论文部分内容阅读
为了提高复杂产品高维不平衡质量特性数据集关键质量特征识别效率,提出CEM-IG识别方法.通过调整CEM(classification EM algorithm)算法的K值输出不同的聚类结果,消除冗余样本后作为IG(information gain)算法的输入,并以IG作为判别质量特性重要程度的标准构建识别模型,最终输出最优关键质量特性集.算例结果表明,该方法将CEM的缺失值处理能力和IG的不相关特性筛选能力优势互补,能够有效降低不平衡和高维度带来的负面影响,正确识别产品关键质量特性.
In order to improve the efficiency of identifying key quality features in high-dimensional unbalanced quality data sets of complex products, a CEM-IG identification method is proposed, and different clustering results are output by adjusting the K value of the CEM (classification EM algorithm) algorithm. After eliminating the redundant samples, IG (information gain) algorithm, IG is used as a criterion to judge the importance of quality features, and finally the optimal key quality feature set is output.Experimental results show that the proposed method can reduce the processing capacity of missing value of CEM and IG Unrelated features Screening capabilities complement each other, effectively reducing the negative impact of imbalances and high dimensions, and correctly identifying key product quality characteristics.