You are on page 1of 7

GRD Journals | Global Research and Development Journal for Engineering | Emerging Research and Innovations in Civil Engineering

(ERICE - 2019) | February 2019


e-ISSN: 2455-5703

Optimizing Reservoir Capacity, Water Allocation


and Crop Yield using Teaching Learning Based
Optimization (TLBO) Technique
1Vijendra
Kumar 2S. M. Yadav
1
Research Scholar 2Professor
1,2
Department of Civil Engineering
1,2
Sardar Vallabhbhai National Institute and Technology, Surat, Gujarat, India

Abstract

In the present study ‘Teaching Learning Based Optimization’ (TLBO) optimization method has been applied to the water resources
engineering problem. TLBO is a population-based natural-inspired evolutionary algorithm comparatively simple, easy and robust.
TLBO algorithm is capable of providing a global solution. Four water resources problem such as optimizing crop water demand,
maximization of benefits, minimization of reservoir capacity and minimization of reservoir capacity with evaporation losses solved
using TLBO technique. The results were compared with linear programming & dynamic programming solutions. TLBO algorithm
has proven to be providing the global and better results. The results obtained from TLBO were better in reservoir capacity problem
with evaporation losses. The results were satisfactory for optimizing crop water demand, maximization of benefits and
minimization of reservoir capacity. The TLBO technique provides a satisfactory solution as other popular optimization techniques.
Keyword- TLBO, LINGO Software, Soft Computing, Linear Programming, Dynamic Programming
__________________________________________________________________________________________________

I. INTRODUCTION
Soft computing techniques are the leading methods for solving water resources complex problems. The water resources problems
are solved using fuzzy logic, artificial neural networks, machine learning, probabilistic reasoning, population-based algorithm and
neighbourhood-based algorithm etc. Population-based algorithms are swarm intelligence and evolutionary computation.
Evolutionary computations are evolutionary programming, differential evolution, genetic algorithm, genetic programming,
evolutionary strategies, artificial immune algorithm and bacteria foraging optimization etc. Swarm intelligence based algorithms
examples are artificial bee colony, particle swarm optimization, ant colony optimization, firefly etc. Neighbourhood-based
algorithms are simulated annealing and tabu search (Venkata Rao 2016; Kumar and Yadav 2018).
Various algorithms have been used to solve reservoir operation problem such as Fuzzy logic (Russell and Campbell 1996),
genetic algorithm (Chang et al. 2010; Fallah-Mehdipour et al. 2012; Ashofteh et al. 2015), particle swarm optimization (Nagesh
Kumar and Janga Reddy 2007; SaberChenari et al. 2016; Bai et al. 2017) and Firefly (Garousi-Nejad et al. 2016).
Soft computing techniques have been applied for studying Rainfall-runoff model, artificial neural network (Nourani
2017), genetic algorithm (Wu et al. 2012), particle swarm optimization (Taormina et al. 2012), fuzzy logic (Talei et al. 2010) and
genetic programming (Rodríguez-Vázquez et al. 2012). Stage forecasting and prediction have been studied using machine learning
(Wu et al. 2008; Taghi Sattari et al. 2013), artificial neural network (Deo et al. 2000) and particle swarm optimization (Chau 2006).
Flood forecasting and prediction have been studied using an artificial neural network (Maier and Dandy 2000; Pramanik and Panda
2009; Yazdani and Zolfaghari 2014; Wu et al. 2010), machine learning (Yu et al. 2006) and genetic algorithm (Sahay and
Srivastava 2014).
The main objective of the paper is to study the preformation of the recently developed algorithm i.e. teaching learning-
based optimization. Four different problems such as optimizing crop water demand, maximization of benefits, minimization of
reservoir capacity and minimization of reservoir capacity with evaporation losses have been solved using TLBO technique. Apart
the manual working procedure has been also shown for better understating of the algorithm.

II. METHODOLOGY

A. Teaching Learning Based Optimization (TLBO)


TLBO mimics a class, based on the teaching-learning process. Just like in class, teachers teach and students learn. To obtain an
excellent result, a good and strong dependency is required between students and the teachers. Thus, every student tries to mimic
or follow the teachers and improve its result. Even in class students try to interact with other students and improve its own result.

All rights reserved by www.grdjournals.com 354


Optimizing Reservoir Capacity, Water Allocation and Crop Yield using Teaching Learning Based Optimization (TLBO) Technique
(GRDJE / CONFERENCE / ERICE - 2019 / 070)

Based on this concept TLBO was developed. The procedure of TLBO is divided into two phases: The First ‘Teaching phase’ and
the second ‘Learning phase’. A flowchart of the TLBO algorithm is given in Figure. 1.
Step 1: Initialization
TLBO is similar to other random search algorithms. The first step is to decide the population size (number of students in
a class) and a number of iterations (termination criteria). Initial solutions are randomly generated using the formula presented in
Eq. (1). Identify the best solution (teacher) from the list of the corresponding objective function. The best solution is taken as the
maximum for maximization problem and minimum for the minimization problem.
Randomly Generated Population = [L + r (U − L)] (1)
Where: L = lower bound on the variable; U = upper bound on the variable; r = random number between [0, 1]
Step 2: Teaching Phase
In teaching phase teacher tries to improve the result of populations using difference mean method based on the best
solution.
Difference mean = r ( Xbest − Xmean ) (2)
Where: X best = best solution (teacher); r = random number between [0, 1]; X mean = mean of all the students.
Based on the difference mean the old solutions are modified using Eq. (3).
Xnew = (Xold + Difference mean) (3)
Where: X new = modified solution; X old = old solution
The better solutions are accepted and replaced with the previous.

Fig. 1: Flowchart of TLBO algorithm

All rights reserved by www.grdjournals.com 355


Optimizing Reservoir Capacity, Water Allocation and Crop Yield using Teaching Learning Based Optimization (TLBO) Technique
(GRDJE / CONFERENCE / ERICE - 2019 / 070)

Step 3: Learning phase


In the second phase, a learner interacts with any random learner to increase their knowledge. For example, select any two
random solutions as Xi and Xj, where i ≠ j than two conditions can occur.
if f (Xi ) < f (X j );
Then, Xnew,i = Xold,i + ri (X j − Xi ) (4)
if f (Xi ) > f (Xj );
Then, Xnew,i = Xold,i + ri (X i − Xj ) (5)
The better solutions are accepted and replaced with the previous. Stop if the maximum generation number is
accomplished, otherwise repeat from teacher’s phase.

III. IMPLEMENTATION OF PROPOSED ALGORITHM AND PERFORMANCE EVALUATION


In the present study, TLBO algorithm has been applied to water resources engineering problems. The optimum solution obtained
using TLBO compared with the LP model and the DP model. Four different test functions are considered to check the performance
of the TLBO algorithm.

A. Linear Programming Problem (LPP)


LPP selected is to identify an optimum pattern of irrigation by maximization of profit to the minimal quantity of water consumed.
Demand for production is observed in linear trend as; A= 0.5* (X-2) + 2 and B = 0.6* (Y-3) + 3. Where A and B are the demand
for production and X and Y are the water requirements for crop 1 and 2, respectively. Constrain as; profit from one unit of crop 1
and 2 are measured four and five units respectively (Nagesh, 2014). Other details are given in Table 1.
Problem Variable bounds Objective function Constraints Remark
LPP X Є [2, 10] Maximize c1(x) = X+Y - 10 ≤ 0
Maximization of crop
Y Є [3, 10] f(x) = 2X +3Y +10
production
DPP xi Є [0, 4] Maximize 𝑐1 (𝑥) = 𝑥1 + 𝑥2 + 𝑥3 = 4
Water allocation
i= 1,….,3 𝑓 = 5 𝑥1 − 0.5 𝑥12
𝑓(2)
= 8 𝑥2 − 0.5 𝑥22
𝑓(3) = 7 𝑥3 − 𝑥32
RCP xi Є [0, 10] Minimize 𝑐1 (𝑥) = 𝑆1 + 4 − 5 = 𝑆2 ;
i= 1,….,6 f(x) = k 𝑐2 (𝑥) = 𝑆2 + 8 − 0 = 𝑆3 ; Reservoir capacity
k Є [0, 10] 𝑐3 (𝑥) = 𝑆3 + 7 − 5 = 𝑆4 ;
𝑐4 (𝑥) = 𝑆4 + 3 − 6 = 𝑆5 ;
𝑐5 (𝑥) = 𝑆5 + 2 − 2 = 𝑆6 ;
𝑐6 (𝑥) = 𝑆6 + 0 − 6 = 𝑆1 ;
𝑆𝑖 ≤ 𝑘; 𝑖 = 1, … … ,6
Generated with Lingo Minimize 𝑐𝑡 (𝑥) = (1 − 𝑎𝑡 ) 𝑆𝑡 + 𝑄𝑡 − 𝐿𝑡 − 𝑅𝑡
RCPEL
software f(x) = k = (1 + 𝑎𝑡 ) 𝑆𝑡+1 Reservoir capacity considering
𝑆𝑡 ≤ 𝑘 losses
𝑅𝑡 ≥ 𝐷𝑡 𝑓𝑜𝑟 𝑎𝑙𝑙 𝑡 = 1, … … ,12;
𝑆𝑡+1 = 𝑆1
Table 1: Test problems used in the study

B. Dynamic Programming Problem (DPP)


The maximum carrying capacity of the canal is 4 units (Nagesh 2014). Find an optimum supply of water to each crop such that net
benefits can be maximized. The objective function and the constraint are shown in Table 1.
Net benefit Z= f (1) + f (2) + f (3)
Here, TLBO algorithm was compared with dynamic programming. This problem can also be solved by discrete TLBO.

C. Reservoir Capacity Problem (RCP)


Find the reservoir capacity without evaporation losses. Required data are given in Table 2 (Vedula and Mujumdar 2005). The
objective function and the constraints are shown in Table 1.
Period, t 1 2 3 4 5 6
Inflow, 𝑄𝑡 4 8 7 3 2 0
Demand, 𝐷𝑡 5 0 5 6 2 6
Table 2: Reservoir Data

All rights reserved by www.grdjournals.com 356


Optimizing Reservoir Capacity, Water Allocation and Crop Yield using Teaching Learning Based Optimization (TLBO) Technique
(GRDJE / CONFERENCE / ERICE - 2019 / 070)

D. Reservoir Capacity Problem with Evaporation Losses (RCPEL)


The main objective is to minimize the reservoir capacity losses. The objective function and the constraints are shown in Table 1.
Required data for the calculation are presented in Table 3 (Vedula and Mujumdar 2005).
e
Where, a t = a ∗ t⁄2; ‘a’ is the surface area per unit active storage or slope = 0.117115; Lt = et ∗ A0 ; ‘A0 ’ Area at the
2
dead storage level (Mm ); a t = Surface per unit time; Lt = Fixed evaporation loss; R t = Release.
Month June July August September October November
Evaporation 𝑒𝑡 (mm) 231.81 147.57 147.57 152.14 122.96 121.76
Inflow, 𝑄𝑡 (𝑀𝑚3 ) 70.61 412.75 348.40 142.29 103.78 45.00
Demand, 𝐷𝑡 (𝑀𝑚3 ) 51.68 127.85 127.85 65.27 27.18 203.99
Month December January February March April May
Evaporation 𝑒𝑡 (mm) 99.89 97.44 106.17 146.29 220.97 246.75
Inflow, 𝑄𝑡 (𝑀𝑚3 ) 19.06 14.27 10.77 8.69 9.48 18.19
Demand, 𝐷𝑡 (𝑀𝑚3 ) 203.99 179.47 89.76 0 0 0
Table 3. Reservoir data
Fixing the upper bounds and lower bounds of the variable is the most important to solve in evolutionary technique. Total
26 unknown variables are present in the RCPEL problem. Since it doesn’t have any upper bounds and lower bounds for the
variables. LINGO software is used to generate the unknown variables. Therefore, the results obtained for the variables from LINGO
are used for making the upper and lower bounds and later run through TLBO.

IV. RESULTS AND DISCUSSION

A. Linear Programming Problem


To show the manual working procedure of TLBO algorithm, linear programming problem has been considered. The steps are
shown below.
Step 1: Assume populations n=10 and one iteration (termination criteria). The initial populations are randomly generated within
the range of variable using Eq. 1. Table 4 shows the variables and the equivalent objective/target function.
Where, f(x) and c(x) represents the objective function and constraint function, respectively. If c(x) is less than or equal to zero,
then it satisfied the condition else it needs to put in for a penalty. The term p(x) is the penalty functions and need to deduct from
f(x). Penalty depends upon the decision maker. It can be observed from Table 4, starting first three population penalties are zero
since it satisfied the condition. For successive population penalty have been applied as per p(x) = 10* (c(x)) 2. The f '(x) is the final
target function after the penalty that needs to consider. Example X=5.2, Y=5.8; f(x) = 2*5.2 + 3*5.8 + 10 = 37.8; c(x) = 5.2 + 5.8
-10 = 1, which is more than zero; p(x) = 10* (1) 2 = 10; f '(x) = f(x) – p(x) = 37.8 – 10 = 27.8.
X Y f(x) c(x) p(x) f '(x)
2.8 3.7 26.7 -3.5 0 26.7
3.6 4.4 30.4 -2 0 30.4
4.4 5.1 34.1 -0.5 0 34.1
5.2 5.8 37.8 1 10 27.8
6 6.5 41.5 2.5 62.5 -21
6.8 7.2 45.2 4 160 -114.8
7.6 7.9 48.9 5.5 302.5 -253.6
8.4 8.6 52.6 7 490 -437.4
9.2 9.3 56.3 8.5 722.5 -666.2
10 10 60 10 1000 -940
Mean 6.4 6.85
Table 4: Initial population
Step 2: Calculate the mean of X and Y, i. e. X= 6.4 and Y= 6.85. Compute the difference mean for X and Y using Eq. (2). As it is
a maximization problem, the highest of f '(x) is selected. Assuming random numbers r1 = 0.102 and r2 = 0.05 for X and Y.
Difference mean (X) = 0.102*(4.4 - 6.4) = -0.204
Difference mean (Y) = 0.05 *(5.1 - 6.85) = -0.0875
The difference mean X and Y is added to the corresponding variables of column X and Y as per Eq. (3). Table 5 shows the new
variables X and Y, and the equivalent target function. Example: The new variant of X1 = 2.8 + (-0.204) = 2.596 and Y1 = 3.7 + (-
0.0875) = 3.6125.
X Y f(x) c(x) p(x) f '(x)
2.596 3.612 26.029 -3.791 0 26.029
3.396 4.312 29.729 -2.291 0 29.729
4.196 5.012 33.429 -0.791 0 33.429
4.996 5.712 37.129 0.708 5.019 32.109
5.796 6.412 40.829 2.208 48.774 -7.945
6.596 7.112 44.529 3.708 137.530 -93.000
7.396 7.812 48.229 5.208 271.285 -223.06

All rights reserved by www.grdjournals.com 357


Optimizing Reservoir Capacity, Water Allocation and Crop Yield using Teaching Learning Based Optimization (TLBO) Technique
(GRDJE / CONFERENCE / ERICE - 2019 / 070)

X Y f(x) c(x) p(x) f '(x)


8.196 8.512 51.929 6.708 450.040 -398.11
8.996 9.212 55.629 8.208 673.795 -618.17
9.796 9.912 59.329 9.708 942.550 -883.22
Table 5: New variables and the equivalent target function (teacher phase)
Step 3: The results obtained in Table 4 and 5 are compared row-wise based on the objective function f ′(x). Since it’s a
maximization problem, the target function should be more. This completes the teacher phase. Table 6 shows the updated variables
and the objective function derived from Table 4 and 5.
X Y f(x) c(x) p(x) f '(x)
2.8 3.7 26.7 -3.5 0 26.7
3.6 4.4 30.4 -2 0 30.4
4.4 5.1 34.1 -0.5 0 34.1
4.99 5.712 37.12 0.70 5.01 32.10
5.79 6.412 40.82 2.20 48.77 -7.94
6.59 7.112 44.52 3.70 137.53 -93.00
7.39 7.812 48.22 5.20 271.28 -223.06
8.19 8.512 51.92 6.70 450.04 -398.11
8.99 9.212 55.62 8.20 673.79 -618.17
9.79 9.912 59.32 9.70 942.55 -883.22
Table 6: Updated variables and the target function
Step 4: Subsequent section is the learner phase. Here any student can interact with any other students to improve their respective
results. From Table 6 randomly any two rows are taken and compared based on f '(x). In this case, interaction has been considered
between 1 and 2, 2 and 3, 3 and 4 etc. Eq. (4), and (5) are used to obtain the new X and Y variables. Assuming random numbers
for X and Y as r1 = 0.81 and r2 = 0.94 respectively. The new variables and the equivalent target function are shown in Table 7. For
example; as it is a maximization problem, higher the f '(x) better is the result. When we compared the learner 1 and 2, it is found
that learner 2 is better. Thus, learner 2 attempts to improve the performance and the result of learner 1. The new variable of learner
1 can be calculated as,
X1 = 2.8 + 0.81* (3.6-2.8) = 3.448
Y1 = 3.7 + 0.94* (4.4-3.7) = 4.358
X Y f(x) c(x) p(x) f '(x) Interaction
3.448 4.358 29.970 -2.194 0 29.97 1-2
4.248 5.058 33.670 -0.694 0 33.67 2-3
3.917 4.524 31.407 -1.558 0 31.40 3-4
4.348 5.054 33.859 -0.597 0 33.85 4-5
5.148 5.754 37.559 0.902 8.145 29.41 5-6
5.948 6.454 41.259 2.402 57.720 -16.46 6-7
6.748 7.154 44.959 3.902 152.295 -107.34 7-8
7.548 7.854 48.659 5.402 291.870 -243.21 8-9
8.348 8.554 52.359 6.902 476.445 -424.09 9-10
1.861 0.594 15.506 -7.544 569.121 -553.61 10-1
Table 7: New variables and the equivalent target function
Step 5: The f '(x) of Table 6 and 7 are compared and the best is selected. The respective variables are presented in Table 8.
X Y f(x) c(x) p(x) f '(x)
3.448 4.358 29.970 -2.194 0 29.970
4.248 5.058 33.670 -0.694 0 33.670
4.400 5.100 34.100 -0.500 0 34.100
4.348 5.054 33.859 -0.597 0 33.859
5.148 5.754 37.559 0.902 8.145 29.414
5.948 6.454 41.259 2.402 57.720 -16.460
6.748 7.154 44.959 3.902 152.295 -107.335
7.548 7.854 48.659 5.402 291.87 -243.210
8.348 8.554 52.359 6.902 476.445 -424.085
1.861 0.594 15.506 -7.544 569.121 -553.614
Table 8: Updated variables and the equivalent target function
This completes one iteration of the TLBO algorithm. At the end of the first iteration, the results of f '(x) have been increasing as
compared to the beginning of the iteration. Increasing the number of iterations will give the optimum solution. Code of the TLBO
algorithm was done in MATLAB R2014b.
The population is taken as 200 and iteration as 50 for TLBO. The results obtained from TLBO, LINGO software and
linear programming (LP) (Nagesh 2014) are tabulated in Table 9. It was found that the result obtained using TLBO was same as a
global solution obtained from the LP model.
LINGO
TLBO (Units) Linear Programming (Units) (Nagesh 2014)
software

All rights reserved by www.grdjournals.com 358


Optimizing Reservoir Capacity, Water Allocation and Crop Yield using Teaching Learning Based Optimization (TLBO) Technique
(GRDJE / CONFERENCE / ERICE - 2019 / 070)

Max f(i) 38.00 38.00 38.00


X 2.00 2.00 2.00
Y 8.00 8.00 8.00
Table 9: Comparison of TLBO and LP model for LPP

B. Dynamic Programming Problem (DPP)


The above problem is solved using TLBO method. The outputs of TLBO and dynamic programming problem are presented in
Table 10. For this problem three variables are present, the population is taken as 200 and the iteration as 3 for TLBO. It was found
that the results obtained using TLBO was same as a global solution obtained from the DP model.
TLBO Dynamic Programming (Nagesh 2014)
Max Z 21.0000 21
𝑥1 0.86097 ~ 1 1
𝑥2 1.00468 ~ 1 1
𝑥3 1.96467 ~ 2 2
Table 10: TLBO and Dynamic Programming Optimization

C. Reservoir Capacity Problem (RCP)


There are 7 variables and problem is to minimize reservoir capacity. The population is taken as 200 and iteration as 100 for TLBO.
The outputs of TLBO and linear programming problem are presented in Table 11. The solution given by both the methods matches
well with each other.
Linear Programming LINGO
TLBO
(Vedula and Mujumdar 2005) software
Min k 10 10 10
Table 11: TLBO and Linear Programming Optimization

D. Reservoir Capacity Problem with Evaporation Losses (RCPEL)


The results obtained from TLBO and LP are tabulated in Table 12. LP has been computed using LINGO software. The total
variables are 25, the population is taken as 200 and iteration as 100 for TLBO. It can be observed from the result that TLBO has
performed better than the LP model.
LINGO Software (Mm3) TLBO (Mm3)
Min k 617.986 609.999
Table 12: Comparison of TLBO and LP model for RCPEL

V. CONCLUSION
TLBO is a population-based algorithm. It has been tested for both constraint and unconstraint, linear and dynamic programming
problem satisfactorily. The TLBO algorithm was used to solve four cases. In each case, the solution matches well with the solution
obtained using other optimization techniques such as Linear and Dynamic programming. The results were satisfactory for
optimizing crop water demand, maximization of benefits and minimization of reservoir capacity. The results obtained from TLBO
were better in reservoir capacity problem with evaporation losses. Based on the results it can be concluded that TLBO algorithm
was 100% compliance with the known global optimal solution.

REFERENCE
[1] Afshar MH (2012) Large scale reservoir operation by Constrained Particle Swarm Optimization algorithms. J Hydro-
Environment Res 6:75–87. doi: 10.1016/j.jher.2011.04.003
[2] Ashofteh P-S, Haddad OB, Loáiciga HA (2015) Evaluation of Climatic-Change Impacts on Multiobjective Reservoir
Operation with Multiobjective Genetic Programming. J Water Resour Plan Manag 141:04015030. doi:
10.1061/(ASCE)WR.1943-5452.0000540
[3] Bai T, Kan Y, Chang J, et al (2017) Fusing feasible search space into PSO for multi-objective cascade reservoir optimization.
Appl Soft Comput 51:328–340. doi: 10.1016/j.asoc.2016.12.005
[4] By E, Thirumalaiah K, Deo MC (2000) Hydrological Forecasting Using Neural Networks. J Hydrol Eng 5:180–189. doi:
10.1061/(ASCE)1084-0699(2000)5:2(180)
[5] Chang LC, Chang FJ, Wang KW, Dai SY (2010) Constrained genetic algorithms for optimizing multi-use reservoir operation.
J Hydrol 390:66–74. doi: 10.1016/j.jhydrol.2010.06.031
[6] Chau KW (2006) Particle swarm optimization training algorithm for ANNs in stage prediction of Shing Mun River. J Hydrol
329:363–367. doi: 10.1016/j.jhydrol.2006.02.025
[7] Fallah-Mehdipour E, Bozorg Haddad O, Mariño MA (2012) Real-Time Operation of Reservoir System by Genetic
Programming. Water Resour Manag 26:4091–4103. doi: 10.1007/s11269-012-0132-z

All rights reserved by www.grdjournals.com 359


Optimizing Reservoir Capacity, Water Allocation and Crop Yield using Teaching Learning Based Optimization (TLBO) Technique
(GRDJE / CONFERENCE / ERICE - 2019 / 070)

[8] Garousi-Nejad I, Bozorg-Haddad O, Loáiciga HA, Mariño MA (2016) Application of the Firefly Algorithm to Optimal
Operation of Reservoirs with the Purpose of Irrigation Supply and Hydropower Production. J Irrig Drain Eng 142:04016041.
doi: 10.1061/(ASCE)IR.1943-4774.0001064
[9] H. Kashani M, Ghorbani MA, Dinpashoh Y, Shahmorad S (2016) Integration of Volterra model with artificial neural networks
for rainfall-runoff simulation in forested catchment of northern Iran. J Hydrol 540:340–354. doi:
10.1016/j.jhydrol.2016.06.028
[10] Kumar V, Yadav SM (2018) Optimization of Reservoir Operation with a New Approach in Evolutionary Computation Using
TLBO Algorithm and Jaya Algorithm. Water Resour Manag 32:4375–4391. doi: 10.1007/s11269-018-2067-5
[11] Maier HR, Dandy GC (2000) Neural networks for the prediction and forecasting of water resources variables: A review of
modelling issues and applications. Environ Model Softw 15:101–124. doi: 10.1016/S1364-8152(99)00007-9
[12] Mukerji A, Chatterjee C, Raghuwanshi NS (2009) Flood Forecasting Using ANN, Neuro-Fuzzy, and Neuro-GA Models. J
Hydrol Eng 14:647–652. doi: 10.1061/(ASCE)HE.1943-5584.0000040
[13] Nagesh Kumar D, Janga Reddy M (2007) Multipurpose Reservoir Operation Using Particle Swarm Optimization. J Water
Resour Plan Manag 133:192–201. doi: 10.1061/(ASCE)0733-9496(2007)133:3(192)
[14] Nourani V (2017) An Emotional ANN (EANN) approach to modeling rainfall-runoff process. J Hydrol 544:267–277. doi:
10.1016/j.jhydrol.2016.11.033
[15] Ostadrahimi L, Mariño MA, Afshar A (2012) Multi-reservoir Operation Rules: Multi-swarm PSO-based Optimization
Approach. Water Resour Manag 26:407–427. doi: 10.1007/s11269-011-9924-9
[16] Pramanik N, Panda RK (2009) Application of neural network and adaptive neuro-fuzzy inference systems for river flow
prediction. Hydrol Sci J 54:247–260. doi: 10.1623/hysj.54.2.247
[17] Rodríguez-Vázquez K, Arganis-Juárez ML, Cruickshank-Villanueva C, Domínguez-Mora R (2012) Rainfall–runoff
modelling using genetic programming. J Hydroinformatics 14:108. doi: 10.2166/hydro.2011.105
[18] Russell SO, Campbell PF (1996) Reservoir Operating Rules with Fuzzy Programming. J Water Resour Plan Manag 122:165–
170. doi: 10.1061/(ASCE)0733-9496(1996)122:3(165)
[19] SaberChenari K, Abghari H, Tabari H (2016) Application of PSO algorithm in short-term optimization of reservoir operation.
Environ Monit Assess 188:667. doi: 10.1007/s10661-016-5689-1
[20] Sahay RR, Srivastava A (2014) Predicting Monsoon Floods in Rivers Embedding Wavelet Transform, Genetic Algorithm and
Neural Network. Water Resour Manag 28:301–317. doi: 10.1007/s11269-013-0446-5
[21] Smith J, Eli RN (1995) Neural-Network Models of Rainfall-Runoff Process. J Water Resour Plan Manag 121:499–508. doi:
10.1061/(ASCE)0733-9496(1995)121:6(499)
[22] Taghi Sattari M, Pal M, Apaydin H, Ozturk F (2013) M5 model tree application in daily river flow forecasting in Sohu Stream,
Turkey. Water Resour 40:233–242. doi: 10.1134/S0097807813030123
[23] Talei A, Chua LHC, Quek C (2010) A novel application of a neuro-fuzzy computational technique in event-based rainfall-
runoff modeling. Expert Syst Appl 37:7456–7468. doi: 10.1016/j.eswa.2010.04.015
[24] Taormina R, Chau KW, Sethi R (2012) Artificial neural network simulation of hourly groundwater levels in a coastal aquifer
system of the Venice lagoon. Eng Appl Artif Intell 25:1670–1676. doi: 10.1016/j.engappai.2012.02.009
[25] Thirumalaiah K, Deo MC (1998) River Stage Forecasting Using Artificial Neural Networks. J Hydrol Eng 3:26–32
[26] Venkata Rao R (2016) Review of applications of TLBO algorithm and a tutorial for beginners to solve the unconstrained and
constrained optimization problems. Decis Sci Lett 5:1–30. doi: 10.5267/j.dsl.2015.9.003
[27] Wu CL, Chau KW, Li YS (2008) River stage prediction based on a distributed support vector regression. J Hydrol 358:96–
111. doi: 10.1016/j.jhydrol.2008.05.028
[28] Wu S-J, Lien H-C, Chang C-H (2012) Calibration of a conceptual rainfall–runoff model using a genetic algorithm integrated
with runoff estimation sensitivity to parameters. J Hydroinformatics 14:497. doi: 10.2166/hydro.2011.010
[29] Yazdani MR, Zolfaghari AA (2017) Monthly River Forecasting Using Instance-Based Learning Methods and Climatic
Parameters. J Hydrol Eng 22:04017002. doi: 10.1061/(ASCE)HE.1943-5584.0001490
[30] Yu PS, Chen ST, Chang IF (2006) Support vector regression for real-time flood stage forecasting. J. Hydrol. 328:704–716

All rights reserved by www.grdjournals.com 360

You might also like