基于近红外光谱技术的常见塑料快速分类方法研究  

Rapid Classification of Common Plastics Based on Near-Infrared Spectral Technology

作　　者：郭慧玲[1] 邓文怡[1] 李晓英[1] 

机构地区：[1]北京信息科技大学仪器科学与光电信息学院,北京100192

出　　处：《现代科学仪器》2012年第5期28-31,44,共5页Modern Scientific Instruments

基　　金：北京市属高等学校人才强教计划资助项目;PHR201107130

摘　　要：对废弃塑料进行回收再生利用具有重要的意义,首要的问题是对塑料进行鉴别分类。本文主要对塑料快速分类方法做了深入的研究。设计并搭建了采集数据的实验系统,编写了处理数据的处理软件。分别采用了全局相关分析、特征相关分析、欧氏距离和k_means聚类分析对数据进行了处理。对常见PE,PET,PP,PVC等塑料进行快速分类鉴别。其中k_means聚类分析法对89个未知样品的预测结果准确率为92.1%,说明近红外光谱结合k_means聚类法进行常见塑料种类鉴别在技术上是可行的。另外提出采用多模式共识可以进一步提高分类的准确率和可靠性。Recycling of waste plastics is of great significance and the primary problem is the identification and classification of plastics.An in-depth research on rapid classification of plastics has been made.The experimental system of data collection were designed and set up,the software of data processing was written,then data were processed by using global correlation analysis,characteristics related correlation analysis,Euclidean distance analysis and k_means cluster analysis.The common waste plastics such as PE,PET,PP,and PVC were identified and classified Rapidly.The accuracy rate is 92.1% on the predicted results to 89 unknown samples using k_means cluster analysis method.It is indicated that using the near-infrared spectral technology combined with k_means cluster analysis method to identify common plastic types are technically feasible.In addition,it is proposed that adopting consensus of multi-mode can further improve the classification accuracy and reliability.

关 键 词：近红外光谱 全局相关 特征相关 欧氏距离 k_means聚类 

分 类 号：TQ320.77[化学工程—合成树脂塑料工业]