组合评价法在烟叶化学质量综合评价中的应用研究

Application of combinatorial evaluation method in comprehensive evaluation of leaf tobacco chemical quality

为解决多种评价模型结论差异性问题,对烟叶质量做出全面、科学的评价,将主成分分析、模糊综合评价和夹角余弦法组合为一种评价模型。该模型首先通过Kendall-w系数法对各评价结果进行事前一致性检验,在保证各方法评价结果具有一致性检验的前提下,运用算术平均组合评价模型、Borda组合评价模型和Copeland组合评价模型3种方法对独立的评价结果进行组合评价,应用等级相关系数进行组合评价方法的事后检验以衡量组合评价与各评价方法的评价结果是否吻合,并根据等级相关系数的大小,选出最优的组合评价方法。

A comprehensive evaluation model based on principal component analysis （PCA）, fuzzy comprehensive evaluation, and angle cosine was proposed to evaluate chemical quality of tobacco leaves. Kendall-w correlation was applied to check coherence of evaluation results from different methods in advance. Mean, Borda, and Copeland combination evaluation models were used for comprehensive evaluation of these independent results. Spearman correlation was then used to test whether comprehensive evaluation results matched independent results, and the best method based on spearman correlation indexes was selected.