Optimizing Slope Stability Assessment Using Hybrid BPSO - SVC with Kernel Function Evaluation

The complex nature of slope engineering presents considerable challenges in accurately predicting slope stability using traditional methodologies. Due to the serious implications that can arise from slope failures, it is crucial to implement the most effective techniques for assessing slope stability. This study investigates a hybrid approach that integrates BPSO with SVC to enhance predictive accuracy in slope stability assessment. The methodology employs BPSO to optimize the selection of features that are critical to the prediction process. Additionally, grid search technique is utilized for fine-tuning the hyperparameters of the SVC. The research evaluates the performance of three SVC kernel functions: linear, polynomial and rbf. For the predictive analysis, six features identified as potentially influential were selected: height of the slope (H), pore water ratio (ru), unit weight of the soil (Ƴ), cohesion of the soil (c), slope angle (β), and angle of internal friction (ɸ). To enhance the generalization capability of the classification models, a 5-fold cross-validation (CV) approach was implemented. The effectiveness of the models was evaluated using various metrics, including the area under the curve (AUC) and overall accuracy of the predictions. The findings of the study indicate that the hybrid approach, particularly the SVC employing the rbf kernel, significantly outperformed the other models in terms of prediction accuracy, achieving an AUC of 0.735 and an accuracy rate of 0.725. This underscores the potential of the proposed hybrid method as a valuable tool for accurately predicting slope stability and mitigating risks associated with slope failures in engineering applications.The complex nature of slope engineering presents considerable challenges in accurately predicting slope stability using traditional methodologies. Due to the serious implications that can arise from slope failures, it is crucial to implement the most effective techniques for assessing slope stability. This study investigates a hybrid approach that integrates BPSO with SVC to enhance predictive accuracy in slope stability assessment. The methodology employs BPSO to optimize the selection of features that are critical to the prediction process. Additionally, grid search technique is utilized for fine-tuning the hyperparameters of the SVC. The research evaluates the performance of three SVC kernel functions: linear, polynomial and rbf. For the predictive analysis, six features identified as potentially influential were selected: height of the slope (H), pore water ratio (ru), unit weight of the soil (Ƴ), cohesion of the soil (c), slope angle (β), and angle of internal friction (ɸ). To enhance the generalization capability of the classification models, a 5-fold cross-validation (CV) approach was implemented. The effectiveness of the models was evaluated using various metrics, including the area under the curve (AUC) and overall accuracy of the predictions. The findings of the study indicate that the hybrid approach, particularly the SVC employing the rbf kernel, significantly outperformed the other models in terms of prediction accuracy, achieving an AUC of 0.735 and an accuracy rate of 0.725. This underscores the potential of the proposed hybrid method as a valuable tool for accurately predicting slope stability and mitigating risks associated with slope failures in engineering applications.

Optimizing Slope Stability Assessment Using Hybrid BPSO - SVC with Kernel Function Evaluation Keywords:

support vector classifier , Machine Learning , slope stability , binary particle swarm optimization , grid search