利用在线近红外光谱鉴别雌雄蚕蛹的方法

A Study on Method of Online Discrimination of Male and Female Silkworm Pupae by Near Infrared Spectroscopy

采用自动识别雌雄蚕蛹的技术与设备,是解决蚕种生产人工鉴蛹人力短缺和保证鉴蛹正确率的有效途径之一。以6个家蚕品种的蚕蛹供试,利用自制的在线近红外光谱检测装置进行雌雄蛹的鉴别。分别以3种采谱条件采集各供试家蚕品种蚕蛹样品的近红外光谱,以采用全谱和经竞争性自适应重加权算法（CARS）、无信息变量消除法（UVE）优选后的波长变量分别建立偏最小二乘判别分析（PLSDA）的雌雄蚕蛹识别模型。测试结果显示,在3种采谱条件下,以单个家蚕品种的光谱数据建模时雌雄蚕蛹鉴别的正确率达100%;多个家蚕品种的蚕蛹混合后采集光谱数据建模,其中采用方法 a（积分时间8 ms,扫描次数5次）、方法 b（积分时间为8 ms,扫描次数10次）、方法 c（积分时间20 ms,扫描次数10次）获取光谱数据建模的鉴别正确率分别达94.2%、95.2%、100.0%。选择高速采集光谱的方法 a,经CARS法优选波长后建模的测试结果优于经UVE法优选波长建模的结果,对校正集、交叉验证集和预测集蚕蛹样品的雌雄识别正确率分别为100%、100%和96.2%。研制的在线近红外光谱检测装置及建立的检测方法与模型,可应用于蚕种生产中快速鉴别分选雌雄蚕蛹。

Implementation of novel technology and equipment for automatic discrimination of male and female pupae is one of the effective ways to solve the problem of labor shortage and to ensure the accuracy of pupa sex identification in silkworm（ Bombyx mori） egg production. Male and female pupae from 6 silkworm varieties were discriminated with a selfmade online near infrared（ NIR） spectroscopic detection device. The NIR spectra of silkworm pupa samples were collected under three different conditions. Pupa sex discrimination models were established by partial least squares discriminant analysis（ PLSDA） with the whole variables and selected variables competitive adaptive reweighted sampling（C ARS） and uninformative variable elimination（UVE）,respectively. The results showed that the accuracy of pupa sex discrimination was all 100% under 3 spectrum collection conditions when the spectral data from pupae of single silkworm variety were applied in modeling. The spectral data modeling from pupae of mixed silkworm varieties showed that the discrimination accuracy of method a（ integration time 8 ms,scanning 5 times）,b（ integration time 8 ms,scanning 10times） and c（ integration time 20 ms,scanning 10 times）was 94. 2%,95. 2% and 100%,respectively. For method a（ high speed spectrum collection）,the test result of model from optimization by CARS was better than that by UVE. The accuracy for calibration set,cross verification set and test set of pupa samples was 100%,100% and96. 2%,respectively. The developed online NIR spectro-scopic discrimination device,detection method and models can be further used in quick discrimination and sorting of male and female silkworm pupae in silkworm egg production.