基于K近邻算法的区域地基处理方案选择研究

Study on Selection of Regional Ground Improvement Plan on K-Nearest Neighbors

环杭州湾地区存在厚覆盖层地质分布,公路工程建设中需要进行大量的地基处理以减少路基病害。基于前期的设计资料与方案,利用K近邻算法进行分类。通过训练,测试集得到的地基处理方案结果较准确。在此基础,将路段分为桥头特殊路段和一般路段再进行计算,准确率接近100%,优化效果更佳。该方法可以在大量得到后期评价筛选的设计资料基础下,建立合理的数据库,在一个地质条件类似的地区内地基处理设计中利用并丰富,形成智能系统。

Around Hangzhou bay region,thick covering layer is widely districted.In construction of highways,it is necessary to conduct ground improvement to reduce disease of embankment.Based on previous design files and plans,K-Nearest Neighbors method was applied for classification.After data training,the results of improvement plans were correctly agreed with the test data.When the same method was used in separate bridge sections and normal section relatively,accuracy approaches 100%,which illustrated a successful optimization.The method can be applied with a great quantity of previous files and data of designs which has been verified valid to construct a database.Eventually the intelligence system will be formed which can be utilized and extended in design of ground improvement in a region with similar geology condition.