Inverse analysis of geotechnical parameters using an improved version of non-dominated sorting genetic algorithm II

The accuracy of slope mechanical parameter values determines the accuracy of slope stability assessments and the effectiveness of protective measures. Traditional methods for obtaining geotechnical parameters cannot entirely take into consideration the size, disturbance, variability, and uncertainty of mechanical parameters. Hence, we aimed to propose two algorithmic models for the geotechnical parameters of highway slope materials based on an improved non-dominated sorting genetic algorithm II (NSGA-II) combined with a GA-optimised backpropagation (BP) neural network (NN),named BPGA- NSGA-Ⅱ model and improved BPGA- NSGA-Ⅱ model.The results indicated that compared with the original NSGA-II model, the improved model showed a substantial shift in the Pareto front, and the errors are smaller than the original model.Among the two improved models, the improved BPGA-NSGA-II model has higher inversion accuracy.Therefore, this study provides a scientifically effective method for improving the accuracy of the geotechnical parameter inversion for highway slope materials.