近红外光谱技术在常用包装材料分类方法的研究与应用

Research and Application of Near-infrared Spectroscopy Technology in Classification Methods of Commonly Used Packaging Materials

中文关键词:  近红外光谱技术  包装材料  分类

英文关键词:near infrared spectroscopy technology  packaging materials  classification

中文摘要:

      目的 探索近红外光谱技术在常见包装材料分类中的应用潜力。方法 首先利用近红外光谱仪对PBS、HDPE、LDPE、LLDPE、PP、PS、POE、PPC、PBAT、PLA、PGA+PBAT这11类常用包装材料进行光谱数据采集，然后采用滑动平均(MA)、标准正态变量变换(SNV)、多元散射校正(MSC)对光谱进行预处理，使用支持向量机(SVM)、K-近邻(KNN)、决策树和随机森林4种模式识别方法建立定性判别模型，最后根据模型评价指标和混淆矩阵对模型的预测性能进行比较。结果 SNV结合K-近邻的模型分类效果较优，准确率达到97.03%。结论 基于近红外光谱仪的塑料鉴别研究为塑料回收和再利用提供了一种便捷、快速、无损的检测方法，具有较好的应用前景。

英文摘要:

      The work aims to explore the potential of using near-infrared spectroscopy to classify common packaging materials. First, a NIR spectrometer was used to collect spectral data from 11 commonly used packaging materials, including PBS, HDPE, LDPE, LLDPE, PP, PS, POE, PPC, PBAT, PLA, and a blend of PGA+PBAT. The spectral data was then preprocessed using three techniques:Moving Average (MA), Standard Normal Variate (SNV) transformation, and Multiplicative Scatter Correction (MSC). Four pattern recognition methods, Support Vector Machine (SVM), K-Nearest Neighbor (KNN), Decision Tree, and Random Forest were utilized to develop qualitative discrimination models. Finally, the prediction performance of the models was compared based on evaluation metrics and confusion matrices. The combination of SNV preprocessing with the KNN algorithm yielded the best classification performance, achieving an accuracy of 97.03%. It is concluded that NIR spectroscopy provides a convenient, fast, and non-destructive method for plastic identification, which is advantageous for plastic recycling and reuse. The results indicate good application prospects for this method in sustainable material management.