Improving Performance of Retaining Walls Under Dynamic Conditions Developing an Optimized ANN Based on Ant Colony Optimization Technique

In this paper, a combination of articial neural network and ant colony optimization (ANNACO) was used for dynamic conditions of retaining wall structures. The retaining walls produce different responses to dynamic loads. The applied data of this study comprising of wall height and thickness, soil density, internal friction angle, and stone density. The walls were designed in a variety of dynamic conditions. Various conditions were considered for the design of the retaining wall structures. Then, an extended data set was created for the next step. After that, the new systems were implemented using optimized articial intelligence techniques. The neural network provided strong relationships between various wall parameters. The design of various networks in the present research led to the best evaluation of the dynamic conditions of the retaining walls. Under these conditions, an ACO was used for optimal design. Effects of parameters varied due to different wall conditions when dynamic loads were considered. Therefore, the impact of parameters was evaluated using hybrid ANN-ACO to increase the efciency. These designs provided more control over dangers by dynamic loads of a retaining wall structure.

ACO, ANN, optimization, prediction, retaining wall, safety factor.