摘要
为解决烟草供应链中信息不透明、数据零散、追溯困难等问题。采用近红外光谱(near-Infrared Spectroscopy,NIR)与区块链技术,结合随机森林(Random Forest,RF)定量分析模型与支持向量机(Support Vector Machine,SVM)定性识别模型,设计了全链路闭环架构,并开展了关键技术整合与应用验证。结果表明:系统检测误差率<3%,数据上链耗时≤2 s,全流程追溯响应时间<5 s,溯源准确率达99.8%;试点应用中,检测效率提升40%,等级误判率从8%降至2%,单样本检测成本降低98%,烟叶优质率提升25%。因此,通过近红外光谱与区块链的深度融合,能有效提升质量数据的可靠性与追溯效率。
To address issues such as information opacity,fragmented data,and difficulty in traceability within the tobacco supply chain,near-infrared spectroscopy(NIR)and blockchain technology were employed,combined with a random forest(RF)quantitative analysis model and a support vector machine(SVM)qualitative identification model.A full-chain closed-loop architecture was designed,and key technology integration and application validation were carried out.The results showed that:the system detection error rate was less than 3%,the time to record data on the blockchain was≤2 s,the full-process traceability response time was less than 5 s,and the traceability accuracy reached 99.8%;in pilot applications,detection efficiency increased by 40%,grade misjudgment rate decreased from 8%to 2%,the cost per sample detection dropped by 98%,and the proportion of high-quality tobacco leaves increased by 25%.Therefore,the deep integration of near-infrared spectroscopy and blockchain can effectively improve the reliability of quality data and the efficiency of traceability.
作者
于闽
YU Min(Nanping Branch of Fujian Tobacco Company,Nanping 353000,China)
出处
《国外电子测量技术》
2025年第11期225-231,共7页
Foreign Electronic Measurement Technology
关键词
近红外光谱
区块链
烟叶质量
精准检测
near-infrared spectroscopy
blockchain
tobacco leaf quality
accurate detection
