Optimization of helical soil nailing behaviors by response surface methodology and hybrid Coot optimization

Abstract

AbstractSoil nailing is one of the effective methods of slope stabilization, and its stability mainly depends on the properties of both the soil and soil nails. Optimizing soil nailing parameters is an effective task to avoid installation issues. This work uses four input parameters: different shaft diameters, surcharge pressure, helical pitch, and inclination angle. The design plan for the experimentation is conducted using Box Behnken design (BBD) of response surface methodology (RSM), performed in design expert software. In addition, RSM is used to determine the optimal design combinations. A total of 25 experimental runs are taking place, and the significance of the developed quadratic model is determined using the analysis of variance (ANOVA) test. Hybrid deep belief network‐based Coot optimization (DBN‐CO) is also used to optimize the pull‐out capacity and the safety factor during installation for more precised prediction. When compared to others the proposed hybrid DBN‐CO predictive model represents the better performance with regression R = 0.99. Also least MSE & RMSE and highest R & R2 values obtained by hybrid DBN‐CO. The obtained experimental outcomes from RSM are 1.46 of safety factor, and 7.55 kN pull‐out capacity. The predicted results from the proposed hybrid DBN‐CO for the safety factor are 1.44 and pull‐out capacity is 7.77 kN. As a result, the optimal results of DBN‐CO are 3% greater than RSM, DBN, and artificial neural network (ANN). Based on prediction approaches the proposed hybrid DBN‐CO results are in perfect agreement with experimental values and are additional superior to the RSM, DBN, and ANN.