Artificial Neural Networks For Settlement Prediction Of Shallow Foundations On Granular Soils

Mohamed Shahin

Artificial neural networks (ANNs) are a form of artificial intelligence, which, by means of their architecture, try to simulate the biological structure of the human brain and nervous system. In recent times, ANNs have been applied to many geotechnical engineering tasks and have demonstrated some degree of success. Over the years, many methods have been developed to predict the settlement of shallow foundations on granular soils. However, methods for making such predictions with the required degree of accuracy and consistency have not yet been developed. This can be attributed to the fact that the problem of estimating the settlement of shallow foundations on granular soils is complex and not yet entirely understood as a result of the uncertainty associated with the factors affecting such settlement. Among these factors are the distribution of applied stress, the stress-strain history of granular soils, soil compressibility and the difficulty of obtaining undisturbed samples of granular soils. Inaccurate estimation of settlement of shallow foundations on granular soils may also be due to the fact that most available methods for predicting such settlement are model driven, in which the structure of the model has to be established a priori before the unknown model parameters can be determined. This may potentially compromise model performance, as the form of the equation chosen may be sup-optimal. On the contrary, ANNs are a technique that uses the data a lone to determine the structure of the model as well as the unknown model parameters. Consequently, the need for predefined mathematical equations is negated and as a result, the use of ANNs may overcome the limitations of the existing methods.In the seminar ANNs were described (how they work and what they do) and some ANN applications in the field of geotechnical engineering were demonstrated. The feasibility of using ANNs to obtain more accurate settlement prediction of shallow foundations on granular soils was explored and their performance, with some of the most commonly used traditional methods, compared. The benefits and limitations of ANNs was also discussed.

Engineers Australia members participating in AGS technical sessions can record attendance on their personal CPD logs. Members should refer to Engineers Australia CPD policy for details on CPD types, requirements and auditing guidelines.