塔里木大学学报

和田224团骏枣与阿克苏10团骏枣在价格上有明显的差别,快速准确的识别二者有很大的实际意义。应用距离匹配法对和田224团骏枣与阿克苏10团骏枣进行聚类分析,预处理方法采用归一化,基线校正和Savitzky-Golay卷积求导法(13点平滑,3点差分宽度),并利用马氏距离剔除异常样本。研究结果表明:和田224团骏枣与阿克苏10团骏枣在水分含量、总糖含量和总酸含量上有明显区别。预测效果最好的是总酸波段的聚类分析模型,其次是水分波段的聚类分析模型,最后是总糖波段的聚类分析模型。在分类效果上,聚类分析模型更适用阿克苏10团骏枣,所以阿克苏10团骏枣比和田224团骏枣效果好。结论:利用距离匹配法对和田骏枣与阿克苏骏枣进行产地的鉴别是可行的,该法对考察南疆红枣的产地鉴别有一定的参考价值。

There is great price difference between the No. 224 Group jun-jujube in Hetian and No. 10 Group in Aksu,and it is practical significance to identify quickly and accurately. The Jun-jujube in No. 224 Group Hetian and No. 10 Group Aksu were clustering analyzed by the distance matching method,and were pretreated with normalized method,baseline correction,Savitzky-Golay convolution derivation method( 13 points smooth,3 points differential width),and abnormal samples were excluded using the Mahalanobis distance. The results showed there were great differences in the moisture content,total sugar content and total acid content between the No. 224 Group Hetian and No. 10 Group Aksu Jun-jujube. The most precise method was the clustering analysis model of total acid band,followed by the cluster analysis model of water wave and the clustering analysis model of total sugar band. The cluster analysis model is more suitable for the Jun-jujube in 10 Group Aksu,because of the more efficiency classification. Conclusion: It is feasible to identify the locality of Hetian and Aksu Jun-jujube using the distance matching method,and this can provide some reference value in identification the producing area of Southern Xinjiang jujube.