基于时空图卷积的汽车配件供应链需求预测与仿真分析

doi:10.16182/j.issn1004731x.joss.25-0659

系统仿真学报 ›› 2025, Vol. 37 ›› Issue (12): 3060-3074.doi: 10.16182/j.issn1004731x.joss.25-0659

• 论文 • 上一篇

基于时空图卷积的汽车配件供应链需求预测与仿真分析

李孝斌¹^,², 胡冰¹^,², 尹超¹^,², 李波¹^,², 马军³

^1.重庆大学机械与运载工程学院，重庆 400044
^2.重庆大学高端装备机械传动全国重点实验室，重庆 400044
^3.重庆铁马工业集团有限公司，重庆 400020

收稿日期:2025-07-09 修回日期:2025-11-02 出版日期:2025-12-26 发布日期:2025-12-24
通讯作者: 尹超
第一作者简介:李孝斌(1987-)，男，副教授，博士，研究方向为智能制造与网络协同制造。
基金资助:
国家重点研发计划(2022YFB3305603);国家自然科学基金(52575561);重庆市科技创新重大研发项目(CSTB2024TIAD-STX0029)

Spatiotemporal Graph Convolution-based Demand Forecasting and Simulation Analysis for Automotive Parts Supply Chain

Li Xiaobin¹^,², Hu Bing¹^,², Yin Chao¹^,², Li Bo¹^,², Ma Jun³

^1.College of Mechanical and Vehicle Engineering, Chongqing University, Chongqing 400044, China
^2.State Key Laboratory of Mechanical Transmission for Advanced Equipment, Chongqing University, Chongqing 400044, China
^3.Chongqing Tiema Industries Group Co. , Ltd. , Chongqing 400020, China

Received:2025-07-09 Revised:2025-11-02 Online:2025-12-26 Published:2025-12-24
Contact: Yin Chao

摘要/Abstract

摘要：

针对汽车售后配件供应网络运行复杂，配件需求预测精度不足、响应不及时，售后服务效率较低等问题，提出一种基于时空图卷积的汽车配件供应链需求预测方法。将汽车配件销售网络的销售数据构建为异构图，融合配件销量、金额等节点特征构建多维节点依赖关系；设计图卷积神经网络的节点迭代更新机制，结合长短时记忆神经网络捕获时序特征，利用时空注意力将时间特征与空间特征融入更新后的节点。仿真实验表明：该方法最优RMSE达到2.09，在动态扰动下误差峰值控制在3.44以内，恢复步数不超过50步，验证了该方法在复杂供应链环境下的预测效果和动态适应性。

关键词: 时空图神经网络, 需求预测, 配件供应链, 仿真分析

Abstract:

To address complex automotive after-sales parts supply network operations with insufficient demand forecasting accuracy, slow response, and low service efficiency, this study proposed a spatiotemporal graph convolution-based method for automotive parts supply chain demand forecasting. Sales network data of the automotive parts sales network was constructed as a heterogeneous graph, integrating node features like parts sales volume and value to build multi-dimensional node dependencies. A node update mechanism of the graph convolutional neural network was designed, combined with long short-term memory neural networks to capture temporal features, using spatiotemporal attention to integrate temporal and spatial features into updated nodes. Simulation experiments show the method achieves an optimal RMSE of 2.09, error peaks within 3.44 under dynamic disturbances, and recovery within 50 steps, validating its forecasting performance and dynamic adaptability in complex supply chain environments.

Key words: spatiotemporal graph neural network, demand forecasting, parts supply chain, simulation analysis

中图分类号:

TP391.9

李孝斌,胡冰,尹超等 . 基于时空图卷积的汽车配件供应链需求预测与仿真分析[J]. 系统仿真学报, 2025, 37(12): 3060-3074.

Li Xiaobin,Hu Bing,Yin Chao,et al . Spatiotemporal Graph Convolution-based Demand Forecasting and Simulation Analysis for Automotive Parts Supply Chain[J]. Journal of System Simulation, 2025, 37(12): 3060-3074.

图/表 13

图1

图2

图3

图4

表1

汽车配件部分数据集

供应商ID	销售商ID	制造商ID	日期	销量	总额/万元
1 154	168	197	2023/1/1	16	489.44
1 649	630	126	2023/1/3	12	1 628.64
1 649	630	119	2023/1/5	9	2 337.93
2 202	28	35	2023/1/7	9	1 320.84
$⋯$

表1

图5

图6

图7

表2

图8

图9

图10

表3

参考文献 15

[1]	Taylor J W, McSharry P E. Short-term Load Forecasting Methods: An Evaluation Based on European Data[J]. IEEE Transactions on Power Systems, 2007, 22(4): 2213-2219.
[2]	Du Shengdong, Li Tianrui, Yang Yan, et al. Deep Air Quality Forecasting Using Hybrid Deep Learning Framework[J]. IEEE Transactions on Knowledge and Data Engineering, 2021, 33(6): 2412-2424.
[3]	Chen Li, Li Gang, Wu Chenguang, et al. Forecasting Seasonal Tourism Demand Using a Multiseries Structural Time Series Method[J]. Journal of Travel Research, 2019, 58(1): 92-103.
[4]	Zhuang E H, Li C L, Wong K C. Discovery of Temporal Associations in Multivariate Time Series[J]. IEEE Transactions on Knowledge and Data Engineering, 2014, 26(12): 2969-2982.
[5]	Farizal, Qaradhawi Yusuf; Caesario Isak Cornelis, et al. Fast Moving Product Demand Forecasting Model with Multi Linear Regression[J]. AIP Conference Proceedings, 2020, 2227(1): 040028.
[6]	Feizabadi Javad. Machine Learning Demand Forecasting and Supply Chain Performance[J]. International Journal of Logistics Research and Applications, 2022, 25(2): 119-142.
[7]	Weng Tingyu, Liu Wenyang, Xiao Jun. Supply Chain Sales Forecasting Based on lightGBM and LSTM Combination Model[J]. Industrial Management & Data Systems, 2019, 120(2): 265-279.
[8]	Svetunkov I, Boylan J E. State-space ARIMA for Supply-chain Forecasting[J]. International Journal of Production Research, 2020, 58(3): 818-827.
[9]	Yang Qin, Chen Yun. Auto Parts Demand Forecasting Based on Nonnegative Variable Weight Combination Model in Auto Aftermarket[C]//2012 International Conference on Management Science & Engineering 19th Annual Conference Proceedings. Piscataway: IEEE, 2012: 817-822.
[10]	Huang Kai, Wang Jian. Short-term Auto Parts Demand Forecasting Based on EEMD-CNN-BiLSTM-attention-combination Model[J]. Journal of Intelligent & Fuzzy Systems, 2023, 45(4): 5449-5465.
[11]	魏静敏, 唐加福. 供应链中供需双方订货批量模型的博弈分析[C]//2006中国控制与决策学术年会论文集. 沈阳: «控制与决策»编辑委员会, 2006: 1020-1025.
	Wei Jingmin, Tang Jiafu. Game Analysis on Lot Sizing Model for Buyer-supplier in a Supply Chain[C]//Proceeding of 2006 Chinese Control and Decision Conference. Shenyang: Editorial Committee of Control and Decision, 2006: 989-994.
[12]	Cigolini Roberto, Pero Margherita, Rossi Tommaso, et al. Linking Supply Chain Configuration to Supply Chain Perfrmance: A Discrete Event Simulation Model[J]. Simulation Modelling Practice and Theory, 2014, 40: 1-11.
[13]	Gao Boquan, Cao Jingjing. GCN-ARIMA Based Sales Demand Prediction[C]//7th International Conference on Human Centered Computing. Cham: Springer Nature Switzerland, 2022: 50-60.
[14]	Sun Junyan, Tang Jianming, Fu Weiping, et al. Hybrid Modeling and Empirical Analysis of Automobile Supply Chain Network[J]. Physica a: Statistical Mechanics and Its Applications, 2017, 473: 377-389.
[15]	Zhang Pengyu, Zhang Yong, Wang Jingcheng, et al. MVMA-GCN: Multi-view Multi-layer Attention Graph Convolutional Networks[J]. Engineering Applications of Artificial Intelligence, 2023, 126, Part A: 106717.

模型	RMSE	MSE
Amount-Influence-Adj	3.684	13.571 9
Single Adjacency Matrix	3.599	12.952 8
Single Adjacency Matrix	3.784	14.318 0
Without TGA	2.485	6.175 2
Without GCN	2.674	7.150 2
AutoTG-LSTM	2.094	4.384 8

场景	模型	RMSE增幅/%	误差峰值	恢复步数	超调比/%
供应中断	完整模型	8.2	3.44	50	0.67
	去GCN模块	21.3	4.69	138	0.75
	去TGA模块	18.6	4.07	84	0.64
需求突增	完整模型	9.1	3.21	44	0.56
	去GCN模块	18.7	4.39	105	0.63
	去TGA模块	17.9	3.87	98	0.58
信息滞后	完整模型	10.4	3.05	26	0.48
	去GCN模块	19.5	4.18	74	0.55
	去TGA模块	16.8	3.54	27	0.59

基于时空图卷积的汽车配件供应链需求预测与仿真分析

Spatiotemporal Graph Convolution-based Demand Forecasting and Simulation Analysis for Automotive Parts Supply Chain

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 13

参考文献 15

相关文章 15

编辑推荐

Metrics

本文评价

[1]	赵曙光, 陈策, 谢孝昌, 栗付平, 韩劲. 基于直线模型提升同轴磁齿轮设计效率的研究[J]. 系统仿真学报, 2025, 37(6): 1574-1581.
[2]	赵振宇, 马旭, 包格日乐图. 基于风速预测模型的风电一次调频仿真研究[J]. 系统仿真学报, 2022, 34(10): 2233-2243.
[3]	孙小庆, 忽伟, 刘宇程, 王智磊, 胡俊. 超磁致伸缩定隔一体驱动系统集成设计与仿真[J]. 系统仿真学报, 2021, 33(2): 358-365.
[4]	张云景, 王鑫鑫, 王洋, 汤光明. 一种基于证据理论的战时备件需求预测方法[J]. 系统仿真学报, 2020, 32(2): 314-322.
[5]	彭本红, 王圆缘, 屠羽. 政府规制视角下EPR契约设计及仿真分析[J]. 系统仿真学报, 2019, 31(9): 1811-1818.
[6]	刘闻, 王晓路, 汪宏昇, 张恒, 王长庆. 基于Agent的低轨预警卫星星座探测仿真分析[J]. 系统仿真学报, 2019, 31(11): 2413-2421.
[7]	周鹏飞, 邢小伟. 回字形布置的双轨轮小车自动化码头仿真[J]. 系统仿真学报, 2018, 30(9): 3264-3273.
[8]	井然, 冯华, 王永庚, 宋海平, 肖艳军. 电池极片电磁热辊的温度控制的分析与仿真[J]. 系统仿真学报, 2018, 30(6): 2245-2250.
[9]	周妍. 列车运行图与车站作业计划协同编制仿真分析[J]. 系统仿真学报, 2018, 30(10): 3665-3670.
[10]	彭本红, 谷晓芬, 鲁倩. 服务型制造项目治理中利益相关者演化博弈研究[J]. 系统仿真学报, 2017, 29(3): 595-608.
[11]	张昊, 陈自力, 邱金刚. 软翼飞行器多体动力学建模与运动特性分析[J]. 系统仿真学报, 2016, 28(8): 1841-1845.
[12]	戚宗锋, 韩山, 李建勋. 基于粗糙集的雷达抗干扰性能评估指标体系研究[J]. 系统仿真学报, 2016, 28(2): 335-342.
[13]	吴先荣, 魏会明, 苏学丰, 吴帆. 基于MAAP5的一类严重事故仿真[J]. 系统仿真学报, 2016, 28(1): 200-204.
[14]	张敏, 郭鑫鑫, 张驰. 积水条件下考虑行车安全的车辙长度仿真分析[J]. 系统仿真学报, 2015, 27(4): 747-754.
[15]	舒彩霞, 严磊, 李磊, 韩彩锐, 蒋亚军, 丁幼春, 廖庆喜. 油菜割晒机液压驱动系统管道设计与仿真优化[J]. 系统仿真学报, 2015, 27(12): 3087-3095.