Research on Prediction Method of Finish Rolling Power Consumption of Multi-Specific Strip Steel Based on Random Forest Optimization Model

The prediction of energy consumption data is of great importance in industrial process, among which the accurate prediction of power consumption at the production line level plays an even MORE essential role for energy conservation and consumption reduction under the background of steel intelligent manufacturing. In this paper, a random forest prediction model based on principal component dimensionality reduction and artificial bee colony dynamic search fusion is proposed to predict the power consumption prediction of multi-standard rolling stock in strip hot rolling finishing control section. The accurate prediction of the power consumption of the rolling section is realized with methods including data dimensionality reduction and parameter optimization. Firstly, principal component analysis is used to extract the principal components of the parameters affecting the strip rolling power consumption. Under the premise of ensuring the prediction accuracy, the computation time cost is reduced, the random forest prediction model based on principal component extraction is established, and artificial bee colony is used. The dynamic search method optimizes the key parameters that affect the accuracy of random forest prediction, and further improves the prediction accuracy. The algorithm verification by using the pre-processed plant-level real-time data shows that the method can accurately predict the power loss of multi-size rolled parts with a low calculation time and a high prediction accuracy. This result indicates that the proposed model can meet the actual production of factory-level rolling demand for forecasting.