粮虫的识别属于参数多、混合度大的分类问题,特征客观权重的自动确定是粮虫可拓分类中的重要环节。提出了用聚类数评价函数自适应确定聚类数目,用FCM进行聚类分析,并依据最大隶属度原则对实值特征进行离散化处理。引入粗集理论中属性重要度的概念,自动确定粮虫特征的客观权重,并对粮仓中危害严重的9类粮虫进行了可拓分类,识别率达到93%,证实了基于FCM离散化的粗集权重在粮虫可拓分类中应用的可行性。 The identification of food insects belongs to the classification problem with many parameters and large mixing degree. The automatic determination of objective weight of features is an important link in the classification of food insects. The clustering number is adaptively determined by using the clustering number evaluation function. Clustering analysis is performed by FCM, and the real value features are discretized according to the principle of maximum membership degree. By introducing the concept of attribute importance in rough set theory, the objective weights of the characteristics of the grain insects were automatically determined, and the nine serious types of grain insects in the grain silos were classified by extension with a recognition rate of 93%. This confirms that based on FCM discretization Feasibility of Applying Rough Set Weight to Extension of Grain Insects.