Beruflich Dokumente
Kultur Dokumente
YaoDeng Chen
The atmospheric science college,
Nanjing information science & technology university Nanjing 210044, China keyu@nuist.edu.cn
and approaches have been developed for inverse transient calibration. Gradient methods like eh Levenberg-Marquardt method were initially most widely used. The main advantage of the gradient type optimization is their computational efficiency. However, several researchers noted that gradient type methods may fail to converge, due to instability problems, or may converge to a local rather than global optimum. Recently, evolutionary and meta-heuristic algorithms as search and optimization tools have been extensively used in various problem domains. These procedures have been built upon the deterministic approach of automatic batch calibration of model parameters as a successful global optimization strategy. Genetic algorithm has been developed as a powerful evolutionary algorithm in various domains including water resource management. Ant colony optimization algorithm, a meta-heuristic optimization algorithm initially proposed by Dorigo[1], is a successful swarm-based search and optimization approach inspired by the behavior of real ants in finding the shortest way to food source. A new evolutionary approach to optimization is artificial fish-swarm algorithm which was first proposed in 2002[2], in which it explores the search space rather like following the movement of a flock of fish out searching for food. It has been successfully applied to nonlinear function optimization[3], three-floor feed-forward neural networks training[4,5], parameter tuning of robust PID controller[6], identification on line of time-variant system, etc and has been proved to
-1-
I. programming, mathematical
non-separable objective function and non-linear constraints. Many different optimization models Supported by Open Function of Huaihe river area (HRM200903) Nature Science Basic Research Foundation of Jiangsu Universities08KJB170007 Open Function of Institute of Heavy Rain,CMA,Wuhan (IHR2009G07)
4415
have many advantages, such as good global stringency, strong robustness, insensitive to initial values and parameters, simplicity of implementation and so on, This paper presents AFSA with an irrigation zone water optimization allocation problem, containing most of the complexity available in the nature of optimization, to demonstrate the efficiency non-convex problems. II. PROBLEM DESCRIPTION of the water algorithm resource in handling management
area,hm2 available
cs (t, j,n) is the canal irrigation water pumping cost,yuan/m3 s (t, j, n) is the available
factor of canal irrigation,[-]; irrigation ds (t, j, n) is the allocated surface water, m3;
d g (t , j , n )
3
is
the
allocated
groundwater,m
C tj is the agriculture
The optimization model includes both the objective function and the constraint conditions. The objective function is economic objective, is approached by maximizing the economic benefit of the water supply. The constraint conditions include the water supply constraints, the user demand capacity constraints, and the regional harmonious development constrain et al. The objective function is as follows:
investment coefficient,[-]; Yj is the yield of crop with deficient irrigation , kg/m2Yjm is the maximum yield with full irrigation,kg/m2 ETjn is the water consumption with deficient irrigation, mm ETmjn is the water consumption with full irrigation, mmNj is the number of crop growing stage,[-]; jn is the water sensitivity index,[-]. The constrain conditions are as follows: 1) The total water supply constrain:
t =1
Q1 (t ) Q
(3)
max F = Fm Fg
Qsk (t ) + Q gk (t ) Q1 (t )
(4)
Fm =
(C
j =1
A j f j Y jm )
d
j =1 n =1
Nj
(t , j , n) Qsk (t )
(5)
K Qg 2 ( j ) Q ( j ) Fg = c g ( j ) + cs ( j ) s2 g s j =1
+ Y j Ct j A j
j =1
d
j =1 n =1
Nj
(t , j , n) Q gk (t )
(6)
(1)
Amin A1 A
(2)
(7)
A( j ) a A
j =1
(8)
is the maximum relative yield,[-]; F is the objective, Yuan; t is the sub-zonej is the crop; n is the growing stage Cj is the unit price of crop, Yuan/kg A j is the crop
-2-
4416
available groundwater supply in planning stage,m3 Qs1 (t ) is the total available surface water supply in planning stage,m3 Q g 1 (t ) is the total available groundwater supply in planning stage ,m3 s(t, j,n) is the available soil
d g (t , j , n)
t =1 j =1
(9)
0 d s (t , j , n) Qs (t , j )
n =1 Nj
Nj
(10)
0 d g (t , j , n) Q g (t , j )
n =1
j, n) is the average water ratio in the planning wetting soil zone,[-]; k(t, j,n) is the leakage,m3/hm2 q(t, j, n) is availability ( rainfall,m3/hm2 ck t, j,n) is groundwater
recharge,m3/hm2 III. ARTIFICIAL FISH SWARM ALGORITHM
(11)
0 Qs (t , j ) Qs1 (t )
0 Q g (t , j ) Q g 1 (t )
(12)
AFSA is a stochastic and effective global optimization algorithm which mainly simulates fish schooling the behavior of prey, swarm and follow. AFSA is an optimizer based on population. The system is initialized firstly in a set of randomly generated potential solutions and then performs the search for the optimum one iteratively. Information about good solutions spreads through the swarm, and thus each artificial fish (AF) tend to move to good areas in the search space. The state of each AF is a potential result. Therefore the AFs fitness can be calculated by putting its state into a designated objective function. During the search process each AF successively adjusts its state toward the good areas according to the behavior of prey, swarm and follow, and finally gains the global optimum. Before constructing the AF model, some definitions are defined first. The objective function can be expressed as under:
(13)
0 s (t , j , n) s (t , j , n) max
(14)
(t , j , n) (t , j , n) (t , j , n) f
soil zone
(15)
Q is the total water ,m3; Q1 (t ) is the total water which can be allocated,m3 Qsk (t )
is the available surface water supply,m3 Q gk (t ) is the available groundwater supply,m3
F = F(X )
for the optimum solution.
(17)
Amin is the planting area of food crop to meet the demand of the irrigation area,hm2 A1 is the
food crop area,hm2 A is the total planting area,hm2 a is cropping index r ( n ) is the groundwater recharge from rainfall, m3 p ( n) is the groundwater recharge from river and
3
X m = ( d s (t , j , n )m , d g (t , j , n )m )
follows:
(18)
lake, m Qs (t,
4417
For each set of decision variables, the integrated simulation models were run to calculate the objective function, the constraints. A subset of the nonzero elements of Artificial
p represents the try number of the behavior of searching food. Visual represent the visual distance of the AF. The position of the fish with the highest food concentration will be the optimal position. IV. CASE STUDY
present. The behavior description of the AF is as follows: The behavior of searching food: assume that
Weihai city in China is experiencing severe water shortage and seawater intrusion. The model is used in water saving irrigation region in Weihai city, which is 672hm2, and belongs to the temperate zone with a half moist monsoon continental climate distinction seasons. It is dry the annual rainfall. In years with 50%, 75% rainfall frequency respectively the surface water is 96*104m3 and 83*104m3, the groundwater is 49*104m3 and 35*104m3. The main food crops are winter wheat and summer maize, and the economy crops are pear, apple, cherry, and vegetable and so on. Sprinkler irrigation is the main irrigation method. Surface water and groundwater are the water source. The control parameters of the algorithm are as
4
F ( q ) < F ( m)
follows:
3
group
size
120,
step
is
0.01*10 m . The AFSA was developed in Visual Basic. The water optimization was soled in current year. The optimization results are shown in table 1. In the current Year (2002) the maximum surface and ground water supply respectively was 94.19*104m3, 48 *104m3 and the production
from
behavior of searching food. If n f = 0 , the AF executes the behavior of searching food. The behavior of follow: Assume that
X m is
value is 1313*104 Yuan. After optimization the production value was 2048*104 Yuan, and the net benefit improved 56%.
If
Fq nf
> F (m )
-4-
4418
Table 1 Initial optimal water allocation monthly for crops obtained from optimization104m3
item 1 total Winter wheat Surface water Ground water total Summer maize Surface water Ground water total Surface fruit water Ground water total Surface vegetable water Ground water total Surface other water Ground water 0.00 0.00 2 0.00 0.00 3 0.00 0.00 4 5.39 5.39 5 5.39 0.00 6 0.00 0.00
month 7 0.00 0.00 8 0.00 0.00 9 0.00 0.00 10 0.00 0.00 11 0.00 0.00 12 5.39 5.39
1.01
1.01
1.01
1.01
1.01
1.01 [1]
1.01
1.01
1.01
1.01
1.01
1.01
12.1
V.
CONCLUSION
Water optimization allocation in irrigation is highly non-linear, non-convex, real-world. The applicability and performance of
[2]
artificial
[3]
[4]
[5]
[6]
Dorigo, M. Optimization, learning and natural algorithm,PhD thesis, Politecnico di Milano,Milan.1992. L.X.Li, Z.J.Shao and J.X.Qian, An optimizing method based on autonomous animals: fish-swarm algorithm, Systems Engineering theory&practice, vol.22, no.11, pp.32-38,2002 X.L.Li, Z.J.Shao and J.X.Qian, An optimizing method based on autonomous animals: fish-swarm algorithm, Systems Engineering Theory& Practice, vol.22, no.11, pp. 32-38, 2002. J.W.Ma, G.L.Zhang, H.Xie, C.L.Zhou and J.Wang, Optimization of feed forward neural networks based on artificial fish-swarm algorithm,Computer Applications, vol.24, no.10, pp.21-23, 2004. J.D.Tang, X.Y.Xiong, Y.W.Wu and X.J.Jiang, Reactive power optimization of power system based on artificial fish-swarm algorithm,Relay, vol.32, no.19, pp.9-12, 2004. X.L.Li, Y.C.Xue, F.Lu and G.H.Tian, Parameter estimation method based on artificial fish school algorithm, Journal of Shangdong University Engineering Science, vol.34, no.3, pp.84-87, 2004.
4419