基于Fisher判别法岩溶塌陷倾向性等级分类预测

Classification Prediction of Karst Collapse Tendency Level Based on Fisher Discriminant Analysis Method

为准确预测岩溶塌陷倾向性的等级分类,通过分析大量观测实例,选取岩性系数、岩体结构系数、地下水系数、覆盖层系数、地形地貌系数和环境条件系数作为模型判别因素。对12个实际观测样本进行训练,建立了基于Fisher判别分析法(FDA)的岩溶塌陷倾向性等级分类预测模型。借助SPSS软件工具,得到判别模型的4个判别函数。根据判别函数对训练样本进行回判,并对2个待判样本进行预测。结果显示:第一、第二判别函数的综合判别效率达到100%,大于规定的85%,满足工程实际应用需求;对训练样本进行回判时,误判率为零,同时对待判样本的分类预测准确率为100%。

In order to predict the Karst collapse tendency level accurately,the rock quality coefficient,rock mass structure coefficient,ground water coefficient,soil stratum coefficient,landform physiognomy coefficient and environment condition coefficient were determined as the major discriminant factors through analyzing a lot of observation examples.By training twelve observation samples,a classification prediction model for Karst collapse tendency level was built based on FDA method.By SPSS software,four discriminant functions were established for the FDA model.According to the functions,the training samples were differentiated and two prediction samples were predicted.The results show that the comprehensive discriminant efficacy of the first and second discriminant function reaches 100% and meets the need of actual engineering well,and that the return discriminant method is used to differentiate the trained samples and its misjudgment rate is zero,also the prediction results of two prediction samples coincide with the practical results exactly.