A search method for probabilistic critical slip surfaces with arbitrary shapes and its application in slope reliability analysis

The Monte Carlo simulation (MCS) is generally accepted as an accurate approximation method in slope reliability analysis. However, low efficiency hinders its use. Based on the online sequential extreme learning machine (OS-ELM), the OS-ELM-MCS method is proposed, which overcomes the requirement of calculating the safety factor, respectively, when using a conventional MCS for reliability analysis. Combined with the multi-initial points sequential quadratic programming (MSQP) algorithm, a search method for probabilistic critical slip surfaces with arbitrary shapes is introduced. It is possible to conduct both a reliability analysis and deterministic analysis under the same algorithm framework. Based on the proposed method, reliability and deterministic analyses of two cases are carried out. Compared with the results of other methods, the accuracy of the reliability and deterministic analysis method is verified. This verifies that the accuracy and time performance of the proposed method are better than those of conventional algorithms. The results of these case studies reveal the differences between the conclusions of the reliability and deterministic analyses. Therefore, it is necessary to combine reliability analysis and deterministic analysis to improve the credibility of stability analysis results.