Prediction of Ultimate Bearing Capacity of Shallow Foundations onCohesionless Soils Using Back Propagation Neural Networks (BPNN)

Abstract

Abstract:
This study explores the potential of back propagation neural networks (BPNN) computing
paradigm to predict the ultimate bearing capacity of shallow foundations on cohesionless
soils. The data from 97 load tests on footings (with sizes corresponding to those of real
footings and smaller sized model footings) were used to train and validate the model. Five
parameters are considered to have the most significant impact on the magnitude of
ultimate bearing capacity of shallow foundations on cohesionless soil and are thus used as
the model inputs. These include the width of the footing, depth of embedment, length to
width ratio, dry or submerge unit weight and angle of internal friction of the soil. The
model output is the ultimate bearing capacity. Performance of the model was
comprehensively evaluated. The values of the performance evaluation measures such as
coefficient of correlation, root mean square error, mean absolute error reveal that the
model can be effectively used for the bearing capacity prediction. BPNN model is
compared with the values predicted by most commonly used bearing capacity theories.
The results indicate that the model perform better than the theoretical methods.
KEYWORDS: Ultimate bearing capacity; Shallow foundations; cohesionless soil; back
propagation neural network (BPNN); prediction