关键词: 粉煤灰, 混凝土, 投影寻踪回归, 蚁群算法, 聚类分析

Abstract: For the problems in classification of fly ash at present, the chemical constituents and physical properties influencing the quality of fly ash was analyzed，and degree of finess，vitric，loss on ignition, K2O, SO3 and CaO are chosen as the criterion indices for classification of fly ash, the pozzolanic activity of fly ash are consider，a classification method of projection pursuit was established，optimized its project direction used Ant Colony Algorithm，and the model is applied to classifying in an example. The evaluation results are accurate，this method is simple，and some errors were eliminated because of the subjective factors for traditional methods. Projection Pursuit is a new evaluation method of judging and classifying of the quality of fly ash，it has provided scientific basis to comprehensive utilization of fly ash.

Key words: fly ash, concrete, projection pursuit, Ant Colony Algorithm, cluster analysis