面向大型工程施工企业的工程物资供应链集成优化研究——以某铁路工程施工集团为例 

摘要：工程物资采购成本通常占大型铁路及基础设施工程项目总成本的六成以上，其供应链管理水平直接决定项目的成本、工期与质量。当前大型工程施工企业普遍受困于多项目并行导致的需求波动剧烈、库存积压与短缺并存、供应商管理粗放以及跨部门信息孤岛等问题，供应链整体协同效率偏低。围绕这些痛点，本文以国内某特大型铁路工程施工集团（以下简称 S 集团）为案例背景，构建 “需求预测 — 库存优化 — 供应商评价 — 协同配送” 四位一体的工程物资供应链集成优化框架。方法层面，采用 ARIMA 与梯度提升树 XGBoost 融合、并以非负最小二乘学习组合权重的混合模型开展物资需求预测；在服务水平约束下建立改进 (s, S) 库存策略与安全库存模型；运用熵权 - 逼近理想解排序法（TOPSIS）构建八维供应商综合评价体系；并设计数字化供应链集成管理平台架构以支撑集中采购与协同配送。基于案例企业业务特征构造的仿真数据集验证表明，混合预测模型将测试期需求预测的平均绝对百分比误差由 ARIMA 的 11.55% 降至 7.54%，相对改进 34.66%；优化后库存周转率由 8.336 次 / 年提升至 10.74 次 / 年，缺货率由 7.78% 降至 2.16%，持货成本下降 34.51%，综合采购成本指数下降 15.75%，准时交付率由 89.09% 提升至 95.82%。灵敏度分析进一步揭示，年化总成本关于服务水平与安全库存系数呈凸碗形，最优区间落在服务水平 0.95、安全库存系数 1.21 附近。研究为大型工程施工企业工程物资供应链的数字化与精益化管理提供了可复用的方法框架与决策依据。

关键词：工程物资供应链；需求预测；库存优化；供应商评价；协同配送；数字化管理

正文内容 / Content：

可下载并阅读全文PDF，请按照本文版权许可使用。

Download the full text PDF for viewing and using it according to the license of this paper.

Integrated Optimization of Engineering Materials Supply Chains for Large Construction Enterprises: A Case Study of a Railway Construction Group

Abstract: Materials procurement typically accounts for over 60% of the total cost of large-scale railway and infrastructure projects, so the maturity of its supply chain management directly shapes project cost, schedule and quality. Large construction enterprises commonly suffer from severe demand fluctuations caused by concurrent multi-project operations, the coexistence of overstock and stockouts, extensive supplier management, and cross-departmental information silos, which together depress overall coordination efficiency. To tackle these problems, this paper takes a large domestic railway construction group (hereinafter Group S) as the case background and builds an integrated optimization framework for the engineering materials supply chain that unifies demand forecasting, inventory optimization, supplier evaluation and collaborative distribution. A hybrid model that fuses ARIMA with the gradient boosting tree XGBoost and learns combination weights via non-negative least squares is used for materials demand forecasting; an improved (s, S) inventory policy with a safety-stock model under a service-level constraint is established; an entropy-weight TOPSIS method constructs an eight-dimensional supplier evaluation system; and a digital integrated management platform architecture supports centralized procurement and collaborative distribution. Validation on a simulated dataset built from the case enterprise's business characteristics shows that the hybrid model reduces the test-period mean absolute percentage error from 11.55% (ARIMA) to 7.54%, a relative improvement of 34.66%. After optimization, inventory turnover rises from 8.336 to 10.74 times per year, the stockout rate drops from 7.78% to 2.16%, holding cost falls by 34.51%, the comprehensive procurement cost index decreases by 15.75%, and the on-time delivery rate improves from 89.09% to 95.82%. A sensitivity analysis reveals that the annualized total cost is convex in the service level and the safety-stock coefficient, with the optimum near a service level of 0.95 and a coefficient of 1.21. The study offers a reusable methodological framework and decision basis for the digital and lean management of engineering materials supply chains.

Keywords: engineering materials supply chain; demand forecasting; inventory optimization; supplier evaluation; collaborative distribution; digital management

参考文献/References

1. 郑学东. 施工企业物资采购成本控制问题探析[J]. 铁道物资科学管理，2012（5）：26-28.
2. 方鹏程. 企业物流集成供应链管理模式建设策略[J]. Journal of Contemporary Social Sciences Research, 2024, 1(2): 35-38.
3. Li Y, Sun T, Han R. Influence of contractual and relational governance on green supply chain management practices and sustainability performance in the construction industry[J]. Engineering, Construction and Architectural Management, 2025.
4. Chopra S, Meindl P. Supply chain management: strategy, planning and operations[M]. Upper Saddle River: Prentice Hall, 2000.
5. Cachon G P, Fisher M. Supply chain inventory management and the value of shared information[J]. Management Science, 2000, 46(8): 1032-1048.
6. O'Brien W J. Construction supply-chain management: a vision for advanced coordination, costing, and control[C]//NSF Berkeley-Stanford Construction Research Workshop, 1999.
7. 吴树贵. 供应链数字化转型初探[J]. 信息技术与网络安全，2020，39（1）：1-5.
8. 唐隆基，潘永刚，张婷. 工业互联网赋能供应链数字化转型研究[J]. 供应链管理，2020（7）：45-53.
9. 张佳慧，马继东. 木材首饰商品零售企业的ARIMA需求预测研究[J]. 森林工程，2016，32（4）：74-78.
10. Sagi O, Rokach L. Approximating XGBoost with an interpretable decision tree[J]. Information Sciences, 2021, 572: 522-542.
11. Qiu Y, Zhou J, Khandelwal M, et al. Performance evaluation of hybrid WOA-XGBoost, GWO-XGBoost and BO-XGBoost models to predict blast-induced ground vibration[J]. Engineering with Computers, 2022, 38(5): 4145-4162.
12. Dong J, Chen Y, Yao B, et al. A neural network boosting regression model based on XGBoost[J]. Applied Soft Computing, 2022, 125: 109067.
13. Minner S. Multiple-supplier inventory models in supply chain management: a review[J]. International Journal of Production Economics, 2003, 81-82: 265-279.
14. Wang C, Shi J, Yang Y, et al. Two methods with bidirectional similarity for optimal selections of supplier portfolio and supplier substitute based on TOPSIS and IFS[J]. IEEE Access, 2024, 12: 1234-1250.
15. Beikkhakhian Y, Javanmardi M, Karbasian M, et al. The application of ISM model in evaluating agile suppliers selection criteria and ranking suppliers using fuzzy TOPSIS-AHP methods[J]. Expert Systems with Applications, 2015, 42(15-16): 6224-6236.
16. 艾欣，秦珺晗，胡寰宇. 基于最优最劣法-熵权-逼近理想解排序法的电网安全与效益综合评价[J]. 现代电力，2021，38（3）：233-241.
17. 金玲玲. 工程物资供应链管理模式及应用研究[J]. 现代经济管理，2024，5（1）：61-63.
18. Lu Y. The current situation and optimization path of enterprise supply chain management construction in the context of digitalization[J]. Advances in Economics, Management and Political Sciences, 2024, 115: 1-8.
19. Jeong S, Park S, Kim S. Analysis of the recall demand pattern of imported cars and application of ARIMA demand forecasting model[J]. Journal of Society of Korea Industrial and Systems Engineering, 2020, 43(4): 93-100.
20. Nafei A, Azizi S P, Edalatpanah S A, et al. Smart TOPSIS: a neural network-driven TOPSIS with neutrosophic triplets for green supplier selection in sustainable manufacturing[J]. Expert Systems with Applications, 2024, 255: 124744.
21. 赵晓飞，鲁楠，李明. 农产品供应链数字化转型：理论框架与实现路径[J]. 云南社会科学，2022（6）：55-63.