Creative Commons source reading · Prediction and forecasting
Hybrid fuzzy clustering and temporal deep learning framework for multi-parameter forecasting in industrial thermal processes
Vignesh V; Narendran GV; Senthil Kumar R; Sitharthan R
Originally published in Front Artif Intell · 2026-08-17
From the article
Selected source text. JournalOS removes markup and shortens excerpts where indicated. The full article provides the complete context.
Abstract
AbstractBlast furnace (BF) operation involves strongly coupled thermal and chemical processes governed by nonlinear heat transfer, gas–solid reactions, and dynamic operational control. Accurate real-time prediction of key thermal and gas parameters is essential for maintaining furnace stability, improving energy efficiency, and reducing carbon emissions. However, most existing studies focus on single-parameter forecasting and fail to adequately capture the heterogeneous temporal dynamics and strong process coupling inherent in blast furnace operations. In this paper, a hybrid fuzzy clustering and temporal deep learning framework is proposed for multi-parameter forecasting. This framework integrates Fuzzy C-Means (FCM) clustering with temporal deep learning models, such as Nonlinear Autoregressive with Exogenous Inputs (NARX),…
Results and discussion
This section analyses the predictive ability of the developed models and explains the implication of the findings. Figures 4–7 show the accurate prediction of CO% and Si content of the four models by plotting their forecasted values against the actual data. The GRU, NARX, and LSTM models deliver impressively accurate predictions. Their forecast lines closely follow the actual data, effectively capturing the high-frequency fluctuations and sharp peaks of these process variables. Whereas the RNN model struggled, predicting data that were too smooth and lagging. They could not train on the dynamic range of the data, especially for the volatile Si content, where crucial peaks and troughs are not forecasted accurately. Figures 8–11 show the comparison of forecasted values to the measured data of HMT and CO₂% for four models. The results…
Discussion
The summary of previous studies such as parameters prediction, algorithms, errors, outcomes and their limitations is given in Table 6. While most current research has focused on forecasting a single blast furnace parameter, e.g., silicon content (Si), HMT, or CO%. There is still a significant gap in integrated multi-parameter predictive models for thermal process monitoring. In this study, several hybrid ML models like NARX, RNN, LSTM, and GRU are integrated into one multi-output dashboard for real-time forecasting of key thermal and process parameters. This method not only increases the reliability of prediction and operational understanding of the thermal performance of the furnace but also remedies some of the limitations in existing research, such as no multi-parameter prediction, limited explainability, and constraints in real…
Editorial reading guide
- What data created and tested the model, and was evaluation independent?
- What is the cost of a false positive, false negative or out-of-distribution decision?
