化学样本数据常为非平衡,用传统方法分析这些数据集时,对于需特别关注的少数类数据,识别能力往往较差。因此,提出建立基于粒计算的分类规则模型(GCCRM),先用改进的自适应共振网络ETM-ART2将性质相近的个体聚合为信息粒,降低样本容量和问题规模,同时又保持较高的纯度;然后将信息粒的属性特征模糊离散化,简化它;最后经关联规则挖掘,得可预测样本的分类规则模型。用于识别玻璃,结果GCCRM能剔除冗余信息,加强关键特征,所提取的分类规则预测正确率高,尤适用于非平衡数据集,其性能比C4.5决策树、支持向量机SVM等算法优良。 Chemical sample data is often unbalanced. When analyzing these datasets by traditional methods, the recognition ability is often poor for a few types of data that require special attention. Therefore, it is proposed to establish classification rules model based on kernel computing (GCCRM). First, the modified adaptive resonance network ETM-ART2 is used to aggregate individuals of similar nature into information particles, reducing sample size and scale while maintaining high purity Then, the attribute features of the information grain are blurred and simplified, and finally the rules of the classification rules of the sample are obtained by mining the association rules. Which is used to identify glass. As a result, GCCRM can eliminate redundant information and strengthen key features. The extracted classification rules have high prediction accuracy and are especially suitable for non-equilibrium data sets. Its performance is better than that of C4.5 decision tree and support vector machine SVM excellent.