Process-Informed Dynamic Graph WaveNet for Industrial Sintering Processes Forecasting

DOI: https://doi.org/10.62517/jike.202604129

Author(s)

Zhili Zhang1, Liya Wang1, Jie Li2

Affiliation(s)

1College of Science, North China University of Science and Technology, Tangshan, China 2College of Metallurgy and Energy, North China University of Science and Technology, Tangshan, China

Abstract

Accurate multi-step forecasting of key state parameters is a critical prerequisite for optimizing energy consumption and reducing emissions in the sintering process, which is characterized by significant nonlinearity and large time-varying lags. However, applying general Spatiotemporal Graph Neural Networks faces a critical mismatch: unbounded dot-product attention lacks robustness to industrial noise, and purely data-driven structures often violate physical consistency. To address this, we propose a Process-Informed Dynamic Graph WaveNet. The model incorporates a Global Feature Broadcasting mechanism to integrate system-wide control variables. Crucially, it employs a Decoupled Two-Stream Topology combining a static adaptive graph initialized with physical time-delay priors and a noise-robust dynamic graph constructor using a bounded Radial Basis Function kernel. This architecture strictly constrains node connectivity to suppress interference while respecting material flows. Across two real-world sintering datasets, the proposed model attains the lowest prediction errors on the more difficult Sinter-B dataset and matches the performance of top-performing baselines on Sinter-A, highlighting its adaptability and reliability under varying industrial conditions.

Keywords

Industrial Internet of Things; Sintering Process; Spatiotemporal Graph Neural Networks; Time Series Forecasting

References

[1] Wen W, Deng Z, Ma X, et al. Analysis of the synergistic benefits of typical technologies for pollution reduction and carbon reduction in the iron and steel industry in the Beijing–Tianjin–Hebei region[J]. Scientific Reports, 2024, 14(1): 12413. [2] Yan F, Zhang X, Yang C, et al. Data‐driven modelling methods in sintering process: Current research status and perspectives[J]. The Canadian Journal of Chemical Engineering, 2023, 101(8): 4506-4522. [3] Du S, Ma X, Fan H, et al. Intelligent prediction and soft-sensing of comprehensive production indicators for iron ore sintering: A review[J]. Computers in Industry, 2025, 165: 104215. [4] Xiong D L, Ning H Y, Xie M, et al. A Predictive Model for Sintering Ignition Temperature Based on a CNN-LSTM Neural Network with an Attention Mechanism[J]. Processes, 2024, 12(10): 2185. [5] Li Y, Yang C, Sun Y. Dynamic time features expanding and extracting method for prediction model of sintering process quality index[J]. IEEE Transactions on Industrial Informatics, 2021, 18(3): 1737-1745. [6] Yu Q, Yang G, Wang X, et al. A review of time series forecasting and spatio-temporal series forecasting in deep learning: Q. Yu et al[J]. The Journal of Supercomputing, 2025, 81(10): 1160. [7] Zhang Z, Wang X, Tan X, et al. DSSA-TCN: Exploiting adaptive sparse attention and diffusion graph convolutions in temporal convolutional networks for traffic flow forecasting[J]. PloS one, 2025, 20(11): e0336787. [8] Zhu K, Zhao C. Dynamic graph-based adaptive learning for online industrial soft sensor with mutable spatial coupling relations[J]. IEEE Transactions on Industrial Electronics, 2022, 70(9): 9614-9622. [9] Zhou H, Zhao J P, Loo C E, et al. Numerical modeling of the iron ore sintering process[J]. ISIJ international, 2012, 52(9): 1550-1558. [10] Li Y, Yu R, Shahabi C, et al. Diffusion Convolutional Recurrent Neural Network: Data-Driven Traffic Forecasting[C]//International Conference on Learning Representations. 2018. [11] Zheng C, Fan X, Wang C, et al. Gman: A graph multi-attention network for traffic prediction[C]//Proceedings of the AAAI conference on artificial intelligence. 2020, 34(01): 1234-1241. [12] Xia Y, Liang Y, Wen H, et al. Deciphering spatio-temporal graph forecasting: A causal lens and treatment[J]. Advances in Neural Information Processing Systems, 2023, 36: 37068-37088. [13] Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need[J]. Advances in neural information processing systems, 2017, 30. [14] Zeng A, Chen M, Zhang L, et al. Are transformers effective for time series forecasting?[C]//Proceedings of the AAAI conference on artificial intelligence. 2023, 37(9): 11121-11128. [15] Bai L, Yao L, Li C, et al. Adaptive graph convolutional recurrent network for traffic forecasting[J]. Advances in neural information processing systems, 2020, 33: 17804-17815. [16] Li F, Feng J, Yan H, et al. Dynamic graph convolutional recurrent network for traffic prediction: Benchmark and solution[J]. ACM Transactions on Knowledge Discovery from Data, 2023, 17(1): 1-21. [17] Wu Z, Pan S, Long G, et al. Graph wavenet for deep spatial-temporal graph modeling[C]//Proceedings of the 28th International Joint Conference on Artificial Intelligence. 2019: 1907-1913.