基于遗传投影寻踪模型的膨胀土胀缩等级分类

由于膨胀土的胀缩特性受多种因素的综合影响,膨胀土胀缩等级分类是膨胀土处理中的一个难题。选取影响膨胀土胀缩等级分类的5个主要因素,通过在各评价等级范围内按均匀分布随机构建标准样本,结合遗传算法和投影寻踪方法,获得最佳投影方向和样本最佳投影值,利用逻辑斯谛曲线函数拟合最佳投影值与经验等级之间的非线性关系,建立了膨胀土胀缩等级分类的遗传投影寻踪模型。最后,将该模型应用到某实际工程中,分类结果与实际情况相吻合,验证了该模型的可行性和适用性,该模型可以作为一种膨胀土胀缩等级分类的新方法。

Classification of swelling-shrinkage grade of expansive soils based on genetic projection pursuit model

Due to the integrated effects of multiple factors, the classification of the swelling-shrinkage grade of expansive soils is one of the most difficult problems in expansive soil treatment. Five key factors were chosen to construct the model. By constructing standard samples through stochastic interpolation, we obtained the best projection direction and best projection values using the genetic algorithm and the projection pursuit method. Then we fitted a non-linear relationship between the best projection values and empirical values using the logistic curve function, and developed a new model for the classification of swelling-shrinkage grade of expansive soils, i.e., the genetic projection pursuit model. Finally, the new model was applied to actual engineering. The results tallied well with the actual situation, and thus verified the feasibility and applicability of the proposed model. This model can be a new method for classifying the swelling-shrinkage grade of expansive soils.