The LDA-KNN model to classify quality of rock mass of a dam abutment
The classification and evaluation of the engineering quality of rock mass has an important impact on the safety,design,and economic benefits of construction projects.In this study,we responded to the problems posed by uncertainty,interference by human factors,and the neglect of traditional qualitative grading in prevalent models used to assess the quality of the rock mass.We collected samples of drift rock at the abutment of a hydropower dam to establish a database of rock mass,selected the spacing between structural planes(D).coefficient of integrity of the rock(Kv),and rock quality index(RQD)as indices of evaluation based on an analysis of actual projects,and introduced the linear discriminant analysis-based method of dimension reduction and the K-Nearest-Neighbor multi-level classification model to develop a nonlinear model to classify rock mass.A combination of qualitative and quantitative analyses was used to comprehensively classify the rock mass based on multiple factors and indicators,and to solve the problems of the redundancy of information and the complexity of judgment based on multiple indicators.The proposed model generated more accurate results than previously developed schemes that were in line with engineering practice,reduced the influence of human factors,and exhibited a strong predictive capability.
rock mass structureclassification of rock mass qualitylinear discriminant analysisK-Nearest-Neighbor analysisclassification model
万方数据
维普
