Rolling Thickness Prediction Based on the Extreme Learning Machine and Clustering

Abstract

The accuracy of thickness is an important standard to measure the strip quality. Therefore, it is crucial to accurately obtain a high precision thickness. The ELM (extreme learning machine) based on clustering forecast method is presented for hot rolled strip thickness prediction. Firstly, strong correlation properties of thickness are obtained by data pretreatment, in order to ensure the effectiveness of the thickness model. Then, a clustering analysis is made about the strong correlation attribute data. Finally, ELM network is performed respectively for each type of prediction. This paper uses filed production data for training and testing, and takes the BP network prediction model as comparison. The simulation results show that this method can predict the thickness more quickly and accurately, and better meet the needs of actual production.