Reliability analysis of foundation pit engineering using support vector machine

The design process in the foundation pit engineering exist many uncertain factors including soil parameters, soil thickness, load and groundwater depth and so on. So the calculated displacement, settlement and safety coefficient are random and uncertain. This paper puts forward a kind of reliability analysis method of foundation pit engineering based on support vector machine (SVM). Firstly, the distributions of the random parameters are determined. The uniform test is used for obtaining random parameter samples. The random parameter samples are calculated to get the corresponding response sample, such as the maximum horizontal displacement, maximum ground settlement and safety factors. Then, the SVM is adopted to establish the response surfaces between the random samples and corresponding response parameters. On the basis of the response surfaces, the random parameters generated by the Latin hypercube sampling method are predicted to obtain the probability density distribution of each response parameter. Finally, the system failure probability of foundation pit engineering is calculated. The case study shows that the method has high computing efficiency and the result is reasonable. It provides the basis for analyzing the reliability of foundation pit engineering.