Data‐driven modelling methods in sintering process: Current research status and perspectives

Sintering is the first step of ironmaking. The amount of returned ore is related to the quality of the sinter, yield, and cost of ironmaking. Herein, a novel and interpreted method is proposed to intelligently predict and optimize the amount of returned ore, which is not the case with the traditional “black box” model. The optimal combination of input variables is determined based on the SHapley Additive exPlanation (SHAP) value and eXtreme Gradient Boosting (XGBoost) selection. The model is constructed considering an R2 of 0.91 and average error of 2.9%, which verify its feasibility. Found by SHAP value analysis, when annular cooler speed and sintering machine speed are below 1.7 m min−1, sintering machine speed is below 2.1 m min−1, coal and coke powder amount are more than 14 and 8 t h−1, respectively, and the amount of the returned ore is reduced. In the optimization process, the joint optimization method with regard to the speed of the annular cooler and sintering machine can help solve 91% of the problem of high‐amount ore return, thereby reducing the returned ore amount 13.3 t h−1 at average and the returned ore rate by 1.8–2.7%.

show abstract

“…The final returned ore was again used as one of the raw materials in the sintering process. [ 26,27 ]…”

Section: Theory and Methodsmentioning

confidence: 99%

The Amount Prediction and Optimization of the Returned Ore Generated from Sintering Process Based on SHAP Value and Ensemble Learning

Zhang

Tang

Chu

et al. 2023

steel research int.

View full text Add to dashboard Cite

show abstract

“…11e and f). 60 3.1.4 Other noble metal atoms. In addition to Pt and Pd, which are widely used in photocatalysis, other noble metal-loaded photocatalysts have been gradually developed by researchers.…”

Section: Noble Metal Dopingmentioning

confidence: 99%

“…(e) H 2 production rates of the as-prepared samples and mass loadings of isolated metal atoms. (f) H 2 production rates calculated based on the isolated metal atom mass loadings 60.…”

mentioning

confidence: 99%

Metal-doped carbon nitride: an all-in-one photocatalyst

Pang,

Li,

et al. 2023

EES. Catal.

View full text Add to dashboard Cite

show abstract

“…Among them, the MMD loss provides information about adjusting the model to fit the data distribution and is the more widely used loss function in domain adaptation. The MMD distance measures the distance between two data distributions in the kernel Hilbert space (RKHS), and the smaller the metric, the higher the degree of similarity between the two distributions, as indicated in Equation (12).…”

Section: Stacked Autoencoder (Sae)mentioning

confidence: 99%

Weighted target feature regression neural networks based soft sensing for industrial process

Guo,

Wang,

2023

Can J Chem Eng

View full text Add to dashboard Cite

There are often nonlinear and time‐varying characteristics in industrial processes. These characteristics cause difficulty in measuring product quality online. To address these issues, a weighted target feature regression neural network (WTFAER) was proposed for soft sensor modelling in this paper. The Pearson correlation coefficient was calculated to assign corresponding weights to process variables and design a weighted objective function. A target feature regression network (TFAER) was constructed using target correlation autoencoder with fully connected layer. After that, the weighted reconstructed information was applied to the TFAER model to extract deep quality‐related features and realize feature reuse. A deep network was formed by layered stacking to fully exploit the deep features for quality prediction. To make the proposed method domain adaptive, a maximum mean squared deviation (MMD) based regularization term was introduced in the loss function. Through the simulation experiments of debutanizer column and industrial polyethylene process, and compared with stacked autoencoder (SAE), variable‐wise weighted stacked autoencoder (VW‐SAE) and stacked target‐related autoencoder (STAE) methods, the effectiveness and generalization performance of the proposed modelling method were verified.

show abstract

Data‐driven modelling methods in sintering process: Current research status and perspectives

Cited by 47 publications

References 124 publications

The Amount Prediction and Optimization of the Returned Ore Generated from Sintering Process Based on SHAP Value and Ensemble Learning

The Amount Prediction and Optimization of the Returned Ore Generated from Sintering Process Based on SHAP Value and Ensemble Learning

Metal-doped carbon nitride: an all-in-one photocatalyst

Weighted target feature regression neural networks based soft sensing for industrial process

Contact Info

Product

Resources

About