Prediction of viscosity of blast furnace slag based on NRBO-DNN model
The viscosity of blast furnace slag significantly impacts operations, slag discharge, and heat recovery. However, accurately measuring or calculating viscosity is challenging due to the complex composition, interactions among variables, and experimental difficulties at high temperatures. To address...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-04-01
|
| Series: | Alexandria Engineering Journal |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S1110016825001541 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The viscosity of blast furnace slag significantly impacts operations, slag discharge, and heat recovery. However, accurately measuring or calculating viscosity is challenging due to the complex composition, interactions among variables, and experimental difficulties at high temperatures. To address this issue, a prediction model was developed based on slag composition. Data preprocessing included isolation forest outlier detection, missing data imputation, normalization, and Generative Adversarial Network (GAN)-based data augmentation, ensuring high-quality data. Among traditional neural network models, the Deep Neural Network (DNN) demonstrated the best accuracy. Optimizing the DNN with an intelligent swarm algorithm resulted in the NRBO-DNN model, which achieved MAE, MSE, RMSE, and R² values of 0.04050, 0.00305, 0.05527, and 0.97599, respectively. Compared to the unoptimized DNN, MAE, MSE, and RMSE decreased by 53.86 %, 50.30 %, and 29.50 %, while R² improved by 8.11 %. Tests on 100 datasets confirmed the NRBO-DNN’s superior accuracy, with an average error of 4.30 %. This study provides theoretical support and practical guidance for optimizing blast furnace operations. |
|---|---|
| ISSN: | 1110-0168 |