Abstract: Collapsibility is an important engineering characteristic of loess, which is indicated by the collapsibility coefficients, dead weight collapsibility coefficients and the initial collapse pressure under the condition of one-dimensional compression stress. There are many influencing factors of collapsible deformation characteristics such as grain size, density and moisture of soil. There is a certain correlation between the influencing factors which are not completely independent. Based on the physical indices statistical analysis and correlation analysis on the collapsible loess fields of loess high platform of Xi'an metro line No.4 and loess tableland of high-speed railway tunnel from Baoji to Lanzhou, the ratio of water content to the liquid limit, and the void ratio are determined to be the factors by the factor analysis method. The multiple linear regression analysis method is used to obtain the relationship between the dead weight collapsibility coefficients, the initial collapse pressure, the modulus of compressibility and the factors according to the test data of two collapsible loess fields. The deviation tests of calculated value and the measured value of the dead weight collapsibility coefficients, the initial collapse pressure and the modulus of compressibility are carried out. The rationality and the accuracy of determining the factors and establishing the relationship between the collapsible parameters and the independent influence factors are verified by the factor analysis method. The valuable significance of the rapid and accurate evaluation of collapsibility and the collapsible deformation of loess foundation is proved by setting up the correlativity of the collapsible parameters of loess in two types of topographic feature of two loess regions.

Key words: collapsible loess, collapsible deformation, physical indices, factor analysis method, correlation analysis