Development of stability evaluation system for retaining walls: Differential evolution algorithm–artificial neural network

The objective of this study is to develop a Stability Evaluation System for retaining walls to assess their safety in real-time during excavation. A ground investigation is typically conducted before construction to gather information about the soil properties and predict wall stability. However, these properties may not accurately reflect the actual ground being excavated. To address this issue, the study employed a differential evolution algorithm to estimate the soil parameters of the actual ground. The estimated results were then used as input for an artificial neural network to evaluate the stability of the retaining walls. The study achieved an average accuracy of over 90% in predicting differential settlement, wall displacement, anchor force, and structural stability of the retaining walls. If implemented at actual excavation sites, this approach would enable real-time prediction of wall stability and facilitate effective safety management. Overall, the developed Stability Evaluation System offers a promising solution for ensuring the stability of retaining walls during construction. By incorporating real-time soil parameter analysis, it enhances the accuracy of stability predictions and contributes to proactive safety management in excavation projects.