论文部分内容阅读
本文用近红外光谱结合最小二乘双胞胎支持向量机(LSTSVM)算法建立了烟叶等级分类模型。从三个等级共210个烟叶样品中,取出120个样品作为建模集,剩余90个样品作为预测集。为了建立最优模型,对光谱预处理方法和模型参数进行筛选优化,最优模型对预测集样品的平均识别率为95.56%,结果表明该方法可以作为烟叶等级分类的一种有效方法。此外,将该算法与SIMCA、PLS-DA、SVM等三种常见的模式识别算法进行了比较,结果表明基于样品的原始光谱,同等条件下,LSTSVM算法的预测效果优于其他三种算法。
In this paper, tobacco leaf classification model was established by using near-infrared spectroscopy combined with least squares twins support vector machine (LSTSVM) algorithm. A total of 210 tobacco leaf samples from three ranks were taken as the modeling set and the remaining 90 samples as the predicted set. In order to establish the optimal model, the spectral pretreatment method and the model parameters were optimized. The average recognition rate of the optimal model to the prediction set samples was 95.56%. The results showed that this method could be used as an effective method to classify tobacco leaves. In addition, compared with the three common pattern recognition algorithms, such as SIMCA, PLS-DA and SVM, the results show that the LSTSVM algorithm outperforms the other three algorithms under the same conditions based on the original spectra of the samples.