用基于邓氏灰色关联度的聚类方法对煤种进行聚类的研究

为实现煤种的准确分类,采用试验研究和定量分析的方法,测量了25组试验煤样的发热量、灰分、挥发分和硫分4项参数;运用基于邓氏灰色关联度的聚类方法,并根据其原理编写VB程序,实现该方法的程序化,对煤样的4项参数进行分析。由计算结果可知,当权值为0.4、临界值为0.68时,获得了最好的聚类结果,据此可从关联性角度研究煤样特性的相似特征,并对具备共同特征的煤种进行聚类。结果表明:在任选的25个试验样本中,误差样本只有一个,准确度为96%。

Study on Coal Cluster by Cluster Method Based on Deng's Grey Relational Degree

In order to cluster the coal samples accurately, four parameters, the calorific value, ash, volatile and sulfur content of 25 coal samples were measured. The cluster method based on Deng’s grey relational degree was used to analyze the parameters, and the method was programed by vision basical language. From the calculation results, it was seen that when the weight was 0.4, the critical value was 0.68, the best clustering results was obtained. The association was paid close attention to, the similarity of coal property was studied and the coal samples with the same characteristics were clustered. The results indicated that there was only one error sample among 25 samples, the precision was 96%.