Temperature Prediction Model for a Regenerative Aluminum Smelting Furnace by a Just-in-Time Learning-Based Triple-Weighted Regularized Extreme Learning Machine

In a regenerative aluminum smelting furnace, real-time liquid temperature measurements are essential for process control. However, it is often very expensive to achieve accurate measurements. To address this issue, just-in-time learning-based triple-weighted regularized extreme learning machine (JITL-TWRELM) soft sensor modeling method proposed prediction. method, weighted JITL (WJITL) adopted updating the online local models deal with time-varying problem. Moreover, model considering both sample similarities and variable correlations was established as method. The effectiveness of demonstrated in an industrial process. results show that can meet requirements prediction accuracy furnace.