840
Proceedings of the 18
th
International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013
representing the distribution of data in which the predicted
value equal to the measured value.
R
2
for models NN_Q
ult
,
Meyerhof’s formula (1976) and Briaud’s formula (1985)
respectively were 0.695: 0.421, and 0.399.
m
for the model
NN_Qult, Meyerhof’s formula (1976), Briaud’s formula (1985)
respectively were 0.673: 0.398, and 0.327.
5. CONCLUSIONS
The
R
2
value from the regression line generated by NN_Qult
closest to the data distribution when compared to the regression
line generated by the Meyerhof ‘s formula (1976) and Briaud’s
formula (1985). Q
ult
generated by NN_Q
ult
closest Q
ult
static
loading test results of the test when compared with Q
ult
that
produced Meyerhof’s formula (1976) and Briaud’s formula
(1985). This condition indicates that the predicted value of the
research model most closely with observed value.
The new calculation of the ultimate bearing capacity by the
artificial neural network model is given in chart form. The
design chart is used as a tool to calculate the ultimate bearing
capacity of a single pile in sand soil. This is particularly due to
the sensitivity analysis results indicated the suitability of
artificial neural network model with existing theories. The
results of the model have the highest performance among the
other methods, even though the difference is not too big.
6. REFERENCES
Bowles, J.E. (1988),
Foundation Analysis and Design
, Mc. Graw Hill
Book Company, Singapore, xix+1004p.
4.2.2 The Analytical Evaluation
In the evaluation of analytically there were 2 (two) values were
reviewed to calculate the mean value and standard deviation.
Mean (
) for the model NN_Qult, Meyerhof’s formula (1976),
and Briaud’s formula (1985) respectively were 1.27; 1.68, and
1.78. Standard deviation (
σ
) for the model NN_Qult,
Meyerhof’s formula (1976) and Briaud’s formula (1985)
respectively were 0.52; 0.34, and 0.33.
Coduto, D.P. (1994),
Foundation Design, Principles and Practices
,
Prentice Hall International, Inc., New Jersey, xx+796p.
Djajaputra, A.A (1997), Konsep Beban Terfaktor Dalam Perancangan
Tiang Bor,
Proceeding Seminar PILE’97
, Bandung, pp.14-1s/d14-3.
Fausett, L.(1994),
Fundamentals of Neural Networks (Architectures,
Algorithms, and Applications)
, Prentice Hall, xv+449p.
In this study, the statistical parameters used to evaluate the
performance of the method are coefficient of determination
(
R
2
), the gradient (
m
), mean (
), and standard deviation (
σ
). The
Rank Index (RI) was made to quantify the total performance of
each method. RI is the algebraic sum of the ratings obtained
from all of the criteria used (Titi and Farsakhs, 1999). RI values
closest to 1 (one) is considered as a method that has the best
performance. Table 4 is a recapitulation of all the statistical
parameters obtained from the calculations that have been done.
Three statistical parameters, namely
R
2
, m
, and
is considered
best when approximately equal to 1 (one), while for
σ
is
considered best when approximately equal to 0 (zero), so for
consistency of the calculation, then the special statistic
parameter
σ
, the value to be is the same compared with the
absolute value (1 -
σ
).
Griffith, D.V., G.A. Fenton, N. Manoharan (2002), Bearing Capacity of
Rough Rigid Strip Footing on Cohesive Soil: Probability Study,
Journal of Geotechnical and Geoenvironmental Engineering
, pp.
743-755.
Grima, M.A. (2000), Neuro-Fuzzy Modeling in Engineering Geology:
Applications to Mechanical Rock Excavation, Rock Strength
Estimation, and Geological Mapping,
PhD thesis
, Delft University
of Technology.
Hashash,Y.M.A., S. Jung, and J. Ghaboussi (2004), Numerical
Implementation of a Neural Network Based Material Model in Finite
Element Analysis,
International Journal for Numerical Methods in
Engineering
, pp. 59:989-1005.
Jeng,D.K., S.M. Bateni, and E. Lockett (2005), Neural Network
Assessment for Scour Depth Around Bridge Piers,
Research Report
No R855, Department of Civil Engineering Sydney NSW, Australia.
Kasabov, N.K. (1998),
Foundations of Neural Networks, Fuzzy Systems,
and Knowledge Engineering
, The MIT Press, x+419p.
Kurup, P.U. and N.K. Dudani (2002), Neural Networks for Profilling
Stress History of Clays from PCPT Data,
Journal of Geotechnical
and Geoenvironmental Engineering
, pp. 569-579.
Nugroho, A.S. (2003), Pengantar Soft Computing, Kuliah umum ilmu
komputer.com.
Prakash, S. and H.D. Sharma (1990),
Pile Foundations in Engineering
Practice
, John Wiley & Sons, Inc., xxx+734p.
Table 4. Perform Evaluation of Some Models
Prakoso, W.A. (2006), Desain Pondasi Dalam Berbasis Keandalan dan
SNI 03-6747-2002,
Prosiding Pertemuan Ilmiah Tahunan – X (PIT
– X) Himpunan Ahli Teknik Tanah Indonesia
, Jakarta, pp. 121-130.
Rahman, M.S., and M. Mulla (2005),
Fuzzy Neural Network Models for
Geotechnical Problems
, NCDOT Research Project, USA.
Samui, P. and B. Kumar (2006), Artificial Neural Network Prediction of
Stability Numbers for Two-Layered Slopes With Associated Flow
Rule,
Electronic
Journal
Geotechnical
Engrg
,
Referring to Table 4, it appears that for the model results
(
NN_Q
ult
) provide RI value is the most closed to 1 (one) or the
optimum value, so that it can be said that the model results of
the research has the highest performance among the methods
are comparable, despite differences in RI values is not too big.
Shahin,M.A., M.B. Jaksa, and H.R. Maier (2002a),
Artificial Neural
Network-Based Settlement Prediction Formula For Shallow
Foundations On Granular Soils
,Australian Geomechanics, pp.45-52.
Shahin,M.A., H.R. Maier, and M.B. Jaksa (2002b), Predicting
Settlement of Shallow Foundations Using Neural Networks,
Journal of Geotechnical and Geoenvironmental Engineering
, pp.
785-793.
4.2.3 Design Chart Based on Final Model
Based network architecture that has been verified by sensitivity
analysis and has been calibrated with the results of static load,
so that created a graph that is expected to be used for initial
design purposes. Model NN_Qult produce design charts. One
example of the design chart shown in Figure 10.
Shahin, M.A., M.B. Jaksa, and H.R. Maier (2005), Neural Network
Based Stochastic Design Charts for Settlement Prediction,
Canadian
Geotechnical Journal
, pp. 42:110-120.
Somantri, A. dan S.A. Muhidin (2006),
Aplikasi Statistika Dalam
Penelitian
, Penerbit Pustaka Setia Bandung, 410p.
Wang, X., B. Li, D. Lockington, D. Pullar, and D.S. Jeng (2005),
Self-
Organizing Polynomial Neural Network for Modeling Complex
Hydrological Processes
, Research Report No. R861, Department of
Civil Engineering Sydney NSW, Australia.
Figure 10. Example of
Design Chart
of
NN_Q
ult
Model
