基于消耗波动性聚类的航材分类研究

针对难以依据航材消耗数据进行分类的问题,建立基于消耗波动性聚类分析的分类模型。基于消耗序列波动性将航材转化为二维图,因该机型服役时间较短,航材样本数量较小,选用无监督分类算法聚类分析对航材分类。针对传统聚类算法的局限性,提出层次划分聚类算法,并使用“容距比”参数为初始中心选择良好环境。仿真结果显示层次划分聚类算法更加稳定高效,同时表明该模型能有效进行航材分类。

Aviation material classification research based on consumption volatility clustering analysis

To solve the problem that it is difficult to classify aviation material consumption data, a classification model based on consumption fluctuation clustering analysis is established. Based on the fluctuation of the consumption sequence, the aviation materials are converted into two-dimensional graphs. Due to its short service time and small quantity of aviation material samples, the unsupervised classification algorithm is used to classify the aviation material. Aiming at the limitation of traditional clustering algorithms, a hierarchical partition clustering algorithm is proposed, and the “tolerance to distance ratio” parameter is used to select a good environment for the initial center. The simulation results show that the hierarchical partition clustering algorithm is more stable and efficient.