Geotechnical Parameters Prediction Based on Multiple Regression Model and Information Entropy

Combining correlation analysis and establishing empirical relationship is a common method for geotechnical parameters prediction in geotechnical engineering. This method can quickly and efficiently predict geotechnical parameters through some easily obtained geotechnical testing data. Taking the geotechnical parameters of the cohesive soil in the lower part of Gehu Formation in Changzhou area as the research object, 102 groups of geotechnical testing data were selected, and the geotechnical parameters prediction model based on the multiple regression model and information entropy was established. Firstly, the cohesion and internal friction angle of quick direct shear test were determined as the prediction objects, and the influencing factors were density, water content, void ratio and liquidity index. Then, through the regression analysis of SPSS software, the optimal single factor fitting model was determined, and calculated the weight of the influencing factors by using the method of information entropy to obtain the improved multiple regression forecast model. Finally, the standard error analysis method was used to analyze the prediction results. The results show that: The density has a positive correlation with the cohesion and the internal friction angle, respectively, and the other factors have a negative correlation with the cohesion and the internal friction angle, respectively. The mean error of the predicted value of cohesion is -1.34 kPa, and the normalized root mean squared error is 12.00%. The mean error of the predicted value of internal friction angle is -0.05°, and the normalized root mean squared error is 18.30%. The study shows that the prediction model has high reliability and can effectively predict the cohesion and internal friction angle of cohesive soil in the lower part of Gehu Formation in this area. This method provides a way of thinking for the prediction of geotechnical parameters.