No abstract

PurposeThis paper aims to develop an effective framework which combines Bayesian optimized convolutional neural networks (BOCNN) with Monte Carlo simulation for slope reliability analysis.Design/methodology/approachThe Bayesian optimization technique is firstly used to find the optimal structure of CNN based on the empirical CNN model established in a trial and error manner. The proposed methodology is illustrated through a two-layered soil slope and a cohesive slope with spatially variable soils at different scales of fluctuation.FindingsThe size of training data suite, T, has a significant influence on the performance of trained CNN. In general, a trained CNN with larger T tends to have higher coefficient of determination (R2) and smaller root mean square error (RMSE). The artificial neural networks (ANN) and response surface method (RSM) can provide comparable results to CNN models for the slope reliability where only two random variables are involved whereas a significant discrepancy between the slope failure probability (Pf) by RSM and that predicted by CNN has been observed for slope with spatially variable soils. The RSM cannot fully capture the complicated relationship between the factor of safety (FS) and spatially variable soils in an effective and efficient manner. The trained CNN at a smaller the scale of fluctuation (λ) exhibits a fairly good performance in predicting the Pf for spatially variable soils at higher λ with a maximum percentage error not more than 10%. The BOCNN has a larger R2 and a smaller RMSE than empirical CNN and it can provide results fairly equivalent to a direct Monte Carlo Simulation and therefore serves a promising tool for slope reliability analysis within spatially variable soils.Practical implicationsA geotechnical engineer could use the proposed method to perform slope reliability analysis.Originality/valueSlope reliability can be efficiently and accurately analyzed by the proposed framework.

This paper proposes a methodology for reliability analysis of seismic slope stability that incorporates interactions among multiple sliding blocks. The primary sliding direction is first determined using the vector sum method and then the imbalance thrust force along the primary sliding direction is calculated using the slice-wise strategy and, finally, the double integration strategy is adopted to calculate the accumulated sliding displacement within the earthquake duration. The interactions among multiple sliding blocks are incorporated by checking the potential of occurrence for each of the multiple sliding modes. The proposed method is applied to a soil slope with two sliding surfaces. The comparative studies demonstrate that the mean and standard deviation of the sliding displacement considering the interaction of multiple sliding blocks are approximately three times larger than that of a single sliding mode, and the COV (mean value divided by standard deviation) of the two are slightly different. For the single sliding mode, the mean and standard deviation of the sliding displacement calculated using the proposed method are about 1/2 of the traditional Newmark sliding block model, and the failure probability obtained by the proposed method is lower than that from the traditional Newmark sliding block model owing to the difference in the sliding direction. The Peak Ground Acceleration (PGA) exhibits a significant effect on the statistics of 10,000 sliding displacements. The interactions among multiple sliding blocks and the PGA are required to be properly considered in seismic slope reliability analysis.