基于近红外网络的小麦品质监测

Wheat Quality Monitoring by NIR Network

【目的】利用自行建立的中国谷物主产区的近红外分析网络,研究小麦的品质分布监测、品种鉴别及种植省份的区域划分。【方法】应用FOSS公司的Infratec 1241型近红外分析仪构建谷物品质近红外分析网络,该网络主要由网络主机、参比实验室、网络管理中心和分布于中国粮食主产省区的网络子机构成。利用该近红外分析网络并结合GPS定位采样技术和GIS技术,对2009年的冬小麦品质分布进行监测;采用软独立建模分类法(SIMCA)对获取的近红外光谱进行分析,建立小麦品种、所在省份的识别模型。【结果】通过网络中心建立和管理模型,可以有效节约成本并具有统一的准确度和精度,该网络的一致性较好,子机与主机的相关系数高于0.92。基于该网络可构建不同尺度区域下的小麦品质分布图,从而获得不同地区小麦品质的分布信息。根据近红外网络提供的光谱数据,可进一步对小麦的品种和所在省份进行区分,对山东省5个小麦品种的识别正确率>80%,对北京、山东、江苏三省所种植的小麦的区分准确率>90%。【结论】近红外网络在小麦的品质区划分布、品种鉴别及种植省份识别方面具有较大应用潜力,可提供大范围的小麦品质数据支撑,从而为优质优价收购和政府决策提供技术支撑。

【Objective】 Based on the NIR network that constructed in main grain producing areas in China,the quality distribution monitoring of wheat,the variety discrimination,and producing area classification were studied.【Method】The Infratec 1241 NIR analyzer from FOSS Company was used to construct the grain NIR network.This network consisted of the main instrument,reference library,network administration center and several satellite instruments that distributed in the main grain producing areas of China.Combined with GPS and GIS technology,the network was used to analyze the quality distribution of winter wheat in 2009.The collected NIR spectra were analyzed by Soft Independent Modeling of Class Analogy（SIMCA）,and the classification models for variety,and producing provinces of wheat were constructed.【Result】The results showed that it could save cost and keep unitive accuracy and precise by using network center to construct and manage the calibration models.All the satellite instruments had good constancy with the main instrument,and their correlation coefficients were over 0.92.Based on the NIR network,the wheat quality distribution map in different scales could be obtained,thus obtained the quality distribution information of wheat.With the spectra collected from NIR network,the wheat variety and producing area could be discriminated.The classification accuracy for five varieties of wheat in Shandong province was over 80%,and for discriminating wheat that produced in Beijing,Shandong and Jiangsu,the classification accuracy was over 90%.【Conclusion】 The results showed that NIR network had a potential in analyzing the quality distribution,variety and producing areas discrimination for wheat.It could provide large scale data support and thus be used to guide the wheat purchase,quality management and decision-making.