通过对标准化数据矩阵加权, 以及对特征向量取绝对值, 对传统主成分分析法进行了改进, 并构建了基于改进主成分分析法的工程材料综合评价模型; 并以5种候选低温储罐用铝合金材料的综合评价为例, 对上述模型的适用性进行了研究。结果表明: 用该模型得出2014-T6铝合金是最佳的候选材料, 这与实际应用以及TOPSIS法的评价结果一致; 改进主成分分析法通过对特征向量取绝对值, 避免了评价结果出现负值, 使得评价结果更为合理, 适用于工程材料的综合评价。

Traditional principal components analysis (PCA) was improved through adding weight to standardizing data matrix and using absolute value of eigenvector, and the comprehensive evaluation model of engineering materials was established on the base of improving traditional PCA. Taking the comprehensive evaluation of low temperature storage pot materials, 5 kinds of aluminum alloys, as an example, the applicability of the model mentioned above was studied. The results show that aluminum alloy 2014-T6 was the best candidate material, which accorded with industry practice and the result obtained by TOPSIS (a technique for order preference by similarity to solution). Through taking the absolute value of the feature vector, improving PCA avoided the negative evaluation results and made them more reasonable. All above showed that improving PCA was feasible in the comprehensive evaluation of engineering materials.