You are on page 1of 13

Machine Learning & Data Analytics (CE784A)

TOPIC
Support Vector Machine Approach for Longitudinal
Dispersion Coefficients in Natural Streams

Presented By
Rajeev Moond
20103089
ABSTRACT
• This paper presents the Support Vector Machine (SVM) approach to
predict the longitudinal Dispersion coefficients in natural rivers.
• Collected published data from the literature for the dispersion
coefficient for wide range of flow conditions are used for the
development and testing of the proposed method
• The proposed SVM approach produce satisfactory results with
coefficient of determination=0.9025 and root mean square error=0.0078
compared to existing predictors for dispersion coefficient
INTRODUCTION
• Longitudinal Dispersion of pollutants in rivers is significant to
practicing hydraulic and environmental engineers for designing
outfalls or water intakes
• They are also used for evaluating risks from accidental releases of
hazardous contaminants
• Fitness of models has been tested using the observed dispersion
coefficient as available in literature
• From the published results, it has been shown that the longitudinal
dispersion coefficients(LDP) vary within a wide range(1.9-2883.5)
• Investigation of quality condition of natural rivers by 1D mathematical
model requires the best estimations for LDP
• When measurements and real data of mixing processes in river are
available, the LDPs is determined simply
• But when they are not available, LDPs are determined by several simple
regressive equations
• Equations derived in the past are valid only in their calibrated ranges of
flow and geometry conditions
• These equations donot give good results for larger or smaller ranges
• Therefor, main aim of this write up is to develop the SVM for dispersion
coefficients and assessing the accuracy of these methods in comparison
with real data
Empirical equations for estimation of longitudinal dispersion coefficient
Reference Equation Author
Tayfour and Singh Kx=5.93HU* Elder
Deng et al Kx=0.58(H/U)2UB McQuivey and Keefer
Fisher et al Kx=0.01U2B2/HU* Fisher et al
Seo and Bake Kx=0.55BU*/H2 Li et al
Seo and Bake Kx=0.18(U/U*)0.5(B/H)2HU* Liu
Tavakollizadeh and Kx=2.0(B/H)1.5HU* Iwasa and Aya
Kashefipur

Seo and Cheong Kx=5.92(U/U*)1.43(B/H)0.62HU* Seo and Cheong

Sedighnezhad et al Kx=0.6(B/H)2HU* Koussis and


Rodriguez-Mirasol

FaghforMaghrebi Kx=0.2(B/H)1.3(U/U*)1.2HU* Li et al
and Givehchi

Rajeev and Dutta Kx/HU*=2(W/H)0.96(U/U*)1.25 Rajeev and Dutta


SUPPORT VECTOR REGRESSION
• When support vector machines were first used for classification, in
1996, another version of SVMs was proposed by Drucker et al
• The new SVM version contains all of the main features that
characterize the maximum margin algorithm, including a nonlinear
function that is leaned by linear learning machine mapping into high
dimensional kernel induced feature space.
• The capacity of the system is controlled by parameters that do not
depend on the dimensionality of the feature space
MODEL DEVELOPMENT
• The scenarios considered in building the SVM model inputs (flow
width (W)/flow depth (H)), flow velocity (U)/shear velocity (U*)) and
output (longitudinal dispersion coefficient (m2/s) Kx/flow depth
(H)×shear velocity (U*)
• From the collected data sets used in this study, around 60% (58 data
set) of these patterns were used for training while the remaining
patterns about 20% (20 data set) were used for testing, and about 20%
(18 data set) for validating, the SVM model
• The model parameters ˛i and ε were initially fixed as 1 and 0
• A genetic algorithm was used to obtain the optimal value of ε
• During the genetic search, an initial population of chromosomes was
created and the fitness of each candidate solution was evaluated
against the fitness function
• Then the population is evolved through multiple generations and the
optimal solution was selected
• Optimal ε is found to be 0.0001 for the present problem. The optimal
values of kernel parameters C and ε are found to be 0.35 and 20.0,
respectively
RESULTS AND DISCUSSION OF SVM

• Overall, particularly for field measurements, the SVM model gives better
predictions than the existing models
• The SVM model produced the least errors (R=0.95, R2 =0.9025 and
RMSE=0.00780)
• From Fig 2 (validation set) it is clear that the traditional predictor under or
over estimate the longitudinal dispersion coefficient
• SVM produced for test data correlation coefficient, R=(0.93), coefficient of
determination R2 (=0.8641) and root mean square error, (RMSE=2.234).
• It can be concluded that for all the data sets the SVM model give either
better or comparable results
Fig 1 Comparison of observed versus predicted Kx/HU* for training data using SVM
Fig. 2. Comparison of observed versus predicted Kx/HU* by SVM and Rajeev and Dutta for validation data set
CONCLUSIONS
• Longitudinal dispersion in rivers is a complex phenomenon
• Natural channels have bends, changes in shape, pools and many other
irregularities, all of which contribute significantly to the dispersion
process
• To overcome the complexity and uncertainty associated with the
dispersion, this research demonstrates that an SVM model can be applied
for accurate prediction of longitudinal dispersion coefficients
• The genetic programming will be used to predict longitudinal dispersion
coefficient in the future with more database
REFERENCES
1. Z.Q. Deng, V.P. Singh, L. Bengtsson, Longitudinal dispersion coefficient in single channel streams, Journal of
Hydraulic Engineering 128 (10) (2001) 901–916.
2. N. Ahsan, Estimating the coefficient of dispersion for a natural stream, World Academy of Science, Engineering
and Technology 44 (2008) 131–135.
3. H. Sedighnezhad, H. Salehi, D. Mohein, Comparison of different transport and dispersion of sediments in mard
intake by FASTER model, in: Proceedings of the Seventh International Symposium River Engineering, 16–18
October, Ahwaz, Iran, 2007, pp. 45–54.
4. I.W. Seo, K.O. Bake, Estimation of the longitudinal dispersion coefficient using the velocity profile in natural
streams, Journal of Hydraulic Engineering 130 (3) (2004) 227–236.
5. H.B. Fisher, E.J. List, R.C.Y. Koh, J. Imberger, N.H. Brooks, Mixing in Inland and Costal Waters, Academic Press
Inc., San Diego, 1979, pp. 104–138.
6. I.W. Seo, T.S. Cheong, Predicting longitudinal dispersion coefficient in natural stream, Journal of Hydraulic
Engineering 124 (1) (1998) 25–32.
7. A.J. Smola, B. Schölkopf, A Tutorial on Support Vector Regression, Royal Holloway College, London, UK,
NeuroCOLT Tech., Rep. TR 1998-030, 1988.
8. NEUROSOLTIONS 5.0, www.neurosolutions.com. NeuroDimension, Inc., 2009.

You might also like