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

Abstract

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

A Dynamic Time Warping Based Locally Weighted LSTM Modeling for Temperature Prediction of Recycled Aluminum Smelting

“…However, the aforementioned soft sensor modeling methods are built by using a global, offline approach, and the models are difficult to update once they are built, which is detrimental to the prediction accuracy of soft sensors for time-varying complex industrial processes. To address these problems and achieve online modeling and updating, strategies of moving window (MV) [13] and just-in-time learning (JITL) [14] have been proposed. The strategy used by JITL builds models related to the entered query samples.…”

“…The JITL strategy based on locally sampleweighted considers the degree of similarity between the query sample and each training sample and uses this degree of similarity as the weight of the corresponding training sample, which greatly improves the JITL model's ability to handle nonlinearities. The traditional approach of JITL uses the Euclidean distance as the similarity measure between the historical and query samples, and the samples usually contain only one sampling point [11], [14], [23], which is unfavorable for complex industrial processes with time-series characteristics. When the sliding window method is used to obtain samples with a certain time length, the original temporal characteristics of the data are effectively retained.…”