With the depletion of high-grade iron ore resources, the efficient utilization of low-grade iron ore has become a critical demand in the steel industry. Due to its uniform particle size and chemical composition, pelletized iron ore significantly enhances both the utilization rate of iron ore and the efficiency of metallurgical processes. This paper presents a deep learning model based on ResUNet, which integrates three-dimensional CT images obtained through industrial computed tomography (ICT) to precisely segment hematite, liquid phase, and porosity. By incorporating residual connections and batch normalization, the model enhances both robustness and segmentation accuracy, achieving F1 scores of 98.37%, 95.10%, and 83.87% for the hematite, pores, and liquid phase, respectively, on the test set. Through 3D reconstruction and quantitative analysis, the volume fractions and fractal dimensions of each component were computed, revealing the impact of the spatial distribution of different components on the physical properties of the pellets. Systematic evaluation of model robustness demonstrated varying sensitivity to different CT artifacts, with the strongest resistance to beam hardening and highest sensitivity to Gaussian noise. Multi-scale resolution analysis revealed that segmentation quality and fractal dimension estimates exhibit phase-dependent responses to resolution changes, with the liquid phase being the most sensitive. Despite these dependencies, the relative complexity relationships among phases remained consistent across resolutions, supporting the reliability of our qualitative conclusions. The study demonstrates that the deep learning-based image segmentation method effectively captures microstructural details, reduces human error, and enhances automation, providing a scientific foundation for optimizing pellet quality and improving metallurgical performance. It holds considerable potential for industrial applications.

The quality control process for sintered ore is cumbersome and time- and money-consuming. When the assay results come out and the ratios are found to be faulty, the ratios cannot be changed in time, which will produce sintered ore of substandard quality, resulting in a waste of resources and environmental pollution. For the problem of lagging sinter detection results, Long Short-Term Memory and Genetic Algorithm-Recurrent Neural Networks prediction algorithms were used for comparative analysis, and the article used GA-RNN quality prediction model for prediction. Through correlation analysis, the chemical composition of the sintered raw material was determined as the input parameter and the physical and metallurgical properties of the sintered ore were determined as the output parameters, thus successfully establishing a GA-RNN-based sinter quality prediction model. Based on 150 sets of original data, 105 sets of data were selected as the training sample set and 45 sets of data were selected as the test sample set. The results obtained were compared to the real value with an average prediction error of 1.24 for the drum index, 0.92 for the low-temperature reduction chalking index (RDI), 0.95 for the reduction index (RI), 0.40 for the load softening temperature T_10%, and 0.43 for the load softening temperature T_40%, with all within the running time thresholds. The study of this model enables the prediction of the quality of sintered ore prior to sintering, thus improving the yield of sintered ore, increasing corporate efficiency, saving energy, and reducing environmental pollution.

Pellet is widely used in blast furnace ironmaking. Pellet quality affects the effect of ironmaking, the existing control system of grating‐rotary kiln mainly adopts manual control mode, and the quality of pellet production largely depends on the experience, fatigue, and sense of responsibility of the site operators. The use of the Internet of things (IoT) technology in the integration and improvement of enterprise information level, to achieve fine, intelligent production management, at the same time, is conducive to promoting steel enterprises to reduce costs and increase efficiency, energy conservation and emission reduction, transformation and upgrading, and taking a new road to industrialization. According to the working principle and technological characteristics of the grate‐rotary kiln at all stages, this paper designs the management system of firing pellets based on convolutional neural network (CNN) and IoT technology, so as to realize automatic recognition of image data obtained by the perceptual layer and make an intelligent analysis of it. The system can classify the working conditions of the current equipment, so as to judge whether the production process parameters of the grate‐rotary kiln are up to the standard, thus achieving the goal of controlling the quality of the finished pellet.

In recent years, deep learning has been widely used in the field of image visualization, and has attracted the attention of researchers. The main component of iron tailings is SiO2, In the process of quenching and tempering slag from iron tailings, melted SiO2 particles will randomly walk in the crucible, so that how to intelligently identify and track targets (SiO2 particles) in sequence image to promote the low consumption and efficient production of high value-added slag cotton has become an urgent problem. First of all, aims at the temperature requirements of slag cotton preparation process, using intelligent algorithm with deep learning and hierarchical clustering, in-depth analysis of the visual information of the high temperature melting process of SiO2 particles. Through centroid tracking and positioning technology, intelligent edge feature extraction technology and fitting method of experimental data through high temperature melting process of SiO2 particles quantitative characterization functions of visual characteristics of SiO2 particles melting process under high temperature environment were obtained. Approximation of the melting process of iron tailings in high temperature environments and embed this function into the static mathematical model of process thermal compensation an accurate thermal dynamic intelligent compensation model for slag cotton preparation process has realized.

Rotary kilns, as critical thermomechanical units in the steel industry, require accurate soft measurement of internal temperature to ensure product quality, enhance production efficiency, and prolong equipment lifespan. Direct measurement inside the kiln is often infeasible due to harsh conditions, while conventional sensing approaches are affected by time delay, nonlinear dynamics, measurement noise, and missing data. To address these challenges, this study proposes a hybrid soft measurement model that integrates Long Short-Term Memory (LSTM) networks, a self-attention mechanism, and the Kalman Filter (KF). LSTM is employed to capture temporal dependencies in temperature variations, attention weights the most informative hidden states, and KF suppresses noise while compensating for data gaps. Experimental results show that, compared with the best-performing baseline model (Informer) among the evaluated conventional baselines, the proposed method reduces Mean Absolute Error (MAE) by at least 28.79% and decreases Root Mean Square Error (RMSE) by at least 26.92%, maintaining estimation errors at consistently low levels. These findings demonstrate the model’s effectiveness as a reliable soft sensor for rotary kiln temperature monitoring, offering robust technical support for real-time process perception, precise control, and intelligent optimization in industrial applications.