The most important control parameters in the methanol distillation process, which are directly related to product quality and yield, are the temperature, pressure and water content of the finished product at the top of the column. In order to adapt to the development trend of modern industrial technology to be more accurate, faster and more stable, the fusion of multi-sensor data puts forward higher requirements. Traditional control methods, such as PID control and fuzzy control, have the disadvantages of low heterogeneous data processing capability, poor response speed and low control accuracy when dealing with complex industrial process detection and control. For the control of tower top temperature and pressure in the methanol distillation industry, this study innovatively combines generative artificial intelligence and a type II fuzzy neural network, using a GAN for data preprocessing and a type II fuzzy neural network for steady-state inverse prediction to construct the GAN-T2FNN temperature and pressure control model for an atmospheric pressure tower. Comparison experiments with other neural network models and traditional PID control models show that the GAN-T2FNN model has a better performance in terms of prediction accuracy and fitting effect, with a minimum MAE value of 0.1828, which is more robust, and an R² Score of 0.9854, which is closer to 1, for the best overall model performance. Finally, the SHAP model was used to analyze the influence mechanism of various parameters on the temperature and pressure at the top of the atmospheric column, which provides a more comprehensive reference and guidance for the precise control of the methanol distillation process.

Against the backdrop of the synergistic advancement of Industry 4.0 and the dual-carbon strategy, traditional university mathematics education struggles to meet the demands for cultivating engineering talents’ integrated competencies in mathematics, specialization, and application. The STEM education paradigm urgently needs innovation. Guided by sustainable development principles, this study explores integrated approaches to university mathematics teaching for advanced manufacturing. It constructs a four-stage cyclical framework, Concept–Algorithm–Equipment–Evaluation (CAEE), and integrates Fourier Transform systems into industrial inspection workflows, using silicon carbide wafer thickness measurement as a case study. Targeting second-year students in Measurement and Control Technology and Instrumentation, a comparative design involving an experimental and a control group was employed. Comprehensive evaluation utilized AI-powered dynamic questionnaires, multimodal eye-tracking and EEG data, along with mixed-methods research. Results indicate that the assessment tools achieved high reliability and validity (0.906). The experimental group demonstrated significantly superior performance in deep learning proficiency and subject-specific educational structure (effect size 0.67) compared to the control group, along with modest positive enhancements in cognitive engagement and social interaction dimensions. This pedagogical model transcends conventional ‘knowledge collage’ integration, transforming mathematics from an external auxiliary tool into an ‘endogenous variable’ within industrial systems. It establishes a replicable and scalable STEM education practice paradigm.

As the demand for deep mineral resource extraction intensifies, optimizing pipeline transportation systems in backfill mining has become increasingly critical. Thus, reducing energy loss while ensuring the filling effect becomes crucial for improving process efficiency. Owing to variations among mines, accurately calculating pipeline resistance loss remains challenging, resulting in significant inaccuracies. The rapid development of Industry 4.0 provides intelligent and data-driven optimization ideas for this challenge. This study introduces a novel pipeline resistance loss prediction framework integrating generative artificial intelligence with a TransKAN model. This study employs generative artificial intelligence to produce physically constrained augmented data, integrates the KAN network’s B-spline basis functions for nonlinear feature extraction, and incorporates the Transformer architecture to capture spatio-temporal correlations in pipeline pressure sequences, enabling precise resistance loss calculation. The experimental data collected from pipeline pressure sensors provides empirical validation for the model. Compared with traditional mathematical formulas, BP neural networks, SVMs, and random forests, the proposed model demonstrates superior performance, achieving an R² value of 0.9644, an RMSE of 0.7126, and an MAE of 0.4703.

One of the hottest topics in current research is the use of different kinds of deep learning models for the identification and forecasting of important indicators in the sintering process. The model's poor prediction accuracy results from the fact that current research approaches mostly focus on chemical composition analysis or single sintering picture data analysis, making it challenging to properly investigate the complementary information between several forms of heterogeneous data. Using a convolutional neural network (CNN) framework, this study builds a Fuzzy Kolmogorov-Arnold Networks (KAN) + Transformer (KanFuzz-Trans) model that offers a precise and effective way to estimate the FeO concentration of sintered ore intelligently. To accurately depict the complex non-linear relationship between FeO content and various influencing factors, the model fully utilises the spatial feature extraction capability of CNN, the network adaptive feature optimisation capability of fuzzy KAN, and the multi-attention mechanism of Transformer. These models can capture the global feature dependency relationship, connection between several influencing elements and FeO content. The prediction accuracy is greatly increased by the model's ability to fully take into account numerous feature information by combining multimodal input, such as text and images, throughout the sintering process. Furthermore, by combining characteristics from several sources of heterogeneous data, this model significantly improves prediction accuracy, as evidenced by its improved performance metrics ( R ² = 0.93, mean squared error = 0.046, root mean squared error = 0.068, mean absolute error = 0.054). The new model presented in this article overcomes the main drawbacks of multimodal information utilisation and prediction accuracy when compared to previous methods. It also offers a fresh concept for achieving intelligent prediction and online detection of FeO content in sintered ores.

In green mine construction and bulk solid-waste utilization, pipeline transport energy efficiency and stope support stability of unclassified cemented tailings backfill exhibit a critical cross-scale mechanical trade-off. However, key process parameters are traditionally determined by empirical heuristics, lacking automated multi-objective quantitative evaluation. To address scale-bar interference and complex boundary degradation in industrial CFD velocity contour maps, this study proposes a physically constitutive-constrained HSV-K-GVF-Snake graded active contour model. Utilizing Bingham three-phase flow priors, the algorithm automatically extracts the cross-sectional flow-core area ratio as a macroscopic geometric proxy for fluid shear-dissipation trends, while three-dimensional micro-porosity is quantitatively resolved via industrial Computed Tomography (CT) reconstruction. Tested across 100 complex contour maps, the proposed segmentation algorithm achieves an Intersection over Union (IoU) of 85.61%, a Dice coefficient of 92.2%, a root mean square error (RMSE) of 0.118, and an average processing time of 1.2908 s per image, reducing GVF iteration steps by over 60% compared with unguided models. Multi-factor analysis across 120 simulated configurations indicated that slurry concentration was the dominant factor affecting the flow-core ratio and the CT-derived porosity indicator. Under the adopted CQI formulation and investigated parameter combinations, the condition with a pipe diameter of 180 mm, a slurry mass concentration of 76%, and a piping aspect ratio of 5 achieved the highest compromise score between the proxy-based transport-efficiency indicator and the porosity-based compactness indicator.

Sintering, as a critical process in iron and steelmaking, has a direct impact on the quality of sintered ore as well as the output and quality of blast furnace ironmaking. The burn-through point (BTP) is a key state parameter for assessing sinter quality, yet it remains difficult to accurately predict in actual production. Given the characteristics of the sintering process—such as multivariable dynamics, large time delays, nonlinearity, and strong coupling—this paper develops an intelligent BTP prediction model based on a hybrid attention FNet architecture.Its diffusion-enhanced preprocessing module can simultaneously collect various sensor time-series data (such as temperature, pressure, etc.) from the windbox, as well as process parameters like ignition temperature and bed thickness.By utilizing an iterative denoising mechanism, it mitigates noise in the windbox data and accomplishes predictive modeling and reconstruction of missing or anomalous sensor data. By integrating an FNet feature fusion network into a hybrid frequency-domain and spatial-domain attention module, the proposed model reduces computational complexity and achieves a 30% improvement in predictive performance over CNN-LSTM models. With an overall data fitting degree of 0.948, the model effectively addresses the scarcity of key abnormal operating condition samples in industrial scenarios and significantly enhances robustness and generalization capability under extreme operating conditions.