Discrete Workshop Production Bottleneck Prediction with CNN-LSTM Spatio-temporal Network Integrated with Dual Attention Mechanism

Yihan Shi; Xu Zhang; Cunbo Zhuang; Jinshan Liu; Jiaxiu Wang; Liansheng Sun

doi:10.3901/JME.260227

Discrete Workshop Production Bottleneck Prediction with CNN-LSTM Spatio-temporal Network Integrated with Dual Attention Mechanism

投稿的翻译标题: 集成双重注意力机制 CNN-LSTM 时空网络的离散车间生产瓶颈预测

Yihan Shi
, Xu Zhang
, Cunbo Zhuang^*
, Jinshan Liu
, Jiaxiu Wang
, Liansheng Sun

^*此作品的通讯作者

机械与车辆学院

Beijing Institute of Technology
Hebei Province Key Laboratory of Big Data Science and Intelligent Technology
Ltd.

科研成果: 期刊稿件 › 文章 › 同行评审

摘要

In the assembly pull production mode, the discrete manufacturing workshop, which is characterized by decentralized production tasks, flexible equipment layout, and complex processes, serves as the core carrier of the multi-factory(shop) production mode. However, the dynamic drift of bottleneck units in both temporal and spatial dimensions has become a critical challenge that limits the improvement of production efficiency and resource utilization. Thus, investigating the bottleneck prediction problem in discrete workshops is crucial for enhancing overall production efficiency in multi-factory settings. To address this, a spatiotemporal network prediction model integrating dual attention mechanisms, CNN-LSTM-DAM (Convolutional neural network-long short term memory-dual attention mechanism), is proposed. First, a composite definition-based bottleneck identification model is constructed, considering the multi-attribute coupling of bottleneck units. Second, the identified historical quasi-bottleneck data are input into the spatial feature perceiver integrated with CNN and spatial attention mechanism, as well as the temporal feature perceiver integrated with LSTM and state attention mechanism, as auxiliary data to further enhance the model's ability to capture spatial and temporal information in production sequence data. Finally, through ablation experiments, the proposed model is compared with other LSTM variants, such as gated recurrent unit (GRU) and bidirectional long short term memory (BiLSTM). The results verify the model's accuracy and effectiveness in predicting bottleneck units and their drift trends within a given time delay.

投稿的翻译标题	集成双重注意力机制 CNN-LSTM 时空网络的离散车间生产瓶颈预测
源语言	英语
期刊	Jixie Gongcheng Xuebao/Chinese Journal of Mechanical Engineering
卷	62
期	5
DOI	https://doi.org/10.3901/JME.260227
出版状态	已出版 - 5 3月 2026

访问文件

10.3901/JME.260227

其它文件与链接

链接到 Scopus 的出版物

引用此

@article{19dce9f52b994674a6e90095520492a2,

title = "Discrete Workshop Production Bottleneck Prediction with CNN-LSTM Spatio-temporal Network Integrated with Dual Attention Mechanism",

abstract = "In the assembly pull production mode, the discrete manufacturing workshop, which is characterized by decentralized production tasks, flexible equipment layout, and complex processes, serves as the core carrier of the multi-factory(shop) production mode. However, the dynamic drift of bottleneck units in both temporal and spatial dimensions has become a critical challenge that limits the improvement of production efficiency and resource utilization. Thus, investigating the bottleneck prediction problem in discrete workshops is crucial for enhancing overall production efficiency in multi-factory settings. To address this, a spatiotemporal network prediction model integrating dual attention mechanisms, CNN-LSTM-DAM (Convolutional neural network-long short term memory-dual attention mechanism), is proposed. First, a composite definition-based bottleneck identification model is constructed, considering the multi-attribute coupling of bottleneck units. Second, the identified historical quasi-bottleneck data are input into the spatial feature perceiver integrated with CNN and spatial attention mechanism, as well as the temporal feature perceiver integrated with LSTM and state attention mechanism, as auxiliary data to further enhance the model's ability to capture spatial and temporal information in production sequence data. Finally, through ablation experiments, the proposed model is compared with other LSTM variants, such as gated recurrent unit (GRU) and bidirectional long short term memory (BiLSTM). The results verify the model's accuracy and effectiveness in predicting bottleneck units and their drift trends within a given time delay.",

keywords = "CNN-LSTM, bottleneck prediction, dual attention mechanism, multi-attribute bottleneck, multi-step time series forecasting",

author = "Yihan Shi and Xu Zhang and Cunbo Zhuang and Jinshan Liu and Jiaxiu Wang and Liansheng Sun",

year = "2026",

month = mar,

day = "5",

doi = "10.3901/JME.260227",

language = "English",

volume = "62",

journal = "Jixie Gongcheng Xuebao/Chinese Journal of Mechanical Engineering",

issn = "0577-6686",

publisher = "Chinese Mechanical Engineering Society",

number = "5",

}

TY - JOUR

T1 - Discrete Workshop Production Bottleneck Prediction with CNN-LSTM Spatio-temporal Network Integrated with Dual Attention Mechanism

AU - Shi, Yihan

AU - Zhang, Xu

AU - Zhuang, Cunbo

AU - Liu, Jinshan

AU - Wang, Jiaxiu

AU - Sun, Liansheng

PY - 2026/3/5

Y1 - 2026/3/5

N2 - In the assembly pull production mode, the discrete manufacturing workshop, which is characterized by decentralized production tasks, flexible equipment layout, and complex processes, serves as the core carrier of the multi-factory(shop) production mode. However, the dynamic drift of bottleneck units in both temporal and spatial dimensions has become a critical challenge that limits the improvement of production efficiency and resource utilization. Thus, investigating the bottleneck prediction problem in discrete workshops is crucial for enhancing overall production efficiency in multi-factory settings. To address this, a spatiotemporal network prediction model integrating dual attention mechanisms, CNN-LSTM-DAM (Convolutional neural network-long short term memory-dual attention mechanism), is proposed. First, a composite definition-based bottleneck identification model is constructed, considering the multi-attribute coupling of bottleneck units. Second, the identified historical quasi-bottleneck data are input into the spatial feature perceiver integrated with CNN and spatial attention mechanism, as well as the temporal feature perceiver integrated with LSTM and state attention mechanism, as auxiliary data to further enhance the model's ability to capture spatial and temporal information in production sequence data. Finally, through ablation experiments, the proposed model is compared with other LSTM variants, such as gated recurrent unit (GRU) and bidirectional long short term memory (BiLSTM). The results verify the model's accuracy and effectiveness in predicting bottleneck units and their drift trends within a given time delay.

AB - In the assembly pull production mode, the discrete manufacturing workshop, which is characterized by decentralized production tasks, flexible equipment layout, and complex processes, serves as the core carrier of the multi-factory(shop) production mode. However, the dynamic drift of bottleneck units in both temporal and spatial dimensions has become a critical challenge that limits the improvement of production efficiency and resource utilization. Thus, investigating the bottleneck prediction problem in discrete workshops is crucial for enhancing overall production efficiency in multi-factory settings. To address this, a spatiotemporal network prediction model integrating dual attention mechanisms, CNN-LSTM-DAM (Convolutional neural network-long short term memory-dual attention mechanism), is proposed. First, a composite definition-based bottleneck identification model is constructed, considering the multi-attribute coupling of bottleneck units. Second, the identified historical quasi-bottleneck data are input into the spatial feature perceiver integrated with CNN and spatial attention mechanism, as well as the temporal feature perceiver integrated with LSTM and state attention mechanism, as auxiliary data to further enhance the model's ability to capture spatial and temporal information in production sequence data. Finally, through ablation experiments, the proposed model is compared with other LSTM variants, such as gated recurrent unit (GRU) and bidirectional long short term memory (BiLSTM). The results verify the model's accuracy and effectiveness in predicting bottleneck units and their drift trends within a given time delay.

KW - CNN-LSTM

KW - bottleneck prediction

KW - dual attention mechanism

KW - multi-attribute bottleneck

KW - multi-step time series forecasting

UR - https://www.scopus.com/pages/publications/105036639088

U2 - 10.3901/JME.260227

DO - 10.3901/JME.260227

M3 - Article

AN - SCOPUS:105036639088

SN - 0577-6686

VL - 62

JO - Jixie Gongcheng Xuebao/Chinese Journal of Mechanical Engineering

JF - Jixie Gongcheng Xuebao/Chinese Journal of Mechanical Engineering

IS - 5

ER -

Discrete Workshop Production Bottleneck Prediction with CNN-LSTM Spatio-temporal Network Integrated with Dual Attention Mechanism

摘要

访问文件

其它文件与链接

学术指纹

引用此