地质与勘探2025，Vol.61Issue(3)：536-544,9.DOI:10.12134/j.dzykt.2025.03.010
支持向量机驱动下的智能化工程地质分区探索
The Intelligent Engineering Geological Zoning Driven by Support Vector Machine(SVM)
摘要
Abstract
To address the challenges of traditional engineering geological zoning methods,such as difficulties in data acquisition,complex classification processes,and inefficiency,this study proposed an intelligent zoning approach based on Support Vector Machine(SVM)technology,and validated its accuracy through a case study of the Jijiawa gold deposit in Luoning County,Henan Province.This deposit exhibits complex engineering geological conditions,with dominant lithologies including biotite-plagioclase gneiss and hornblende-plagioclase gneiss,locally intruded by diabase dikes with poor engineering geological properties.The area was traversed by north-south trending faults,with unevenly distributed and well-developed joint fractures.Within a supervised learning framework,this study first constructed an engineering geological feature dataset.Based on known zoning data from exposed areas above the 700 m section,an SVM model was trained and subsequently applied to predict zoning in unexploited regions below the 700 m section.The research demonstrates that SVM,employing a Gaussian kernel function,effectively maps nonlinear geological features to a high-dimensional space,achieving linear separability of complex geological data.Under optimal parameters,the model achieved classification accuracies of 99.72%and 99.82%on the training and test sets,respectively,confirming its reliability and accuracy in complex geological conditions.The innovations of this study include:(1)establishing an SVM-based intelligent engineering geological zoning method;(2)validating its applicability under data-limited conditions;and(3)providing a scientific foundation for mining operations in complex geological environments.The results not only offer direct guidance for safe production at the Jijiawa gold deposit but also present a transferable technical solution for intelligent geological assessment in similar mining areas.关键词
支持向量机(SVM)/智能化/工程地质分区/监督学习/吉家洼金矿/河南省Key words
support vector machine(SVM)/intelligence/engineering geological zoning/supervised learning/Jijiawa gold mine/Henan Province分类
天文与地球科学引用本文复制引用
赵福权,孙斌堂,王秀丽..支持向量机驱动下的智能化工程地质分区探索[J].地质与勘探,2025,61(3):536-544,9.基金项目
中国冶金地质总局科研项目(编号:CMGBKY202208)资助. （编号:CMGBKY202208）
