基于近红外光谱学的机采籽棉杂质含量快速检测方法研究

Abstract: The impurity content is an important factor affecting the price of cotton and the national standard GB1103 stipulates that it should be settled according the weight of seed cotton deducting the weight of impurity and water in the purchase of seed cotton. However, it is a pity that there are no rapid detection methods of impurity content in seed cotton. The near infrared spectroscopy is known for it is rapid, non-invasive and simultaneous detection of multiple detection indexes. Therefore, our team plans to study the rapid detection method of impurity content in machine-harvested seed cotton based on near-infrared spectroscopy. We will mainly focus on: The first, rapid and effective spectral acquisition method of seed cotton and its impurities. The second, rapid feature extraction method of the near infrared spectra. The last, the key algorithms of the standard near infrared spectral database of seed cotton and its impurities. At last, the reliable and stable prediction models should be built to classify and quantify the impurities in machine-harvested seed cotton. This research can provide a new impurity detection method of machine-harvested seed cotton and lay the theoretical basis for the development of new equipment for the impurity detection in the machine-harvested seed cotton. From the social perspective, this work plays the positive role protecting benefits of workers, peasants and businessmen, especially in protecting the benefit of peasants and the implementing of preferential policies for farmers taken by the government.

杂质含量是籽棉收购定价的重要依据，国标GB1103规定籽棉收购时按公定重量结算，实际收购时通常按籽棉重量扣补水和杂质的重量后结算。目前，棉花含水率检测方法已较为成熟，但尚缺乏籽棉含杂率的快速检测方法。近红外光谱分析技术具有快速、无损和多指标同时检测等优点。课题组拟研究采用近红外光谱分析技术检测机采籽棉杂质类别和含量的方法，着重研究：一、机采籽棉及其杂质的有效近红外光谱采集方法；二、机采籽棉及其杂质的近红外光谱特征提取方法；三、标准机采籽棉及其杂质近红外光谱数据库的关键算法。最终建立可靠、稳定的机采籽棉杂质含量和类别的快速检测模型。该项目研究将为机采籽棉杂质检测提供新方法，为研发新型棉花杂质含量检测装备奠定理论基础，另外对维护棉花贸易过程中工、农、商各方利益，尤其是维护农民利益，落实国家的惠农政策起到积极的作用。

针对缺乏机采籽棉植物性杂质含量快速检测技术的问题，提出基于机采籽棉对近红外光谱的吸收特性，建立机采籽棉植物性杂质的近红外光谱检测模型的检测思路。基于试验探索和参照文献资料，规范了籽棉样品近红外光谱的采集方法；对光谱进行多元散射矫正、平滑、基线移除、归一化等预处理，降低光谱噪声提高光谱质量；使用样品分布空间、杠杆值和学生残差等信息识别和剔除异常样品；由于籽棉样品密度与体积性状难以控制，导致光谱中存在显著的基线平移和旋转变化，使用一阶微分、二阶微分消除基线平移和旋转变化，改善建模效果；设计和开发了棉花近红外光谱数据库原型系统，应用基于杰卡德相似性原理的全光谱匹配算法探索基于数据库查询分析模式的籽棉杂质含量快速检测方法；拓展性的研究了棉花品质在典型锯齿加工过程中的变化规律，构造了棉花综合品质指标。

内容获取失败，请点击重试