电子固废处理析取过程溶解氧浓度预测模型

韩红桂; 任晨昊; 李方昱; 刘峥

doi:10.16383/j.aas.c250658

电子固废处理析取过程溶解氧浓度预测模型

doi: 10.16383/j.aas.c250658 cstr: 32138.14.j.aas.c250658

韩红桂^{1, 2, 3,},
任晨昊^{2, 3,},
李方昱^{1, 3,},
刘峥^{1, 3,}

1.
北京工业大学信息科学技术学院北京 100124
2.
北京工业大学环境科学与工程学院北京 100124
3.
数字社区教育部工程研究中心北京 100124

基金项目: 2030新一代人工智能重大专项项目(2025ZD0123700), 国家自然科学基金(62125301, 62021003, 62373014, 92467205), 北京市科技新星计划(20240484694), 青年北京学者计划(037)资助

详细信息

作者简介:
韩红桂：北京工业大学教授. 主要研究方向为典型资源循环利用过程智能优化控制, 神经网络结构设计与优化. 本文通信作者. E-mail: rechardhan@bjut.edu.cn

任晨昊：北京工业大学环境科学与工程学院博士研究生. 主要研究方向为电子固废析取过程智能建模, 神经网络设计与优化. E-mail: renchenhao@emails.bjut.edu.cn

李方昱：北京工业大学信息科学技术学院教授. 主要研究方向为智能感知与建模, 复杂系统分析与控制, 分布式智能检测优化. E-mail: fangyu.li@bjut.edu.cn

刘峥：北京工业大学信息科学技术学院讲师. 主要研究方向为复杂系统智能特征建模, 智能优化控制. E-mail: liuzheng@bjut.edu.cn

计量
- 文章访问数: 14
- HTML全文浏览量: 8
- 被引次数: 0
出版历程
- 收稿日期: 2025-11-18
- 录用日期: 2026-03-21
- 网络出版日期: 2026-07-22

Prediction Model of Dissolved Oxygen Concentration in Electronic Waste Treatment Disjunction Process

HAN Hong-Gui^{1, 2, 3
,},
REN Chen-Hao^{2, 3
,},
LI Fang-Yu^{1, 3
,},
LIU Zheng^{1, 3
,}

1.
School of Information Science and Technology, Beijing University of Technology, Beijing 100124
2.
College of Environmental Science and Engineering, Beijing University of Techno-logy, Beijing 100124
3.
Engineering Research Center of Digital Community, Ministry of Education, Beijing 100124

Funds: Supported by 2030 New Generation Artificial Intelligence National Science and Technology Major Project (2025ZD0123700), National Natural Science Foundation of China (62125301, 62021003, 62373014, 92467205), Beijing Nova Program (20240484694), and Beijing Youth Scholar (037)

More Information

Author Bio:
HAN Hong-Gui　Professor at Beijing University of Technology. His research interests include intelligent optimization and control of typical resource recycling processes, structure design and optimization of neural networks. Corresponding author of this paper

REN Chen-Hao　Ph.D. candidate at the College of Environmental Science and Engineering, Beijing University of Technology. His research interests include intelligent modeling of electronic waste disjunction process, and neural network design and optimization

LI Fang-Yu　Professor at the School of Information Science and Technology, Beijing University of Technology. His research interests include intelligent perception and modeling, complex system analysis and control, and distributed intelligent detection and optimization

LIU Zheng　Lecturer at the School of Information Science and Technology, Beijing University of Technology. His research interests include intelligent feature modeling of complex systems and intelligent optimization control

摘要

摘要: 电子固废处理析取过程涉及多变量强耦合的多相反应, 其原料异质性使工艺变量间呈现动态因果特性, 导致表征反应平衡的溶解氧浓度难以准确预测. 为解决该问题, 设计一种电子固废处理析取过程溶解氧浓度自适应因果推理预测模型. 首先, 提出一种基于多头注意力的自适应因果发现方法, 更新工艺变量的因果注意力分布, 实现动态因果结构的建模. 其次, 设计一种基于动态掩码的因果干预策略, 抑制工艺参数的虚假因果相关, 实现模型真实因果关系重构. 最后, 构建一种基于元更新的门控网络结构, 调节时空特征与因果特征的门控融合权重, 实现析取过程溶解氧浓度的稳定预测. 在电子固废处理析取过程数据集上验证该预测模型, 实验结果表明, 该模型能够准确描述工艺变量间动态因果关系, 提升溶解氧浓度的预测精度.
- 电子固废处理 /
- 析取过程 /
- 多头注意力 /
- 因果干预 /
- 元更新
Abstract: The electronic waste treatment disjunction process involves multiphase reactions with multivariable strong coupling. The heterogeneity of raw materials results in dynamic causal characteristics in process variables, complicating the accurate prediction of dissolved oxygen concentration at reaction equilibrium. To address the issue, an adaptive causal inference prediction model is designed for the dissolved oxygen concentration in electronic waste treatment disjunction process. First, an adaptive causal discovery method based on multi-head attention is proposed, which updates the causal attention distribution among process variables to model dynamic causal structure. Second, a causal intervention strategy based on dynamic mask is established to suppress spurious causal correlations among process parameters, enabling the reconstruction of true causal relationships in the model. Finally, a gating network structure based on meta-update is constructed to adjust the gating fusion weights of spatiotemporal and causal features, achieving stable prediction of dissolved oxygen concentration in the disjunction process. The proposed prediction model is validated on the electronic waste treatment disjunction process dataset. Experimental results demonstrated that the model can accurately characterize dynamic causal relationships among process variables and improve the prediction accuracy of dissolved oxygen concentration.
- electronic waste treatment /
- disjunction process /
- multi-head attention /
- causal intervention /
- meta-update

HTML全文

图 1 电子固废处理析取过程示意图

Fig. 1 Schematic diagram of electronic waste treatment disjunction process

下载: 全尺寸图片幻灯片

图 2 自适应因果推理预测模型总体框架

Fig. 2 The overall framework of the adaptive causal inference prediction model

下载: 全尺寸图片幻灯片

图 3 ZL数据集的因果关系热力图((a)真实因果关系; (b) CISTGNN; (c)未因果干预ACIPM; (d) ACIPM)

Fig. 3 Causal relationship heatmaps on the ZL dataset ((a) Truth causal relationship; (b) CISTGNN; (c) ACIPM without causal intervention; (d) ACIPM)

下载: 全尺寸图片幻灯片

图 4 酸碱度预测曲线((a) P1; (b) P2)

Fig. 4 Predicted curve of pH value ((a) P1; (b) P2)

下载: 全尺寸图片幻灯片

图 5 溶解氧浓度预测曲线((a) D1; (b) D2)

Fig. 5 Prediction curve of dissolved oxygen concentration ((a) D1; (b) D2)

下载: 全尺寸图片幻灯片

图 6 消融实验结果((a) ZL数据集; (b) TE数据集)

Fig. 6 Ablation experiment results ((a) ZL dataset; (b) TE dataset)

下载: 全尺寸图片幻灯片

表 1 TE数据集上的对比实验

Table 1 Comparison experiments on the TE dataset

	MAE	RMSE	MAPE (%)
DCRNN	10.75	19.55	6.31
STGCN	10.14	23.42	5.48
ASTGCN	9.15	18.21	4.97
CISTGNN	9.60	19.79	5.01
ACIPM	8.49	17.91	4.17

下载: 导出CSV

表 2 ZL数据集上的类GLCM特征

Table 2 GLCM-like features on the ZL dataset

	CISTGNN	无CIDM	ACIPM
正样本灰度均值$\uparrow$	0.50	0.77	0.79
负样本灰度均值$\downarrow$	0.43	0.30	0.21
正样本能量$\uparrow$	0.24	0.42	0.49
负样本能量$\uparrow$	0.14	0.18	0.25
正样本熵$\downarrow$	2.19	1.55	1.42
负样本熵$\downarrow$	2.89	2.62	2.33
Bhattacharyya$\uparrow$	0.21	0.86	1.07

下载: 导出CSV

表 3 ZL数据集上的对比实验

Table 3 Comparison experiments on the ZL dataset

	MAE	RMSE	MAPE (%)
DCRNN	2.75	4.32	5.41
STGCN	2.84	4.17	5.47
ASTGCN	1.93	3.19	5.81
CISTGNN	1.63	3.40	5.32
ACIPM	1.34	2.05	5.02

下载: 导出CSV

表 4 消融实验说明

Table 4 Ablation experiment description

方案	ACDM	CIDM	GNSM
A	$\times$	$\times$	$\times$
B	$\times$	$\times$	$\checkmark$
C	$\checkmark$	$\times$	$\checkmark$
D	$\checkmark$	$\checkmark$	$\times$
E	$\checkmark$	$\checkmark$	$\checkmark$

下载: 导出CSV

参考文献(30)

[1]	Foo G, Kara S, Pagnucco M. Artificial learning for part identification in robotic disassembly through automatic rule generation in an ontology. IEEE Transactions on Automation Science and Engineering, 2023, 20(1): 296−309 doi: 10.1109/TASE.2022.3149242
[2]	Han H G, Liu Y M, Li F Y, Du Y P. Sparse convolutional model with semantic expression for waste electrical appliances recognition. Science China Technological Sciences, 2024, 67(9): 2881−2893 doi: 10.1007/s11431-023-2650-x
[3]	Han H G, Zhang Q Y, Li F Y, Du Y P, Gu Y F, Wu Y F. Metallic product recognition with dual attention and multi-branch residual blocks-based convolutional neural networks. Circular Economy, 2022, 1: Article No. 100014 doi: 10.1016/j.cec.2022.100014
[4]	曹力丰, 阎高伟, 肖舒怡, 董珍柱, 董平. 基于域适应物理信息神经网络的时间序列预测方法. 自动化学报, 2025, 51(6): 1329−1346 Cao Li-Feng, Yan Gao-Wei, Xiao Shu-Yi, Dong Zhen-Zhu, Dong Ping. Time series prediction method based on domain adaptation physics-informed neural network. Acta Automatica Sinica, 2025, 51(6): 1329−1346
[5]	Zhu S Y, Zheng J W, Ma Q L. MR-Transformer: Multiresolution Transformer for multivariate time series prediction. IEEE Transactions on Neural Networks and Learning Systems, 2025, 36(1): 1171−1183 doi: 10.1109/TNNLS.2023.3327416
[6]	Duan C M, Wu Z C, Zhu L, Xu X R, Zhu J M, Wei Z Q, et al. DSTN: Dynamic spatio-temporal network for early fault warning in chemical processes. Knowledge-Based Systems, 2024, 296: Article No. 111892 doi: 10.1016/j.knosys.2024.111892
[7]	Han H G, Sun C X, Wu X L, Yang H Y, Zhao D Z. Self-organizing stacked type-2 fuzzy neural network with rule generation. IEEE Transactions on Neural Networks and Learning Systems, 2025, 36(6): 10128−10142 doi: 10.1109/TNNLS.2025.3544495
[8]	He Z H, Xu L Y, Yu J, Wu X R. Dynamic multi-fusion spatio-temporal graph neural network for multivariate time series forecasting. Expert Systems With Applications, 2024, 241: Article No. 122729 doi: 10.1016/j.eswa.2023.122729
[9]	Fang H, Wang H S, Gao Y, Zhang Y G, Bu J J, Han B, et al. InsGNN: Interpretable spatio-temporal graph neural networks via information bottleneck. Information Fusion, 2025, 119: Article No. 102997 doi: 10.1016/j.inffus.2025.102997
[10]	Qi X Y, Hu W J, Li B S, Han K. STGNN-FAM: A traffic flow prediction model for spatiotemporal graph networks based on fusion of attention mechanisms. Journal of Advanced Transportation, 2023, 2023(1): Article No. 8880530
[11]	Xu Y, Zhang C, Luo Y, Ke W, Zhu Q X, He Y L. Multiattention spatiotemporal fusion graph neural network for chemical process fault diagnosis. IEEE Transactions on Instrumentation and Measurement, 2025, 74: Article No. 3530113
[12]	Yu B, Yin H T, Zhu Z X. Spatio-temporal graph convolutional networks: A deep learning framework for traffic forecasting. In: Proceedings of the 27th International Joint Conference on Artificial Intelligence. Stockholm, Sweden: IJCAI, 2018. 3634−3640
[13]	Qian L P, Zuo Q, Liu H G, Zhu H. Multivariate time series classification based on spatial-temporal attention-dynamic graph neural network. Applied Intelligence, 2025, 55(2): 1−15 doi: 10.1007/s10489-024-06014-8
[14]	Qu Y H, Jia X L, Guo J H, Zhu H R, Wu W B. MSSTGNN: Multi-scaled spatio-temporal graph neural networks for short- and long-term traffic prediction. Knowledge-Based Systems, 2024, 306: Article No. 112716 doi: 10.1016/j.knosys.2024.112716
[15]	Sun H Z, Tang X H, Lu J G, Liu F J. Spatio-temporal graph neural network for traffic prediction based on adaptive neighborhood selection. Transportation Research Record, 2024, 2678(6): 641−655
[16]	Zheng Y, Koh H Y, Jin M, Chi L H, Phan K T, Pan S R. Correlation-aware spatial-temporal graph learning for multivariate time-series anomaly detection. IEEE Transactions on Neural Networks and Learning Systems, 2023, 35(9): 11802−11816 doi: 10.1109/tnnls.2023.3325667/mm1
[17]	Li Z L, Yu J, Zhang G W, Xu L Y. Dynamic spatio-temporal graph network with adaptive propagation mechanism for multivariate time series forecasting. Expert Systems With Applications, 2023, 216: Article No. 119374 doi: 10.1016/j.eswa.2022.119374
[18]	Nauta M, Bucur D, Seifert C. Causal discovery with attention based convolutional neural networks. Machine Learning and Knowledge Extraction, 2019, 1(1): 312−340 doi: 10.3390/make1010019
[19]	Zhu Q H, Chen S W, Guo T, Lv Y S, Du W B. A spatio-temporal approach with self-corrective causal inference for flight delay prediction. IEEE Transactions on Intelligent Transportation Systems, 2024, 25(12): 20820−20831 doi: 10.1109/TITS.2024.3443261
[20]	Miraki A, Parvinian P, Arghandeh R. Electricity demand forecasting at distribution and household levels using explainable causal graph neural network. Energy and AI, 2024, 16: Article No. 100368 doi: 10.1016/j.egyai.2024.100368
[21]	Wang H, Liu R N, Ding S X, Hu Q H, Li Z X, Zhou H K. Causal-trivial attention graph neural network for fault diagnosis of complex industrial processes. IEEE Transactions on Industrial Informatics, 2024, 20(2): 1987−1996 doi: 10.1109/TII.2023.3282979
[22]	Luo X X, Yin W, Li Z. Spatio-temporal graph neural network with hidden confounders for causal forecast. In: Proceedings of the 32nd International Conference on Case-Based Reasoning. Yucatán, Mexico: Springer, 2024. 157−169
[23]	Song P Y, Zhao C H, Huang B. Addressing heterogeneous time-frequency causality: Source consistency exploring for industrial root cause alignment and diagnosis. IEEE Transactions on Cybernetics, 2025, 55(3): 1107−1120 doi: 10.1109/TCYB.2024.3515174
[24]	Song P Y, Zhao C H, Huang B. MPGE and rootrank: A sufficient root-cause characterization and quantification framework for industrial process faults. Neural Networks, 2023, 161: 397−417 doi: 10.1016/j.neunet.2023.01.030
[25]	Li Y G, Yu R, Shahabi C, Liu Y. Diffusion convolutional recurrent neural network: Data-driven traffic forecasting. In: Proceedings of the 6th International Conference on Learning Representations. Vancouver, Canada: ICLR, 2018. 52−57
[26]	Guo S N, Lin Y F, Feng N, Song C, Wan H Y. Attention based spatial-temporal graph convolutional networks for traffic flow forecasting. In: Proceedings of the AAAI Conference on Artificial Intelligence. Hawaii, USA: AAAI, 2019. 922−929
[27]	Portes L, Pirot G, Nzikou M M, Giraud J, Lindsay M, Jessell M, et al. Feature-fusion enhanced t-SNE image atlas for geophysical features discovery. Scientific Reports, 2025, 15(1): Article No. 17152 doi: 10.1038/s41598-025-01333-3
[28]	Li J C, Fan Y, Wang Z Y, Wang Y J. A graph neural network with dual-stage feature aggregation for industrial soft sensors. IEEE Transactions on Instrumentation and Measurement, 2025, 7: Article No. 3557410 doi: 10.1109/tim.2025.3606039
[29]	Zheng J L, Saxena D, Cao J N. COIN-GNN: Inductive spatial-temporal prediction for continuous distribution shifts via graph neural networks. IEEE Transactions on Knowledge and Data Engineering, 2025, 37(11): 6320−6333 doi: 10.1109/TKDE.2025.3606629
[30]	Wang Y L, Wang X L, Zhou J Y, Yang C H, Yang Y. Long sequence multivariate time-series forecasting for industrial processes using SASGNN. IEEE Transactions on Industrial Informatics, 2024, 20(10): 12407−12417 doi: 10.1109/TII.2024.3424214