Use Of Artificial Neural Networks In Stability Control Of Cantilever Retaining Walls

Stability conditions effective on the design of the wall are safety factors of sliding, overturning and slope stability. Design of a cantilever retaining wall depends on great numbers of parameters such as soil properties and wall dimensions. In addition, design of the cantilever retaining wall must be satisfies stability conditions like safety factors of sliding, overturning and slope stability. It takes time to design wall considered all effective parameters with traditional methods. For this reason, some methods like artificial neural networks have been preferred due to reach the result in brief time, nowadays. In this study, algorithms of multi-layer artificial neural networks, generalized regression artificial neural networks and radial based artificial neural networks have been handled to investigate the model which provides to gain safety factors of sliding, overturning and slope stability. 1024 (45) different wall designs have been composed, for parameters of the length of base, the toe extension, the thickness of base, the angle of front face, the angle of internal friction which have four levels each of them. Training stage of artificial neural networks method has been completed by using this 1024 data set and testing stage of the method has been achieved for random 100 different wall designs. end of testing stage, safety factors of sliding, overturning and slope stability have been obtained for random 100 data set. Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and Determination Coefficient (R2) of obtained safety factors according to developed models by using the artificial neural networks algorithms have been compared for three algorithms. The model improved by algorithms of multi-layer artificial neural networks and radial based artificial neural networks have proper R2 value (0.99) when is compared the generalized regression artificial neural networks algorithm. This result show that improved model by means of Artificial Neural Networks can be used reliably and effectively in design of the cantilever retaining wall.