论文部分内容阅读
采用近红外光谱技术结合化学计量学方法对菜籽油中多效唑残留进行定性检测。在4000~10000 cm-1光谱范围内采集126个菜籽油样本的近红外透射光谱。对原始光谱进行初步分析后,分别采用线性判别分析(LDA)、簇类独立软模式法(SIMCA)和最小二乘支持向量机(LSSVM)三种不同方法建立菜籽油中多效唑残留的定性检测模型,并对不同多效唑残留的菜籽油样本的分类正确率进行分析。研究结果表明,LDA,SIMCA及LSSVM 3种方法建立的检测模型均具有较高的判别能力,其校正集和预测集的正确率分别为93.33%,91.11%,95.56%和86.11%,88.89%,83.33%。此外,高多效唑残留样本的分类正确率大致趋于100%,而低多效唑残留样本的分类正确率则有一定波动。由此可知,利用近红外光谱技术可对菜籽油中多效唑残留进行快速、无损的定性检测。
Near infrared spectroscopy and chemometric methods were used to qualitatively detect the residues of paclobutrazol in rapeseed oil. Near-infrared transmission spectra of 126 rapeseed oil samples were collected in the spectral range 4000 ~ 10000 cm-1. After preliminary analysis of the original spectra, the qualitative detection of paclobutrazol (PP333) in rapeseed oil was established by three different methods: linear discriminant analysis (LDA), cluster independent soft mode (SIMCA) and least squares support vector machine (LSSVM) Model, and analyzed the classification accuracy of rapeseed oil samples with different MMZ residues. The results show that the detection models established by the three methods of LDA, SIMCA and LSSVM all have higher discriminant ability, and the correct rates of the corrected set and the predicted set are 93.33%, 91.11%, 95.56% and 86.11%, 88.89% respectively, 83.33%. In addition, the classification accuracy of the samples of the paclobutrazol generally approaches 100%, but the classification accuracy of the samples of the paclobutrazol is somewhat fluctuating. It can be seen, the use of near infrared spectroscopy of rapeseed oil in the paclobutrazol fast, nondestructive qualitative detection.