摘要
同一土层中的岩土参数会随着位置改变而变化,在一定距离范围内存在一定程度的相关关系,该相关性随着距离的增大而衰减,直至不相关。地层划分是岩土工程勘察设计的重要内容,在地层划分中应考虑土性的空间相关性,且有必要提高地层划分的智能化程度。该研究发展一套能“自然”地考虑土性空间变异特征的智能地层划分方法。首先分析某场地岩土工程勘察数据的空间变异性特征,并据此生成一个随机场实例,构建“完备地层”模型;对构建的地层模型进行6个孔静力触探模拟,基于静力触探结果展示基于贝叶斯压缩感知的数据扩展和基于支持向量机、高斯模型和隐马尔科夫随机场的地层智能划分方法,将划分结果与“完备地层”模型对比,论证地层划分方法的准确性。
Geotechnical parameters in the same soil layer will change with the change of location.There is a certain degree of correlation within a certain distance.This correlation will attenuate with the increase of distance until it is irrelevant.Stratum division is an important part of geotechnical engineering survey and design.The spatial correlation of soil properties should be considered in stratum division,and it is necessary to improve the intelligence level of stratum division.This study developed a set of intelligent stratigraphic classification methods that can"naturally"consider the spatial variation characteristics of soil properties.First,the spatial variability characteristics of geotechnical engineering survey data of a certain site are analyzed,and a random field example is generated based on this to build a"complete stratum"model;Six hole static penetration simulations were carried out on the constructed stratum model.Based on the static penetration results,data expansion based on Bayesian compressive sensing and intelligent stratum classification based on support vector machines,Gaussian models and hidden Markov random fields were demonstrated.The classification results were compared with the"complete stratum"model to demonstrate the accuracy of the stratum classification method.
出处
《科技创新与应用》
2024年第36期139-144,共6页
Technology Innovation and Application
关键词
土性变异性
地层划分
相关距离
随机场理论
静力触探
soil variability
stratum division
correlation distance
random field theory
static sounding