actual model application has also been simulated indicating that the proposed system is adequate to handle energy prediction.