摘要
在“双碳”与数字化转型背景下,塑料制品生产供应链面临原料价格波动、能耗偏高和运输不确定等多重风险。为提高风险管控的规范性与可量化程度,本文基于多源大数据,构建涵盖原料采购、生产运行、运输调度与市场波动的风险指标体系。研究采用随机森林提取关键特征,结合长短期记忆网络开展时序预测,形成静态特征识别与动态序列建模相结合的风险预测模型。结果显示,该模型在预测精度与稳定性方面优于单一算法,能够识别高风险时段及其主要驱动因素。在此基础上,构建成本-风险双目标优化模型,模拟不同策略下的资源配置方式与风险响应路径,为塑料制品企业的供应链稳定运行与决策优化提供量化依据。
Under the background of the"dual carbon"strategy and digital transformation,the supply chain of plastic products production faces multiple risks such as raw material price fluctuations,high energy consumption,and transportation uncertainty.To enhance the scientific management of these risks,this paper constructs a risk indicator system covering raw material procurement,production operation,transportation scheduling,and market volatility based on multi-source big data.The study employs a random forest algorithm to identify key features and integrates it with a long short-term memory(LSTM)network for time-series prediction,forming a hybrid risk prediction model that combines static feature recognition with dynamic sequence fitting.The results show that the proposed model outperforms single algorithms in both prediction accuracy and stability,effectively identifying high-risk periods and major driving factors.On this basis,a cost-risk bi-objective optimization model is established to simulate resource allocation and risk response strategies under different scenarios,providing a quantitative foundation for supply chain stability and decision-making optimization in the plastic products industry.
作者
杨涛
李雅
Yang Tao;Li Ya(Shaanxi Polytechnic University,Xian'yang,Shaanxi 712000,China)
出处
《塑料助剂》
2025年第6期88-93,共6页
Plastics Additives
基金
陕西工业职业技术大学生鲜电商与冷链物流创新团队(项目编号:KCTD2022-02)。
关键词
塑料制品
供应链风险
大数据
机器学习
优化模型
plastic products
supply chain risk
big data
machine learning
optimization model
