基于投影寻踪模型的纸和纸板质量综合评价

Comprehensive Quality Evaluation of Paper and Cardboard Based on Projection Pursuit Model

采用基于遗传算法的投影寻踪模型(GA-PP)用于纸和纸板质量分类与评价。该模型依据样本数据类内密集度和类间散开度最大构建目标函数,寻求最优的投影方向,将多维分类指标综合成一维指标,然后利用一维投影值特征信息,根据相近原则进行样本的聚类与排序。实例研究表明:投影寻踪模型得到的各指标权重客观地反映了各评价指标之间的影响关系,用于纸和纸板质量分类与评价切实可行,分析结果和其它方法具有较好的一致性,并更为合理和可靠。

The model based on genetic algorithm and projection pursuit （GA-PP） was applied to study the classification and comprehensive quality evaluation of different paper and cardboard. Determined by the characteristic of density and variance of the sample datum, GA-PP could project high dimensional data to low dimensional space through an optimized projecting direction, the values of the evaluation index were synthesized into a one -dimension projection value. Through studying the main characteristics in 1D space,the key information could be used to classify and evaluate the samples according to the principle of similarity. The results of case study showed that the obtained weighting coefficients by GAPP related the interrelationship of the indexes and the model was feasible in practice in quality evaluation of paper and cardboard. The analysis results were rightly consistent with the other methods as well as more reasonable and reliable.