基于投影寻踪的灰色聚类分析在 煤样识别中的应用*

煤炭种类识别是一个非线性灰色系统，鉴于煤样特征指标较多、其意义、量纲不同，且在数量上差异较悬殊，选用了灰色定权聚类评判模型。因指标权重对模型评判结果具有重要影响，采用了投影寻踪法计算出各指标的客观权重，以此求出煤样隶属各灰类即无烟煤、烟煤及褐煤的聚类系数。分析表明用基于投影寻踪算法的灰色定权聚类分析识别煤样种类完全可行。

Application of grey cluster analysis based on projection pursuit in coal sample recognition

Coal type classification is a non-linear grey system. In view of the large number of characteristic indexes, different meanings and dimensions of coal samples, and the great difference in quantity, the grey fixed weight clustering evaluation model is selected. Because the weight of the index has an important influence on the model evaluation results , the projection pursuit method is used to calculate the objective weight of each index, and then the clustering coefficients of anthracite coal, bituminous coal and lignite coal are obtained. The empirical analysis shows that it is feasible to identify the coal sample by using the gray-weighted clustering analysis based on the projection pursuit algorithm.