Novel global harmony search algorithm for unconstrained problems

Novel global harmony search algorithm for unconstrained problems
Novel global harmony search algorithm for unconstrained problems

Letters

Novel global harmony search algorithm for unconstrained problems

Dexuan Zou a,?,Liqun Gao a,Jianhua Wu a,Steven Li b

a School of Information Science and Engineering,Northeastern University,Shenyang,Liaoning110004,PR China

b Division of Business University of South Australia GPO Box2471,Adelaide,SA5001,Australia

a r t i c l e i n f o

Article history:

Received3January2010

Received in revised form

30July2010

Accepted31July2010

Communicated by Y.Liu

Available online27August2010

Keywords:

Novel global harmony search algorithm

Position updating

Genetic mutation

Convergence

Stability

Harmony search algorithm

a b s t r a c t

In this paper,we use a recently proposed algorithm—novel global harmony search(NGHS)algorithm to

solve unconstrained problems.The NGHS algorithm includes two important operations:position

updating and genetic mutation with a low probability.The former can enhance the convergence of the

NGHS,and the latter can effectively prevent the NGHS from being trapped into the local optimum.

Based on a large number of experiments,the NGHS has demonstrated stronger convergence and

stability than original harmony search(HS)algorithm and its two improved algorithms(IHS and SGHS).

&2010Elsevier B.V.All rights reserved.

1.Introduction

Due to the computational drawbacks of existing numerical

methods,researchers have to rely on metaheuristic algorithms

based on simulations to solve some complex optimization

problems.A common feature in meta-heuristic algorithms is that

they combine rules and randomness to imitate natural phenomena.

These phenomena include the biological evolutionary process(e.g.,

the genetic algorithm(GA)[1]and the differential evolution(DE)

[2]),animal behavior(e.g.,particle swarm optimization(PSO)[3]

and ant colony algorithm(ACA)[4]),and the physical annealing

process(e.g.,simulated annealing(SA)[5]).Over the last decades,

many meta-heuristic algorithms and their improved algorithms

have been successfully applied to various engineering optimization

problems[6–12].They have outperformed conventional numerical

methods on providing better solutions for some dif?cult and

complicated real-world optimization problems.

Geem et al.proposed a new metaheuristic algorithm—harmony

search(HS)[13]in2001,which is inspired by the phenomenon of

musician attuning.The HS algorithm is based on natural musical

performance processes that occur when a musician searches for a

better state of harmony,such as during jazz improvisation.Jazz

improvisation seeks to?nd musically pleasing harmony as

determined by an aesthetic standard,just as the optimization

process seeks to?nd a global solution as determined by an

objective function.Due to its own advantage and potential,the HS

has been successfully applied to many engineering optimization

problems[14–17]during the last decade.Recently,many improved

harmony search algorithms have been proposed,such as:improved

harmony search(IHS)algorithm[18],global-best harmony search

(GHS)algorithm[19],self-adaptive global-best harmony search

(SGHS)algorithm[20],and local-best harmony search algorithm

with dynamic subpopulations(DLHS)[21].In addition,Das et al.

presented a simple mathematical analysis of the explorative search

behavior of the HS[22],and proposed a simple but very useful

modi?cation to the classical HS in light of the mathematical

analysis.

On the other hand,the novel global harmony search(NGHS)

algorithm[23]is a recently proposed algorithm,and it has been

successfully applied into constrained reliability problems.The

NGHS is an improved version of the HS,and it has shown better

performance than the HS on solving constrained reliability

problems.In this paper,we use the NGHS to solve unconstrained

problems so as to profoundly study and analyze the performance

of the NGHS.We also compare the NGHS with the other three

harmony search algorithms on solving unconstrained problems.

Experimental results indicate that the NGHS has stronger

convergence and stability than the HS and its two improved

algorithms[18,20].

This paper is organized as follows.In Section2,a brief

overview of three harmony search algorithms is provided,and

they are the original harmony search(HS)algorithm,the

Contents lists available at ScienceDirect

journal homepage:https://www.360docs.net/doc/966346959.html,/locate/neucom

Neurocomputing

0925-2312/$-see front matter&2010Elsevier B.V.All rights reserved.

doi:10.1016/j.neucom.2010.07.010

?Corresponding author.

E-mail address:zoudexuan@https://www.360docs.net/doc/966346959.html,(D.Zou).

Neurocomputing73(2010)3308–3318

improved harmony search(IHS)algorithm,and the self-adaptive global-best harmony search(SGHS)algorithm,respectively.In Section3,a novel global harmony search(NGHS)is introduced, and the procedure of the NGHS is described adequately.In Section 4,a large number of experiments are carried out to test and compare the performance of the NGHS and the other three algorithms.We end this paper with some conclusions and comments for further research in Section5.

2.HS,IHS and SGHS algorithms

2.1.The HS algorithm

In this section,we will introduce three harmony search algorithms.The?rst is the original harmony search(HS) algorithm,and it works as follows:

Step1:Initialize the HS algorithm parameters.

The parameters are as follows:the harmony memory size (HMS),or the number of solution vectors in the harmony memory;harmony memory considering rate(HMCR);bandwidth (bw);pitch adjusting rate(PAR);and the number of improvisa-tions(K),or stopping criterion.

Step2:Initialize the harmony memory.

The initial harmony memory(HM)is generated from a uniform distribution in the ranges[x iL,x iU]ei?1,2,...,NT,as

HM?

x1

1

x1

2

áááx1

N

x2

1

x2

2

áááx2

N

^^&^

x HMS

1

x HMS

2

áááx HMS

N

B B

B B

@

1

C C

C C

A

e1T

Step3:Improvise a new harmony.

Generating a new harmony is called improvisation.The new harmony vector x u?ex u1,x u2,...,x u NTis determined by three rules: memory consideration,pitch adjustment and random selection. The procedure works as follows:

for each i A?1,N do

if randeTr HMCR then

x u i?x j

i

ej?1,2,...,HMST%memory consideration

if randeTr PAR then

x u i?x u i7r?bw%pitch adjustment

if x u i4x iU

x u i?x iU

elseif x u i o x iL

x u i?x iL

end

end

else

x u i?x iLtrandeT?ex iUàx iLT%random selection end

end

x u iei?1,2,...,NTis the i th component of x u,and x i j ej?1,2,...,HMSTis the i th component of the j th candidate solution vector in HM.Both r and rand()are uniformly generated random numbers in the region of[0,1],and bw is an arbitrary distance bandwidth.

Step4:Update harmony memory.

If the?tness of the improvised harmony vector x u?ex u1,x u2,...,x u NTis better than that of the worst harmony,replace the worst harmony in the HM with x u.

Step5:Check the stopping criterion.

If the maximal iteration number(K)is satis?ed,computation is terminated.Otherwise,Steps3and4are repeated.

The most important step of the HS algorithm is Step3,and it includes memory consideration,pitch adjustment and random selection.PAR and bw have profound effects on the performance of the HS.Specially,suitable PAR and bw values can not only diversify harmony memory(HM),but also improve the quality of harmonies in HM.

2.2.The IHS algorithm

Mahdavi et al.[18]proposed a new variant of the HS,called the improved harmony search(IHS).The IHS dynamically updates PAR and bw according to

PARekT?PAR mint

PAR maxàPAR min

K

ke2TbwekT?bw max exp

ln bw min

bw max

k

@

1

Ae3T

where K is the maximum number of iterations,and k is the current number of iterations;PAR min and PAR max are the minimum adjusting rate and the maximum adjusting rate,respectively; bw min and bw max are the minimum bandwidth and the maximum bandwidth,respectively.A large number of experiments and studies show that the IHS based on improved PAR and bw has better optimization performance than the HS in most cases.

2.3.The SGHS algorithm

Pan et al.[20]proposed a self-adaptive global-best harmony search(SGHS)algorithm,which is inspired by the GHS algorithm [19].The best harmony vector is utilized to produce a new vector, and a dynamically changing BW is introduced to well balance the exploration and exploitation of the SGHS algorithm.In addition, HCMR and PAR are dynamically adapted to a suitable range by recording their historic values corresponding to generated harmonies entering the HM.Specially,The HMCR(PAR)value is normally distributed in the range of[0.9,1.0]([0.0,1.0])with mean HMCRm(PARm)and standard deviation0.01(0.05). Initially,HMCRm(PARm)is set at0.98(0.9).And then SGHS starts with a HMCR(PAR)value generated according to the normal distribution.During the evolution,the HMCR(PAR)value associated with the generated harmony successfully replacing the worst member in the HM is recorded.After a speci?ed number of generations LP(i.e.,the learning period which is100in their experiment),HMCRm(PARm)is recalculated by averaging all the recorded HMCR(PAR)values during this period.With the new mean and the given standard deviation of0.01(0.05),new HMCR (PAR)value is produced and used in the subsequent iterations. The above procedure is repeated.As a result,an appropriate HMCR(PAR)value can be gradually learned to suit the particular problem and the particular phases of the search process.

3.The novel global harmony search(NGHS)algorithm

A feature of PSO is that the individual is inclined to mimic its successful companion.Inspired by the swarm intelligence of particle swarm,a new variation of HS is proposed in this paper. The new approach,called NGHS,modi?es the improvisation step of the HS such that the new harmony can mimic the global-best harmony in the HM.The NGHS and the HS are different in three aspects as follows:

(1)In Step1,harmony memory considering rate(HMCR)and pitch adjusting rate(PAR)are excluded from the NGHS,and genetic mutation probability(p m)is included in the NGHS;

D.Zou et al./Neurocomputing73(2010)3308–33183309

(2)In Step 3,the NGHS modi?es improvisation step of the HS,and it works as follows:

for each i A ?1,N do x R ?2?x best i àx worst i if x R 4x iU x R ?x iU

elseif x R o x iL x R ?x iL end

x u i ?x worst i tr ?ex R àx worst i T%position updating if rand eTr p m then

x u i ?x iL trand eT?ex iU àiL T%genetic mutation end end

Here,‘‘best’’and ‘‘worst’’are the indexes of the best harmony

and the worst harmony in HM,respectively.r and rand ()are all uniformly generated random number in [0,1].

Fig.1is used to illustrate the principle of position updating:step i ?j x best i àx worst i

j is de?ned as adaptive step of the i th decision variable.The region between P and R is de?ned as trust region for the i th decision variable.The trust region is actually a region near the global-best harmony.A reasonable explanation is as follows:In the early stage of optimization,all solution vectors are sporadic in solution space,so most adaptive steps are large and most trust regions are wide,which is bene?cial to the global search of the NGHS;while in the late stage of optimization,all non-best solution vectors are inclined to move to the global-best solution vector,so most solution vectors are close to each other.In this case,most adaptive steps are small and most trust regions are narrow,which is bene?cial to the local search of the NGHS;the reasonable design for step i can guarantee that the proposed algorithm has strong global search ability in the early stage of optimization,and has strong local search ability in the late stage of optimization.Dynamically adjusted step i keeps a balance between the global search and the local search.

Genetic mutation operation with a small probability is carried out for the worst harmony of harmony memory after updating position,for it can enhance the capacity of escaping from the local optimum for the proposed algorithm.

(3)In Step 4,the NGHS replaces the worst harmony x worst in HM with the new harmony x u even if x u is worse than x worst .

Now,we will illustrate that the modi?cations introduced in this paper are compatible with the HS framework.Detailed illustrations are as follows:

(1)The HS randomly selects harmony components from the harmony memory (HM)with the probability HMCR.On the other hand,the NGHS selects the best harmony components from the HM with probability 100%.Both algorithms aim at using existing good harmonies to improvise a new harmony.

(2)For the HS,pitch adjustment will be carried out if both HMCR and PAR are satis?ed,and the adjusting step bw is de?ned as distance bandwidth.For the NGHS,location updating is carried out with probability 100%,and adjusting step step i ei ?1,2,...,N Tis de?ned as adaptive step.Similarly,both bw and step i are

utilized to determine the promising search ranges of their corresponding algorithms.

(3)The HS carries out random selection with the probability 1-HMCR,and the NGHS carries out genetic mutation with the probability p m .In fact,both operations are exactly the same,and they are used to diversify the harmony memory,which can effectively prevent both algorithms from being trapped into the local optimum.

4.Experimental results and analysis

Sixteen test functions are considered to extensively investigate the performance of the NGHS algorithm,and they are presented as follows:

The ?rst is Sphere function,de?ned as min f 1?

X N i ?1

x 2i

e4T

where global optimum x *?(0,0,y ,0)and f (x *)?0for à100r x i r 100.

The second is Rosenbrock function,de?ned as min f 2?

X N à1i ?1

e100ex i t1àx 2i T2tex i à1T2

T

e5T

where global optimum x *?(1,1,y ,1)and f (x *)?0for à100r

x i r 100.

The third is generalized Rastrigrin function,de?ned as min f 3?

X N i ?1

ex 2i à10cos e2p x i Tt10T

e6T

where global optimum x *?(0,0,y ,0)and f (x *)?0for à10r

x i r 10.

The fourth is generalized Griewank function,de?ned as

min f 4?14000X N i ?1x 2

i àY N i ?1cos x i ?i p t1

e7Twhere global optimum x *?(0,0,y ,0)and f (x *)?0for à600r

x i r 600.

The ?fth is Ackley’s function,de?ned as

min f 5?20te à20exp à0:2

???????????????????P N i ?1x 2i N

s 0@1A àexp P N i ?1

cos e2p x i TN !e8T

where global optimum x *?(0,0,y ,0)and f (x *)?0for à32r

x i r 32.

The sixth is Schwefel’s problem 2.22,de?ned as min f 6?

X N i ?1

j x i jt

Y N i ?1

j x i j

e9T

where global optimum x *?(0,0,y ,0)and f (x *)?0for à100r

x i r 100.

The seventh is Schwefel’s problem 2.26,de?ned as min f 7?418:9829N à

X N i ?1

ex i sin e

???????

j x i j p TT

e10T

where global optimum x *?(420.9687,420.9687,y ,420.9687)and f (x *)?0for à500r x i r 500.

The eighth is rotated hyper-ellipsoid function,de?ned as

min f 8?

X N i ?1

X i j ?1

x j 0@1A 2e11

TFig.1.The schematic diagram of position updating.

D.Zou et al./Neurocomputing 73(2010)3308–3318

3310

where global optimum x*?(0,0,y,0)and f(x*)?0forà100r x i r100.

The ninth[24]is as follows:

min f9?

X N

i?1z2

i

à450e12T

where z?xào;o?[o1,o2,y,o n]is the shifted global optimum; global optimum x*?o and f(x*)?à450forà100r x i r100.

The10th[24]is as follows:

min f10?

X N

i?1

X i

j?1

z j

@

1

A

2

à450e13T

where z?xào;o?[o1,o2,y,o N]is the shifted global optimum; global optimum x*?o and f(x*)?à450forà100r x i r100.

The11th[24]is as follows:

min f11?

X N

i?1e106Teià1T=eNà1Tz2

i

à450e14T

where z?exàoT?M,and o?[o1,o2,y,o N]is the shifted global optimum,M is orthogonal matrix;global optimum x*?o and f(x*)?à450forà100r x i r100.

The12th[24]is as follows:

min f12?

X N

i?1

X i

j?1

z j

@

1

A

2

e1t0:4?j Ne0,1TjTà450e15T

where z?xào;o?[o1,o2,y,o N]is the shifted global optimum; global optimum x*?o and f(x*)?à450forà100r x i r100.

The13th[24]is as follows:

min f13?X

Nà1

i?1

e100ez it1àz2

i

T2tez ià1T2Tt390e16T

where z?xào+1;o?[o1,o2,y,o N]is the shifted global optimum; global optimum x*?o and f(x*)?390forà100r x i r100.

The14th[24]is as follows:

min f14?

X N

i?1z2

Y N

i?1

cos

z i

?

i

p

t1à180e17T

where z?exàoT?M,and o?[o1,o2,y,o N]is the shifted global optimum,M is linear transformation matrix;global optimum x*?o and f(x*)?à180forà600r x i r600.

The15th[24]is as follows:

min f15?20teà20expà0:2??????????????????P N

i?1

z2

i

s 0

@1 A

àexp P N

i?1

cose2p z iT

!

à140e18T

where z?exàoT?M,and o?[o1,o2,y,o N]is the shifted global optimum,specially,o2jà1?à32,and o2j are randomly distributed in the search range,for j?1,2,...b N=2c.M is linear transformation matrix;global optimum x*?o and f(x*)?à140forà32r x i r32.

The16th[24]is as follows:

min f16?

X N

i?1ez2

i

à10cose2p z iTt10Tà330e19T

where z?xào;o?[o1,o2,y,o N]is the shifted global optimum; global optimum x*?o and f(x*)?à330forà5r x i https://www.360docs.net/doc/966346959.html,parison between NGHS,HS,IHS and SGHS algorithms

We experimentally compare the NGHS with the other three harmony search algorithms on a large number of problem instances(f1–f16).In the experiments,the settings of the compared HS algorithms are as follows:For the HS algorithm, HMS?5,HMCR?0.9,PAR?0.3and bw?0.01;For the IHS algorithm,HMS?5,HMCR?0.9,PAR min?0.01,PAR max?0.99, bw min?0.0001,and bw max?(x Uàx L)/20;For the SGHS algorithm, HMS?5,HMCRm?0.98,PARm?0.9,bw min?0.0005,bw max?(x Uàx L)/10,and LP?100;For the NGHS,HMS?5,p m?0.005.

According to Sections2.2and2.3,both the IHS and the SGHS dynamically adjust their algorithm parameters throughout the whole iterations,which adds additional computing time.On the other hand,the NGHS uses position updating to replace the harmony memory considering and pitch adjusting of the HS,and its mutation operation is the same as the random selection of the HS.It seems that the HS and the NGHS have similar computational complexity. To provide a more fair comparison,we increase the maximal iteration numbers of both the HS and the NGHS when the maximal iteration numbers of both the IHS and the SGHS are?xed.For each function,three different dimension sizes are tested,and they are N?10,30and50,respectively.Thirty independent experiments are carried out in each case,and the optimization results using four harmony search algorithms are reported in Tables1–3.

Here,‘‘K’’is the maximal iteration number,‘‘AOT’’represents average optimization time,and‘‘SD’’represents standard devia-tion.From Table1,we can see that the NGHS always uses the least average optimization time to solve any test function even if its maximal iteration number is larger than those of the IHS and the SGHS.The NGHS can?nd the best objective function values for all test functions except f14and f15.The IHS found the best objective function values for f14and f15,and the NGHS results are comparable to both best IHS results.

From Table2,it can be observed that the average optimization times using the NGHS are always less than those of the other three harmony search algorithms for all test functions except f12, and the AOT of the NGHS is slightly larger than that of the SGHS for f12.According to four criterions(Best,Worst,Mean and SD), the NGHS outperforms the other three harmony search algo-rithms for all test functions except f2and f15.More specially,the Mean and SD obtained by the NGHS are better than those of the other three approaches for most test functions,which indicates the strong convergence and stability of the NGHS.Based on the above simulation analysis,the NGHS can generate better solutions with smaller computational requirements compared to the other three approaches in most cases.

It can be seen from Table3,that NGHS performs better than the other three approaches in terms of solution accuracy on the utilized16test functions.Specially,the NGHS algorithm gen-erates15best results out of16functions,and the worst results using the NGHS are even better than the best results using three harmony search algorithms for f1,f5,f6,f7,f8,f10,f12,f14and f16.So, the HS,the IHS and the SGHS cannot show better results than NGHS even on one run for any of these test functions.

According to the above observations,we concluded that the performance of the NGHS is superior to the other three methods when used to solve most of these unconstrained optimization problems.More experiments,especially with larger problems may, however,be needed in order to con?rm such good performance.

4.2.Effects of p m and HMS

In this section,all the above16test functions are used to investigate the effects of p m and HMS on the performance of the

D.Zou et al./Neurocomputing73(2010)3308–33183311

Table 1

The optimization results of four harmony search algorithms for f 1àf 16(N ?10).Function Algorithm K AOT

Best

Worst

Mean

SD

f 1

HS 30000 5.7660e à001 2.9129e à008 6.1045e à007 2.0999e à007 1.2421e à007IHS 10000 6.3643e à001 6.1585e à009 3.3831e à008 2.1536e à008 6.4556e à009SGHS 100008.7347e à001 3.1385e à009 5.1117e à0028.6474e à003 1.1695e à002NGHS 20000 3.6770e à001 3.9296e à041 2.9471e à031 2.1333e à032 6.9650e à032f 2

HS 30000 5.8433e à001 1.6436e à002 4.9914e+003 2.3417e+0029.0439e+002IHS 10000 6.3910e à001 6.5349e+0009.4594e+003 1.5642e+003 2.8488e+003SGHS 100008.8440e à001 5.0470e à001 5.1306e+003 2.5233e+0029.2618e+002NGHS 20000 3.7760e à001 5.3463e à004 5.4967e+003 4.9368e+002 1.3524e+003f 3

HS 30000 6.3130e à001 1.1661e à005 1.3006e à004 4.5934e à005 2.5863e à005IHS 10000 6.6923e à001 2.4508e à006 2.9849e+0009.7833e à0018.8010e à001SGHS 100008.9847e à001 5.4866e à008 2.3348e+000 5.2420e à0017.9395e à001NGHS 20000 4.0100e à0010

9.9496e à001 3.3165e à002 1.8165e à001f 4

HS 30000 6.8443e à001 2.6485e à007 1.4754e à001 6.0457e à002 3.6097e à002IHS 10000 6.7507e à0019.0619e à003 2.0167e à0017.6168e à002 4.6718e à002SGHS 100009.1817e à001 5.4412e à002 2.6813e à001 1.4351e à001 5.7705e à002NGHS 20000 4.1453e à0010

1.2046e à001 6.8529e à002 3.3204e à002f 5

HS 30000 6.2083e à001 2.6202e à004 1.0237e à003 5.4888e à004 1.7181e à004IHS 10000 6.8070e à0019.7298e à005 1.8063e à001 6.1802e à003 3.2949e à002SGHS 100009.1300e à001 2.8494e à005 5.4666e à002 5.0686e à003 1.1268e à002NGHS 20000 4.1933e à0017.9936e à015 5.0626e à014 2.6468e à0149.4872e à015f 6

HS 30000 6.2397e à001 5.9582e à004 2.1651e à003 1.1033e à003 3.4149e à004IHS 10000 6.6090e à001 2.6277e à004 4.9911e à004 3.7745e à004 6.3608e à005SGHS 100009.1510e à001 2.0507e à002 4.7466e à001 1.4771e à0019.3862e à002NGHS 20000 3.9327e à001 5.4770e à021 2.6652e à017 2.5681e à018 6.4570e à018f 7

HS 300007.8127e à001 1.2731e à0049.8712e à003 6.5087e à004 2.0524e à003IHS 100008.1307e à001 1.2728e à004 1.2728e à004 1.2728e à004 1.0104e à009SGHS 10000 1.0687e+000 3.8367e à003 4.4140e+001 4.3980e+000 1.0872e+001NGHS 20000 5.0827e à001 1.2728e à004 1.1844e+002 3.9481e+000 2.1624e+001f 8

HS 300007.9223e à001 2.9975e+001 3.3575e+002 1.1424e+0027.7109e+001IHS 100008.1667e à001 1.2325e+001 6.8309e+002 2.4584e+002 1.9314e+002SGHS 10000 1.0832e+000 1.1472e+000 2.2166e+002 4.4718e+001 5.0450e+001NGHS 20000 5.2767e à001 1.0224e à0019.4449e+000 2.5160e+000 2.3538e+000f 9

HS 30000 6.4173e à001à4.5000e+002à4.5000e+002à4.5000e+002 1.2193e à007IHS 100007.1923e à001à4.5000e+002à4.5000e+002à4.5000e+002 5.7304e à009SGHS 100009.7077e à001à4.5000e+002à4.4993e+002à4.4999e+002 1.6373e à002NGHS 20000 4.5263e à001à450

à450

à450

f 10

HS 300008.6130e à001à4.4180e+002à1.5849e+002à3.3589e+0027.3644e+001IHS 100007.4760e à001à4.4870e+002 5.3709e+002à1.9303e+002 2.3832e+002SGHS 100009.9853e à001à4.4627e+002à2.6309e+002à4.0418e+002 5.1129e+001NGHS 20000 5.7227e à001à4.4983e+002à4.1824e+002à4.4334e+0028.0172e+000f 11

HS 30000 1.0529e+000 3.0837e+005 1.6868e+007 4.6601e+006 4.1172e+006IHS 100008.5843e à001 1.9761e+005 1.3515e+007 3.4793e+006 3.2989e+006SGHS 10000 1.1334e+000 2.1771e+005 1.0386e+007 2.8299e+006 2.2564e+006NGHS 200007.5890e à001 1.8886e+005 4.2554e+006 1.2129e+006 1.0627e+006f 12

HS 300009.2593e à001à4.3759e+002à8.4163e+000à2.9502e+002 1.1333e+002IHS 100008.0577e à001à4.4782e+002 1.0248e+003à4.6045e+001 4.3384e+002SGHS 10000 1.0329e+000à4.4131e+002à3.2378e+002à3.9487e+002 3.2845e+001NGHS 20000 5.7863e à001à4.4984e+002à4.1765e+002à4.4158e+0028.4856e+000f 13

HS 300007.3793e à001 3.9254e+002 1.0272e+004 1.1280e+003 2.2521e+003IHS 100007.0680e à001 3.9014e+0028.2056e+003 1.6553e+003 2.3782e+003SGHS 100009.3850e à001 3.9118e+002 1.1485e+004 3.1457e+003 4.2465e+003NGHS 20000 4.3240e à001 3.9002e+0029.6400e+003 2.0091e+003 3.3979e+003f 14

HS 30000 1.1510e+000à1.7988e+002à1.7716e+002à1.7915e+002 6.5803e à001IHS 100008.8337e à001à1.7994e+002à1.7131e+002à1.7800e+002 2.1163e+000SGHS 10000 1.1942e+000à1.7973e+002à1.7841e+002à1.7913e+002 4.2704e à001NGHS 200007.5427e à001à1.7987e+002à1.7147e+002à1.7639e+002 2.4661e+000f 15

HS 30000 1.1292e+000à1.1996e+002à1.1960e+002à1.1981e+0029.4518e à002IHS 100008.7710e à001à1.1997e+002à1.1936e+002à1.1969e+002 1.7267e à001SGHS 10000 1.1870e+000à1.1993e+002à1.1946e+002à1.1978e+002 1.1715e à001NGHS 200007.2650e à001à1.1996e+002à1.1954e+002à1.1978e+002 1.2150e à001f 16

HS 300007.4210e à001à3.3000e+002à3.3000e+002à3.3000e+002 1.4282e à005IHS 100007.0420e à001à3.3000e+002à3.2901e+002à3.2995e+002 1.8624e à001SGHS 100009.3697e à001à3.3000e+002à3.2501e+002à3.2800e+002 1.3135e+000NGHS

20000

4.7507e à001

à330

à3.2901e+002

à3.2997e+002

1.8165e à001

D.Zou et al./Neurocomputing 73(2010)3308–3318

3312

D.Zou et al./Neurocomputing73(2010)3308–33183313

Table2

The optimization results of four harmony search algorithms for f1–f16(N?30).

Function Algorithm K AOT Best Worst Mean SD

f1HS90000 4.5290e+000 5.9433eà002 3.1769e+000 6.9084eà0017.5714eà001 IHS30000 4.6819e+000 1.6633e+000 1.6300e+0017.1249e+000 3.4260e+000

SGHS30000 3.0687e+000 5.2824eà007 5.5217eà002 1.3529eà002 1.3744eà002

NGHS60000 2.2387e+000 5.7294eà018 3.9233eà015 5.9866eà016 1.0009eà015

f2HS90000 4.4417e+000 1.7814e+002 1.0026e+004 1.0228e+003 1.8587e+003 IHS30000 4.6833e+000 2.4381e+003 3.6220e+004 1.1604e+0047.7731e+003

SGHS30000 3.0953e+000 3.8906e+001 6.2642e+003 4.9478e+002 1.3856e+003

NGHS60000 2.2937e+000 2.1716eà0037.2347e+003 1.1163e+003 2.0391e+003

f3HS90000 4.7698e+000 2.5679eà002 3.0349e+0009.3737eà0019.3018eà001 IHS30000 4.8902e+0007.5656e+000 1.7846e+001 1.1807e+001 2.4043e+000

SGHS30000 3.1735e+000 2.3796eà0067.2831e+000 2.5508e+000 1.5670e+000

NGHS60000 2.5150e+0000 4.1353eà010 1.5694eà0117.5579eà011

f4HS90000 4.9225e+0007.2972eà001 1.0516e+0009.7764eà0017.4106eà002 IHS30000 4.9886e+000 1.0572e+000 1.3285e+000 1.1516e+000 6.4516eà002

SGHS30000 3.2848e+000 4.1045eà002 3.4484eà001 1.6738eà0017.8560eà002

NGHS60000 2.6151e+0000 1.8765eà001 5.5000eà002 4.9350eà002

f5HS90000 4.7823e+0008.4348eà003 6.3452eà001 6.1483eà002 1.3074eà001 IHS30000 4.9312e+000 1.0436eà001 2.6117e+000 1.8408e+000 4.5176eà001

SGHS30000 3.1797e+000 4.5161eà005 1.2398eà002 1.1591eà003 3.2218eà003

NGHS60000 2.4079e+0007.3700eà010 2.3065eà008 5.0506eà009 4.8921eà009

f6HS90000 4.7161e+000 3.1756eà001 3.2593e+000 1.5317e+0007.0225eà001 IHS30000 4.8510e+000 4.1647eà001 6.3598e+000 2.4069e+000 1.4358e+000

SGHS30000 3.1668e+000 1.2858eà0018.7032eà001 4.1507eà001 1.7614eà001

NGHS60000 2.5135e+000 1.3387eà0099.0607eà008 2.0414eà008 2.3302eà008

f7HS90000 5.3157e+000 3.7153e+000 1.9469e+0018.2052e+000 3.5276e+000 IHS30000 5.0390e+000 3.4680e+000 3.6983e+001 1.5504e+0019.4119e+000

SGHS30000 3.3552e+000 1.0536eà001 6.8470e+001 6.4187e+000 1.7870e+001

NGHS60000 2.8839e+000 3.8183eà004 3.8183eà004 3.8183eà004 1.0080eà012

f8HS90000 6.3901e+000 1.4499e+003 4.4411e+003 2.6404e+0038.9778e+002 IHS30000 5.3772e+000 3.1668e+0039.7564e+003 6.4600e+003 1.8053e+003

SGHS30000 3.7547e+000 3.8900e+002 2.0936e+0039.5742e+002 3.9842e+002

NGHS60000 3.6587e+000 1.4414e+001 2.4507e+0028.1775e+001 5.5860e+001

f9HS90000 4.8208e+000à4.4994e+002à4.4779e+002à4.4947e+002 4.6568eà001 IHS30000 5.0301e+000à4.4993e+002à4.3415e+002à4.4308e+002 4.1650e+000

SGHS30000 3.3276e+000à4.5000e+002à4.4993e+002à4.4999e+002 1.6583eà002

NGHS60000 2.4313e+000à450à4.5000e+002à450 5.0623eà014

f10HS900007.2657e+000 1.1187e+003 4.9553e+003 2.4254e+0039.1445e+002 IHS30000 5.9093e+000 4.1374e+003 1.7549e+0047.7504e+003 3.0011e+003

SGHS30000 4.1807e+000 1.7644e+002 2.8867e+0039.7098e+002 6.3131e+002

NGHS60000 4.1526e+000à3.9857e+002à2.9674e+000à3.0693e+0028.5679e+001

f11HS900007.6589e+000 1.3191e+007 5.0627e+007 3.3232e+007 1.0172e+007 IHS30000 6.1760e+000 1.6372e+0078.1516e+007 5.1501e+007 1.8883e+007

SGHS30000 4.7110e+000 5.8176e+006 2.1411e+007 1.2625e+007 4.1470e+006

NGHS60000 4.5360e+000 1.1506e+0068.1823e+006 3.4080e+006 1.5773e+006

f12HS900007.0880e+000 1.3297e+003 6.9905e+003 3.2367e+003 1.1122e+003 IHS30000 5.6652e+000 4.8642e+003 1.7237e+0048.5947e+003 2.8593e+003

SGHS30000 4.0302e+000 3.1185e+002 3.5944e+003 1.3484e+0038.0101e+002

NGHS60000 4.1833e+000à4.1304e+002à7.8426e+001à2.7939e+0028.7641e+001

f13HS90000 4.8709e+000 4.8810e+002 1.1203e+004 2.6713e+003 3.1923e+003 IHS30000 5.2942e+000 3.1273e+003 2.6193e+0049.8223e+003 5.1948e+003

SGHS30000 3.5121e+000 4.3287e+002 1.1863e+004 2.8462e+003 4.0781e+003

NGHS60000 2.7963e+000 3.9047e+002 1.1109e+004 1.6664e+003 2.8307e+003

f14HS900007.4739e+000à1.7867e+002à1.7686e+002à1.7819e+002 4.1038eà001 IHS30000 6.1693e+000à1.7775e+002à1.7065e+002à1.7517e+002 1.6656e+000

SGHS30000 4.6812e+000à1.7928e+002à1.7891e+002à1.7900e+0029.3573eà002

NGHS60000 4.3907e+000à1.8000e+002à1.7992e+002à1.7998e+002 1.7561eà002

f15HS900008.4621e+000à1.1911e+002à1.1893e+002à1.1900e+002 4.4332eà002 IHS300007.6494e+000à1.1907e+002à1.1888e+002à1.1896e+002 4.8804eà002

SGHS30000 5.6287e+000à1.1988e+002à1.1954e+002à1.1972e+0027.8421eà002

NGHS60000 3.6126e+000à1.1987e+002à1.1952e+002à1.1968e+0028.8529eà002

f16HS90000 5.2827e+000à3.2998e+002à3.2897e+002à3.2987e+002 3.0290eà001 IHS30000 5.0381e+000à3.2794e+002à3.2207e+002à3.2544e+002 1.5255e+000

SGHS30000 3.3412e+000à3.2891e+002à3.2478e+002à3.2695e+002 1.2779e+000

NGHS60000 2.8833e+000à330à3.3000e+002à3.3000e+0028.5981eà013

Table 3

The optimization results of four harmony search algorithms for f 1–f 16(N ?50).Function Algorithm K AOT

Best

Worst

Mean

SD

f 1

HS 100000 6.3780e+000 6.0076e+001 2.5205e+002 1.7603e+002 4.4359e+001IHS 50000 1.2900e+001 2.3228e+0027.5158e+002 4.9421e+002 1.1559e+002SGHS 50000 6.1854e+000 1.8998e à006 1.6033e à001 2.8402e à002 3.2096e à002NGHS 80000 4.7459e+000 2.2757e à0107.0950e à009 1.8091e à009 1.9488e à009f 2

HS 100000 6.5327e+0009.6629e+0048.9231e+005 3.3964e+005 1.9909e+005IHS 50000 1.2686e+0017.5693e+005 5.7741e+006 2.4476e+006 1.1118e+006SGHS 50000 6.2088e+000 1.5578e+0029.6556e+003 1.2443e+003 2.7350e+003NGHS 80000 4.7349e+000 2.4458e à0018.1098e+003 5.3162e+002 1.7208e+003f 3

HS 1000007.2161e+000 2.9511e+001 5.2810e+001 4.2476e+001 5.7278e+000IHS 50000 1.3192e+001 5.0142e+001 1.0121e+0027.6374e+001 1.0771e+001SGHS 50000 6.4016e+000 1.0903e à001 1.0875e+001 5.7834e+000 2.2214e+000NGHS 80000 4.9380e+000 3.7993e à0099.9496e à001 3.3166e à002 1.8165e à001f 4

HS 1000008.0198e+000 2.0566e+000 3.2913e+000 2.5946e+000 3.2134e à001IHS 50000 1.3743e+001 3.9497e+0009.6680e+000 5.8018e+000 1.2152e+000SGHS 50000 6.8679e+0007.6282e à002 3.1431e à001 1.8212e à001 6.0839e à002NGHS 80000 5.5714e+000 4.4744e à010 1.3045e à001 3.7959e à002 3.8783e à002f 5

HS 1000007.4250e+000 3.2228e+000 4.4313e+000 3.7857e+000 3.2904e à001IHS 50000 1.3500e+001 4.7196e+000 5.9813e+000 5.2778e+000 2.9940e à001SGHS 50000 6.5453e+000 6.8535e à0059.4342e à0038.1216e à004 2.0654e à003NGHS 80000 4.8640e+000 2.1342e à006 1.7391e à0057.7723e à006 3.4193e à006f 6

HS 1000007.0240e+000 3.1353e+001 5.0629e+001 4.2350e+001 4.9070e+000IHS 50000 1.3060e+001 3.5501e+0018.4063e+001 5.8769e+001 1.0974e+001SGHS 50000 6.4645e+000 4.7367e à001 1.3476e+0009.0272e à001 2.2061e à001NGHS 80000 5.0724e+000 3.6976e à005 2.2687e à0048.1858e à005 3.7660e à005f 7

HS 1000008.4332e+000 2.9427e+002 6.7011e+002 4.1774e+0029.5933e+001IHS 50000 1.3977e+001 2.8093e+002 1.1779e+0037.4093e+002 1.9622e+002SGHS 500007.2183e+000 2.0680e à001 1.5987e+001 1.4906e+000 2.9687e+000NGHS 80000 6.3162e+000 6.3638e à004 6.3681e à004 6.3644e à0048.7793e à008f 8

HS 100000 1.1377e+001 1.2951e+004 3.2445e+004 2.1537e+004 5.1131e+003IHS 50000 1.5177e+001 2.1066e+004 4.4783e+004 2.9112e+004 4.8624e+003SGHS 500008.6996e+000 2.3241e+003 6.2041e+003 3.9058e+0039.2316e+002NGHS 800008.7239e+000 3.0708e+002 1.2443e+003 6.9208e+002 2.1710e+002f 9

HS 100000 6.4484e+000à3.5043e+002à1.6889e+002à2.5642e+002 4.4840e+001IHS 50000 1.2916e+001à1.5674e+002 4.8751e+0029.7197e+001 1.4285e+002SGHS 50000 6.0876e+000à4.5000e+002à4.4989e+002à4.4998e+002 2.3295e à002NGHS 80000 4.5729e+000à4.5000e+002à4.5000e+002à4.5000e+002 1.2822e à009f 10

HS 100000 1.2535e+001 1.6278e+004 3.5761e+004 2.3503e+004 5.1672e+003IHS 50000 1.6052e+001 1.8107e+004 5.3281e+004 3.3209e+0047.0436e+003SGHS 500009.3468e+000 2.6940e+003 1.0618e+004 5.3262e+003 1.8789e+003NGHS 800009.7527e+000 3.9438e+002 2.3064e+0039.2796e+002 4.4507e+002f 11

HS 100000 1.1155e+001 6.0448e+007 1.8797e+008 1.2265e+008 3.2867e+007IHS 50000 1.5359e+0017.9658e+007 2.3900e+008 1.4841e+008 4.0123e+007SGHS 500009.7390e+0008.0843e+006 3.7497e+007 1.8978e+007 6.1005e+006NGHS 800008.6995e+000 3.8070e+006 1.4028e+007 6.6177e+006 1.9688e+006f 12

HS 100000 1.2926e+001 1.5686e+004 4.0686e+004 2.5813e+004 6.0171e+003IHS 50000 1.6085e+001 2.1726e+004 5.4777e+004 3.8369e+0047.7422e+003SGHS 500009.5588e+000 3.0714e+003 1.3240e+004 6.4690e+003 2.0085e+003NGHS 80000 1.0260e+001 3.1526e+002 2.6327e+003 1.0597e+003 6.0928e+002f 13

HS 100000 6.6739e+000 1.7211e+005 1.8291e+006 4.0277e+005 3.0845e+005IHS 50000 1.3041e+001 1.2263e+006 4.5379e+006 2.8450e+0068.9783e+005SGHS 50000 6.3203e+000 4.4911e+002 1.6233e+004 3.2408e+003 5.0484e+003NGHS 80000 4.8637e+000 3.9004e+002 1.1862e+004 2.0039e+003 3.3640e+003f 14

HS 1000009.8458e+000à1.6607e+002à1.4317e+002à1.5772e+002 6.2514e+000IHS 50000 1.4928e+001à1.5332e+002à8.2651e+001à1.2930e+002 1.6724e+001SGHS 500009.0807e+000à1.7896e+002à1.7880e+002à1.7891e+002 3.8661e à002NGHS 800007.3657e+000à1.7998e+002à1.7990e+002à1.7994e+002 2.1517e à002f 15

HS 1000009.1332e+000à1.1891e+002à1.1874e+002à1.1883e+002 3.7003e à002IHS 50000 1.4616e+001à1.1885e+002à1.1873e+002à1.1878e+002 2.8639e à002SGHS 500008.9495e+000à1.1984e+002à1.1966e+002à1.1975e+002 5.0678e à002NGHS 80000 6.9954e+000à1.1976e+002à1.1946e+002à1.1961e+0027.0579e à002f 16

HS 1000007.1546e+000à3.1671e+002à3.0395e+002à3.1038e+002 2.5907e+000IHS 50000 1.3261e+001à2.9968e+002à2.7562e+002à2.8991e+002 5.2346e+000SGHS 50000 6.4401e+000à3.2997e+002à3.2266e+002à3.2592e+002 1.8642e+000NGHS

80000

4.9802e+000

à3.3000e+002

à3.3000e+002

à3.3000e+002

4.0850e à005

D.Zou et al./Neurocomputing 73(2010)3308–3318

3314

NGHS.The dimension sizes of all functions are set to30,and the maximal iteration number is set to30000accordingly.Thirty independent runs are carried out in each case,and the mean and standard deviation(7SD)obtained using different p m and HMS values are reported in Tables4and5.

As can be seen from Table4,that a large p m(Such as0.05and 0.1)or a small p m(Such as0.0005and0.0001)is harmful to the convergence of the NGHS in most cases,and will deteriorate the performance of the NGHS.In contrast,using p m?0.005(or0.01) can improve the convergence and stability of the NGHS signi?cantly on solving most test functions.In short,there is no single choice for p m,and it should be adjusted according to practical optimization problems.

According to Table5,there is no indication that one setting of HMS is superior to the other settings.Generally speaking,small HMS is reasonable and logical,for it is similar to the short memory of musician.

4.3.Mann–Whitney U test

In this section,Mann–Whitney U test is employed to con?rm that a statistical signi?cant difference in performance is obtained by the NGHS algorithm when compared to the other three algorithms.The Mann–Whitney U test(also called as Mann–Whitney Wilcoxon test)[25,26]is a non-parametric rank-based test for identifying a difference between populations.The test statistic U is given by

U1?R1à

n1en1t1T

e20TU2?R2à

n2en2t1T

2

e21Twhere n1and n2are the sizes of sample1and sample2, respectively;R1and R2are the sums of the ranks in sample1and sample2,respectively.U1is the number of sample1observations beaten by sample2observations,and U2is the number of sample 2observations beaten by sample1observations.The smaller value of U1and U2is the one used when consulting signi?cance tables.The sum of the two values is given by

U1tU2?R1à

n1en1t1T

2

tR2à

n2en2t1T

2

e22TKnowing that R1+R2?n(n+1)/2and n?n1+n2,and doing some algebra,we?nd that the sum is

U1tU2?n1n2e23T

Before using Mann–Whitney U test,it is necessary to set a rule to determine who is the winner in a comparison.Suppose x i(i?1,y,n1)is a solution obtained by A algorithm for a problem, and y j(j?1,y,n2)is a solution obtained by B algorithm for the same problem.If x i(i?1,y,n1)is better than y j(j?1,y,n2),then A algorithm is the winner,otherwise,B algorithm is the winner.If

Table4

The effect of p m on the performance of the NGHS(HMS?5).

Function p m?0.0005p m?0.001p m?0.005p m?0.01p m?0.05p m?0.1

f1 1.1672eà012 4.9313eà021 3.4411eà0129.6784eà007 2.6428e+0017.9350e+002

(6.3851eà012)(1.9566eà020)(1.0646eà011)(9.0368eà007)(8.6339e+000)(1.5884e+002)

f2 1.6314e+003 1.3591e+0039.5434e+0029.2395e+002 1.4643e+004 6.1203e+006

(3.1221e+003)(2.9679e+003)(2.3201e+003)(2.2704e+003)(6.2798e+003)(2.9657e+006)

f3 1.1377e+001 3.6987e+000 1.9908eà001 2.3350eà001 5.0222e+001 1.5816e+002

(4.1109e+000)(2.1059e+000)(4.0474eà001)(4.2735eà001)(8.3951e+000)(2.0305e+001)

f4 1.7727eà001 1.2763eà001 6.4404eà002 3.6069eà002 1.2331e+0008.2507e+000

(1.7375eà001)(9.1172eà002)(4.1008eà002)(3.2762eà002)(7.2446eà002)(2.0658e+000)

f57.8514eà001 6.6771eà007 5.6981eà007 2.2327eà004 2.6356e+0007.4782e+000

(6.8145eà001)(3.1800eà006)(4.3733eà007)(9.2461eà005)(2.4915eà001)(4.8084eà001)

f6 6.9220eà0128.4738eà012 1.1684eà006 1.7768eà003 2.0990e+0017.0245e+003

(2.2300eà011)(3.0308eà011)(8.2811eà007)(8.3204eà004)(3.1491e+000)(2.2161e+004)

f78.5539e+002 2.8030e+002 1.1738eà003 3.9484e+000 4.1721e+002 2.6858e+003

(3.0424e+002)(1.4748e+002)(4.3166eà003)(2.1624e+001)(1.4368e+002)(3.4031e+002)

f89.7757e+003 5.4738e+003 5.9327e+002 3.4043e+002 2.5208e+0038.0722e+003

(2.2508e+003)(1.5381e+003)(2.4670e+002)(1.6069e+002)(7.7844e+002)(2.0146e+003)

f9à4.5000e+002à450à4.5000e+002à4.5000e+002à4.2585e+002 3.2059e+002

(1.2655eà008)(1.1942eà013)(1.4911eà011)(1.2235eà006)(4.5090e+000)(1.9590e+002)

f10 2.2198e+004 1.1990e+0048.9591e+002 1.0115e+002 2.2995e+0038.8052e+003

(5.5295e+003)(3.8401e+003)(5.7904e+002)(2.2891e+002)(1.0218e+003)(2.0989e+003)

f11 4.1851e+007 2.0056e+007 6.7207e+006 6.3989e+006 2.6575e+007 6.8280e+007

(1.7194e+007)(8.2315e+006)(2.6293e+006)(2.5094e+006)(8.3295e+006)(2.4975e+007)

f12 3.0884e+004 1.8353e+004 1.0991e+003 2.9116e+002 3.7482e+0038.9484e+003

(1.0873e+004)(7.4429e+003)(5.4102e+002)(3.5765e+002)(1.2450e+003)(3.2689e+003)

f13 4.7132e+003 2.9687e+003 1.4566e+003 3.0345e+003 1.3774e+004 4.3094e+006

(6.3945e+003)(5.1419e+003)(2.3845e+003)(4.7491e+003)(7.7463e+003)(1.8392e+006)

f14à1.7636e+002à1.7898e+002à1.7989e+002à1.7975e+002à1.7650e+002à1.3888e+002

(2.7628e+000)(3.3633eà001)(5.4303eà002)(9.7887eà002)(9.7624eà001)(1.0368e+001)

f15à1.1947e+002à1.1951e+002à1.1957e+002à1.1946e+002à1.1896e+002à1.1897e+002

(8.9905eà002)(1.1339eà001)(1.0229eà001)(1.1727eà001)(6.0986eà002)(5.8540eà002)

f16à3.1812e+002à3.2688e+002à3.2993e+002à3.2983e+002à3.2987e+002à2.2752e+002

(4.4830e+000)(1.9506e+000)(2.5243eà001)(3.7710eà001)(3.4405eà001)(1.3046e+001)

D.Zou et al./Neurocomputing73(2010)3308–33183315

x i (i ?1,y ,n 1)is the same as y j (j ?1,y ,n 2),then we should consider the optimization times of two algorithms.If the optimization time of A algorithm is less than that of B algorithm,then A algorithm is the winner,otherwise,B algorithm is the winner.

After learning the above rule,three groups of Mann–Whitney U tests are carried out,and they are (U NGHS ,U HS ),(U NGHS ,U IHS )and (U NGHS ,U SGHS ),respectively.All 16unconstrained problems (f 1àf 16,N ?50)are selected to carry out these tests.The parameters settings of the four harmony search algorithms are the same as those of Section 4.1,and 30independent experiments are carried out in each case.The three groups of Mann–Whitney U tests can be stated as Tables 6–8.

As can be seen from Table 6,the values of U NGHS are all equal to 0,in contrast,the values of U HS are all equal to 900which is much larger than 0.Statistically,the NGHS can beat the HS on solving all 16unconstrained functions,which indicates that the performance of the NGHS is better than that of the HS.

As can be seen from Table 7,the values of U NGHS are all smaller than those of U IHS for all functions.Specially,the values of U NGHS and U IHS are 0and 900,respectively,for any function.In other words,the numbers of NGHS solutions beaten by IHS solutions are 0for all 16unconstrained problems.Thus,we can safely conclude that the IHS has been badly defeated by the NGHS on solving all 16functions.

As can be seen from Table 8,the values of U NGHS are smaller than 450for all functions except f 15,in the mean time,the values of U SGHS are larger than 450for all functions except f 15.In short,the

NGHS has obtained a larger number of wins when compared to the SGHS,which indicates that the NGHS is superior to the SGHS.

5.Conclusions

A novel global harmony search (NGHS)algorithm is introduced in this paper.Unlike the HS algorithm,the NGHS designs a novel

Table 5

The effect of HMS on the performance of the NGHS (p m ?0.005).Function HMS ?5

HMS ?10HMS ?20HMS ?40HMS ?80f 1 3.7455e à012 6.6008e à013 6.4703e à012 6.7822e à011 1.0727e à009(1.1753e à011)(8.1519e à013)(1.0961e à011)(1.3287e à010)(1.1924e à009)f 2 5.5693e+002 1.8485e+002 6.6884e+001 2.8060e+001 4.3644e+001(1.6224e+003)(5.6518e+002)(7.5718e+001)(2.7791e+001)(4.3756e+001)f 3 6.6345e à002 6.6406e à002 1.6608e à001 2.3356e à001 6.8403e à001(2.5247e à001)(2.5241e à001)(4.5872e à001)(5.0089e à001)(8.0939e à001)f 48.8581e à002 4.9286e à002 3.7184e à002 3.9671e à002 2.3203e à002(5.9176e à002)(4.7725e à002)(3.2936e à002)(4.2763e à002)(3.5180e à002)f 5 5.3823e à007 2.7800e à007 6.7582e à007 1.6858e à0069.2713e à006(6.2054e à007)(1.7985e à007)(3.8313e à007)(7.5629e à007)(3.5933e à006)f 6 1.9709e à006 1.2651e à0067.0816e à006 3.5219e à005 2.2310e à004(3.3050e à006)(8.0173e à007)(6.7054e à006)(2.3731e à005)(1.0196e à004)f 7 3.8189e à004 3.8193e à0047.8963e+0007.8963e+000 1.1847e+001(2.6756e à007)(4.6899e à007)(3.0049e+001)(3.0049e+001)(3.6138e+001)f 8 5.4900e+002 5.2437e+001 1.5879e+0009.8972e à001 1.2598e+000(2.3962e+002)(2.7859e+001)(7.1301e à001)(6.2141e à001)(7.4038e à001)f 9à4.5000e+002à4.5000e+002à4.5000e+002à4.5000e+002à4.5000e+002(4.6250e à012)(1.9141e à012)(1.6327e à012)(3.9962e à011)(1.0251e à009)f 10 6.8696e+002à3.8233e+002à4.4801e+002à4.4887e+002à4.4805e+002(4.9857e+002)(4.0714e+001)(1.0446e+000)(6.7255e à001)(1.1716e+000)f 117.0622e+006 3.5850e+006 3.1335e+006 2.4065e+006 2.7199e+006(2.7765e+006)(1.7565e+006)(1.6418e+006)(1.1816e+006)(1.4497e+006)f 128.1370e+002à3.7498e+002à4.4778e+002à4.4827e+002à4.4767e+002(5.3745e+002)(3.8696e+001)(8.7792e à001)(1.3109e+000)(1.6420e+000)f 13 2.1212e+003 1.1022e+003 5.1917e+002 4.9355e+002 4.6619e+002(3.0182e+003)(1.7786e+003)(2.2059e+002)(1.3083e+002)(9.5065e+001)f 14à1.7988e+002à1.7995e+002à1.7996e+002à1.7998e+002à1.7998e+002(5.3773e à002)(2.6152e à002)(1.9606e à002)(1.8976e à002)(1.8882e à002)f 15à1.1956e+002à1.1963e+002à1.1971e+002à1.1972e+002à1.1974e+002(1.0329e à001)(7.5334e à002)(9.8542e à002)(9.4335e à002)(7.2315e à002)f 16

à3.2997e+002à3.2993e+002à3.2997e+002à3.2977e+002à3.2978e+002(1.8165e à001)

(2.5243e à001)

(1.8165e à001)

(5.0145e à001)

(4.1944e à001)

Table 6

Mann–Whitney U test result obtained using the NGHS and the HS.Function U NGHS U HS f 10900f 20900f 30900f 40900f 50900f 60900f 70900f 80900f 90900f 100900f 110900f 120900f 130900f 140900f 150900f 16

900

D.Zou et al./Neurocomputing 73(2010)3308–3318

3316

location updating strategy,which makes it easier to converge.In addition,the genetic mutation operation of the NGHS is the same as the random selection of the HS,and it can effectively prevent the NGHS from being trapped into local optima.Based on a large number of experiments,the NGHS has demonstrated better performance on solving most unconstrained problems when compared to the other three harmony search algorithms. Acknowledgement

This work was supported by National Science Foundation of P.R.China under Grants60674021.

References

[1]J.H.Holland,Adaptation in Natural and Arti?cial Systems,University of

Michigan Press,Ann Arbor,MI,1975.

[2]R.Storn,Differential evolution design of an IIR-?lter,in:IEEE International

Conference on Evolutionary Computation,Nagoya,1996,pp.268–273. [3]J.Kennedy,R.C.Eberhart,Particle swarm optimization,in:Proceedings of IEEE

International Conference on Neural Networks,1995,pp.1942–1948.

[4]M.Dorigo,V.Maniezzo,A.Golomi,Ant system:optimization by a colony of

cooperating agents,IEEE Transactions on SMC26(1)(1996)29–41.

[5]S.Kirkpatrick,C.Gelatt,M.Vecchi,Optimization by simulated annealing,

Science220(1983)671–680.

[6]J.Sanchis,M.A.Martanez,X.Blasco,Integrated multiobjective optimization

and a priori preferences using genetic algorithms,Information Sciences178

(4)(2008)931–951.

[7]J.J.Zhang,Y.H.Zhang,R.Z.Gao,Genetic algorithms for optimal design of

vehicle suspensions,in:IEEE International Conference on Engineering of Intelligent Systems,2006,pp.1–6.

[8]A.Y.Qing,Dynamic differential evolution strategy and applications in

electromagnetic inverse scattering problems,IEEE Transactions on Geoscience and Remote Sensing44(1)(2006)116–125.

[9]J.H.V.Sickel,K.Y.Lee,J.S.Heo,Differential Evolution and its Applications to

Power Plant Control,in:International Conference on Intelligent Systems Applications to Power Systems,ISAP,2007,pp.1–6.

[10]Y.Marinakis,M.Marinaki,G.Dounias,Particle swarm optimization for pap-

smear diagnosis,Expert Systems with Applications35(4)(2008)1645–1656.

[11]K.M.Sim,W.H.Sun,Ant colony optimization for routing and load-balancing:

survey and new directions,IEEE Transactions on Systems,Man and Cybernetics,Part A33(5)(2003)560–572.

[12]M.Serrurier,H.Prade,Improving inductive logic programming by using

simulated annealing,Information Sciences178(6)(2008)1423–1441. [13]Z.W.Geem,J.H.Kim,G.V.Loganathan,A new heuristic optimization

algorithm:harmony search,Simulation76(2)(2001)60–68.

[14]Z.W.Geem,J.H.Kim,G.V.Loganathan,Harmony search optimization:

application to pipe network design,International Journal of Model Simulation 22(2)(2002)125–133.

[15]Z.W.Geem,C.L.Tseng,Y.Park,Harmony search for generalized orienteering

problem:best touring in China,in:Springer Lecture Notes in Computer Science,vol.3412,2005,pp.741–750.

[16]J.H.Kim,Z.W.Geem, E.S.Kim,Parameter estimation of the nonlinear

Muskingum model using harmony search,Journal of American Water Resource Association37(5)(2001)1131–1138.

[17]K.S.Lee,Z.W.Geem,A new structural optimization method based on the

harmony search algorithm,Computer and Structures82(9–10)(2004)781–798.

[18]M.Mahdavi,M.Fesanghary, E.Damangir,An improved harmony search

algorithm for solving optimization problems,Applied Mathematics and Computation188(2)(2007)1567–1579.

[19]M.G.H.Omran,M.Mahdavi,Global-best harmony search,Applied Mathe-

matics and Computation198(2)(2008)643–656.

[20]Q.K.Pan,P.N.Suganthan,M.F.Tasgetiren,J.J.Liang,A self-adaptive global best

harmony search algorithm for continuous optimization problems,Applied Mathematics and Computation216(3)(2010)830–848.

[21]Q.K.Pan,P.N.Suganthan,J.J.Liang,M.F.Tasgetiren,A local-best harmony

search algorithm with dynamic subpopulations,Engineering Optimization42

(2)(2010)101–117.

[22]S.Das,A.Mukhopadhyay,A.Roy,A.Abraham,B.K.Panigrahi,Exploratory

power of the harmony search algorithm:analysis and improvements for global numerical optimization,IEEE Transactions on Systems,Man,and Cybernetics,Part B:Cybernetics(99)(2010)1–18.

[23]D.X.Zou,L.Q.Gao,J.H.Wu,S.Li,Y.Li,A novel global harmony search

algorithm for reliability problems,Computers&Industrial Engineering58(2) (2010)307–316.

[24]P.N.Suganthan,N.Hansen,J.J.Liang,K.Deb,Y.P.Chen,A.Auger,S.Tiwari,

Problem De?nitions and Evaluation Criteria for the CEC2005Special Session on Real-Parameter Optimization,Technical Report,Nanyang Technological University,Singapore,May2005,KanGAL Report#2005005,IIT Kanpur,India.

[25]F.Wilcoxon,Individual comparisons by ranking methods,Biometrics1

(1945)80–83.

[26]H.B.Mann,D.R.Whitney,On a test of whether one of two random variables is

stochastically larger than the other,Annals of Mathematical Statistics18 (1947)

50–60.

Dexuan Zou received the B.Sc.degree in electronic

information science and technology from Liaoning

University in2005,and he obtained his M.S.degree

in signal and information processing from North-

eastern University,Shenyang,China,in2008.He is

currently working towards the Ph.D.degree in the?eld

of control theory&control Engineering at North-

eastern University.His current research interests are

evolutionary algorithms,arti?cial intelligence,and

optimal

control.

Liqun Gao received the M.S.and Ph.D.degrees in auto-

matic control from Northeastern University,Shenyang,

China,in1985and1991,respectively.Currently,he is a

professor in the control theory and navigation technology

department,Northeastern University.His current research

interests are arti?cial intelligence,control theory and

control methods,and pattern recognition.

Table7

Mann–Whitney U test result obtained using the NGHS and the IHS.

Function U NGHS U IHS

f10900

f20900

f30900

f40900

f50900

f60900

f70900

f80900

f90900

f100900

f110900

f120900

f130900

f140900

f150900

f160900

Table8

Mann–Whitney U test result obtained using the NGHS and the SGHS.

Function U NGHS U SGHS

f10900

f2182718

f30900

f422878

f50900

f60900

f70900

f80900

f90900

f100900

f1120880

f120900

f13172728

f140900

f1584555

f160900

D.Zou et al./Neurocomputing73(2010)3308–33183317

Jianhua Wu received the B.Sc.and M.S.degrees in automatic control from Northeastern University, Shenyang,China,in1978and1986,respectively. Currently,she is a professor in the electric power system and automation department,Northeastern University.Her current research interests are digital image processing,research and development of power converter,and pattern

recognition.Steven Li received the Ph.D degree from Delft University of Technology,The Netherlands.Currently, he is an associate professor in the division of business, University of South Australia.His current research interests are quantitative?nance,corporate?nance, asset pricing,ef?ciency of?nancial markets,and investments.

D.Zou et al./Neurocomputing73(2010)3308–3318 3318

英语精读第二册课文翻译

UNIT 2-1 一场关于男人是否比女人勇敢的激烈的讨论以一个意外的方式。晚宴我最初听到这个故事是在印度,那儿的人们今天讲起它来仍好像实有其事似的——尽管任何一位博物学家都知道这不可能是真的。后来有人告诉我,在第一次世界大战之后不久就出现在一本杂志上。但登在杂志上的那篇故事, 以及写那篇故事的人,我却一直未能找到。故事发生在印度。某殖民官员和他的夫人举行盛行的晚宴。跟他们一起就座的客人有——军官和他人的夫人,另外还有一位来访的美国博物学家——筵席设在他们家宽敞的餐室里,室内大理石地板上没有铺地毯;屋顶明椽裸露;宽大的玻璃门外便是阳台。席间,一位年轻的女士同一位少校展开了热烈的讨论。年轻的女士认为,妇女已经有所进步,不再像过去那样一见到老鼠就吓得跳到椅子上;少校则不以为然。“女人一遇到危急情况,”少校说,反应便是尖叫。而男人虽然也可能想叫,但比起女人来,自制力却略胜一筹。这多出来的一点自制力正是真正起作用的东西。”那个美国人没有参加这场争论,他只是注视着在座的其他客人。在他这样观察时,他发现女主人的脸上显出一种奇异的表情。她两眼盯着正前方,脸部肌肉在微微抽搐。她向站在座椅后面的印度男仆做了个手势,对他耳语了几句。男仆两眼睁得大大的,迅速地离开了餐室。在座的客人中,除了那位美国人以外论证也没有注意到这一幕,也没有看到那个男仆把一碗牛奶放在紧靠门边的阳台上。那个美国人突然醒悟过来。在印度,碗中的牛奶只有一个意思——引蛇的诱饵。他意识到餐室里一定有条眼镜蛇。他意识到餐室里一定有条眼镜蛇。他抬头看了看屋顶上的椽子——那是最可能有蛇藏身的地方——但那上面空荡荡的。室内的三个角落里也是空的,而在第四个角落里,仆人们正在等着下一道菜。这样,剩下的就只有一个地方了餐桌下面。他首先想到的是往后一跳,并向其他人发警告。但他知道这样会引起骚乱,致使眼镜索受惊咬人。于是他很快讲了一通话,其语气非常威严,竟使所有的人安静了下来。我想了解一下在座的诸位到底有多大的克制能力,我数三百下——也就五分钟——你们谁都不许动一动。动者将罚款五十卢比。准备好!”在他数数的过程中,那2 0 个人像一尊尊石雕一样端坐在那儿。当他数到“……280……”时,突然从眼然处看到那条眼镜蛇钻了出来,向那碗牛奶爬去。在他跳起来把通往阳台的门全都砰砰地牢牢关上时,室内响起了一片尖叫声。“你刚才说得很对,少校!”男主人大声说。一个男子刚刚为我们显示了从容不迫、镇定自若的范例。”“且慢”,那位美国人一边说着一边转向女主人。温兹太太,你怎么知道那条眼镜蛇是在屋子里呢?”女主人的脸上闪现出一丝淡淡的微笑,回答说:“因为它当时正从我的脚背上爬过去。” UNIT2 杰斐逊很久以前就死了,但是我们仍然对他的一些思想很感兴趣,杰斐逊的箴言, 布鲁斯.布利文、托马斯.杰斐逊美国第三任总统,也许不像乔治.华盛顿和亚伯拉罕.林肯那样著名,但大多数人至少记得有关他的一件事实:《独立宣言》是他起草的。虽然杰斐逊生活在二百多年以前,但我们今天仍可以从他身上学到很多东西。他的许多思想对当代青年特别有意义。下面就是他讲过和写到过的一些观点:自己去看。杰斐逊认为,一个自由的人除了从书本中获取知识外,还可以从许多别的来源获得知识;亲自做调查是很重要的。当他还年轻的时候,他就被任命为一个委员会的成员,去调查詹姆斯河南部支流的水深是否可以通行大型船只。委员会的其他成员都坐在州议会大厦内,研究有关这一问题的文件,而杰斐逊却跳进一只独木舟去做现场观测。你可以向任何人学习。按出身及其所受的教育,杰斐逊均属于最高的社会阶层。然而很少跟出身卑贱的人说话的年代,在那个贵人们除了发号施令以外。杰斐逊却想尽办法跟园丁、仆人和侍者交谈。有一次杰斐逊曾这样对法国贵族拉斐特说:你必须像我那样到平民百性的家里去,看看他们的烧饭锅,吃吃他们的面包。只要你肯这样做,你就会发现老百姓为什么会不满意,你就会理解正在威胁着法国的革命。”自已作判断。未经过认真的思考,杰斐逊绝不接受别人的意见。“不要相信它或拒绝它。

土耳其电力市场分析

土耳其电力市场土耳其电力市场概况概况 一一、土耳其电力市场发展现状土耳其电力市场发展现状 随着土耳其经济的高速发展,土耳其电力市场正成为土耳其经济发展最快的领域之一。自1980年起,土耳其的电力需求就开始快速增长,2009年达到了194兆瓦特-小时(TWh)(图1)。目前土耳其的人均用电量低于2200千瓦时,远低于欧盟6602千瓦时的平均水平。根据土耳其能源部最近的调查显示,土耳其的电力需求将从2008年的198 TWh 增加到2017年的363TWh。 图1 19801 1980--2009年土耳其电力需求年土耳其电力需求((TWh TWh)) 数据来源:TEDA ?,E üA ?。 注:7.8%、3.6%是复合年均增长率(CAGR)。 电力产业可分为四个垂直分工的部门:发电、传输、分配和零售。在土耳其,目前除了传输环节仍完全由国有公司TEIA ?控制外,其他环节均引入了私营企业(图2)。

数据来源:EüA?,TE?A?,TEDA?。 土耳其电力市场自由化时间表 二、土耳其电力市场自由化时间表 在土耳其电力工业的发展初期,曾有外国企业参与,之后由地方公共团体承担。1950年以后,私营企业逐渐参与。1970年10月,根据国家第1312号法令,设立土耳其电力局(TEK),垄断性的经营发电、输电、配电业务。根据土耳其3096号法令,从1984年开始,允许私营部门进入电力市场,但只有极少数的民营企业参与经营电力。1994 年,一贯垄断经营发电、输电、配电的TEK被分割成发电、输电公司TEA?和配电公司TEDA?。2001年 TEA?解体为EüA?、TE?A?和TEDA?,这三家公司的主营业务分别是发电、输电和零售。2005-2010年,土耳其配电领域的私有化开始,预计在2005-2010的5年中TEDA?将被21个私营配电公司所取代。2007年,土耳其发电领域开始了私有化进程。2008年,拥有总装机容量141MW的ADüA?公司成功完成了私有化,这是土耳其政府私有化管理局(Privatisation Administration)

中东国家有哪些风俗

中东包括那些国家? “中东地区”或“中东”是指地中海东部与南部区域,从地中海东部到波斯湾的大片地区,“中东”地理上也是非洲东北部与欧亚大陆西南部的地区。 中东名称的来源据说是:16-17世纪欧洲殖民者向东殖民时,把距离欧洲的地理位置按远近划分成了:近东,中东,远东。 中东大部分为西亚,但与西亚的区别是:1.中东不包括阿富汗。2.中东包括非洲国家埃及。3.中东包括了土耳其的欧洲部分。 中东是两洋三洲五海之地,其处在联系亚欧非三大洲,沟通大西洋和印度洋的枢纽地位。其三洲具体指亚欧非三大洲,五海具体指里海,黑海,地中海,红海,阿拉伯海。交通便利,海陆空的路线,可顺利运送石油到各国。 “中东”不属于正式的地理术语。一般说来包括巴林、埃及、伊朗、伊拉克(专题,图库)、以色列、约旦、科威特、黎巴嫩、阿曼、卡塔尔、沙特、叙利亚、阿联酋和也门,巴勒斯坦、马格里布国家(阿尔及利亚、利比亚、摩洛哥、突尼斯)以及苏丹、毛里塔尼亚和索马里,由于其历史文化原因一般认为属于中东国家,土耳其和塞浦路斯尽管地理上属于中东地区的一部分,但是他们自身认为属于欧洲;北边的阿富汗有时也与中东联系密切。 政治概念上的中东问题系指阿拉伯国家(包括巴勒斯坦)与以色列之间的冲突问题,也称阿以冲突。中东问题是列强争夺的历史产物,也是世界上持续时间最长的地区热点问题,至今已半个多世纪。中东问题的核心是巴勒斯坦问题。 中东包括那些国家?使用的语言? “中东”概念究竟包括哪些国家和地区,国内外尚无定论,但一般泛指西亚,北非地区,约24个国家,1500余万平方公里,3.6亿人口。 西亚国家包括沙特、伊朗、科威特、伊拉克、阿联酋、阿曼、卡塔尔、巴林、土耳其、以色列、巴勒斯坦、叙利亚、黎巴嫩、约旦、也门和塞浦路斯。 北非包括:苏丹、埃及、利比亚、突尼斯、阿尔及利亚、摩洛哥、马德拉群岛、亚速尔群岛 阿拉伯国家主要用阿拉伯语、库尔德语、土耳其语、波斯语,以色列主要用希伯来语. 中东包括北非的埃及和西亚各国中除了阿富汗以外的其他国家。主要使用阿拉伯语。 中东地区都有什么语言?各自国家的母语都是什么? 中东地区国家包括巴林、埃及、伊朗、伊拉克、以色列、约旦、科威特、黎巴嫩、阿曼、卡塔尔、沙特、叙利亚、阿联酋和也门,巴勒斯坦、马格里布国家(阿尔及利亚、利比亚、摩洛哥、突尼斯)以及苏丹、毛里塔尼亚和索马里,这里面大多数阿拉伯国家的通用语言是阿拉伯语,以色列的通用语言是希伯来语,伊朗的通用语言是波斯语,土耳其和塞浦路斯虽然地理上属于中东地区,但他们一般认为属于欧洲,而阿富汗虽地理上不属中东地区,但与中东联系密切.此外,在这些国家还有大量的民族,比如说库尔德族、贝督因各部族,这些民族说本民族的语言。而苏丹等东非洲国家通用英语. 因信奉伊斯兰教,通用官方语言是阿拉伯语,只是类似于我国的方言,发音有区别,以色列说的是希伯来语,伊朗说的是波斯语,土耳其说土耳其语. 中东国家有哪些风俗? 阿联酋虽然是穆斯林国家,信仰伊斯兰教,但国家实行对外全方位开放,政策较开明,对外国人在衣食住行等方面没有太多的限制,超级市场可以买到猪肉及其制品,基本可满足居住在阿联酋各国人士的需求,值得注意的是:

大学英语精读第二册翻译

翻译 Unit1 1.她砰地关上门,一声不吭地走了,他们之间那场争执就此结束。 Their argument ended when she slammed the door and left without a word. 2. 出席晚宴的客人对那个美国人威严的语气感到有点意外。 The guests at the dinner party were slightly surprised at the commanding tone of the American. 3. 约翰尼已长大成熟,不再害怕独自呆在家里了。 Johnny has outgrown the fear of staying at home alone. 4. 当全部乘客都向出口处(exit) 走去时,他却独自留在座位上,好像不愿意离开这架飞机似的。 While all the other passengers made for the exit, he alone remained in his seat as if unwilling to leave the plane. 5. 这封信必须交给威尔逊博士本人。 The letter is to be handed to Dr. Wilson himself. 6. 南希虽然很想参加辩论,但腼腆得不敢开口。 While she felt like joining in the argument, Nancy was too shy to open her mouth. 7. 你觉得什么时候最有可能在家里找到他? What do you think is the likeliest time to find him at home? 8. 猎人一看见有只狐狸从树丛中出现并向他设下(lay) 的陷阱(trap) 方向跑去,脸上顿时闪出了兴奋的表情。 The hunter’s face lit up with excitement as soon as he saw a fox emerge from among the bushes and run in the direction of / make for the trap he had laid. Unit 2 1) 会上有人建议任命一个十一人委员会来制定新章程。 It was suggested at the meeting that a committee of eleven be appointed to make a new constitution. 2) 这些青年科学家通过现场观察,获得了研究工作所需的第一手资料。 By making on-the-spot observations, the young scientists obtained first-hand information they needed in their research work. 3) 他很可能会因视力不好而被拒收入伍。 It is very likely that he will be rejected by the army because of his bad eyesight.

中东一带一路国家、土耳其市场分析

二、中线 路线走向:中国(北京、郑州、西安、乌鲁木齐)——中亚——西亚及中东——中欧——西欧 中亚 一句话概况:(机遇)中亚地区电商起步晚,发展相对落后,算是一块未开垦的处女地。 从相关资料了解到,目前,(优势)中亚大部分国家(包括哈萨克斯坦、乌兹别克斯坦、土库曼斯坦、吉尔吉斯斯坦、塔吉克斯坦和阿富汗斯坦)已经有宽带接入,网络基础设施建设有助于互联网用户的增长,但(劣势)中亚地区互联网的商业用途还远远低于亚洲其他国家。同时,由于物产相对匮乏,中亚地区居民生活所需品严重依赖进口,跨境电商是其目前主要的电商形式。 根据市场调研来看,2015年中亚地区的跨境电商已有了重大改变,这主要得益于哈萨克斯坦进入了WTO提高了商业透明度、吉尔吉斯斯坦对欧盟消费进一步扩展等。而促进中亚地区电子商务迅速发展的原因主要有两点: 1、电子商务可以有效降低商品价差,同时由于信息不透明和贸易水平落后,市场上存在严重的价差,而中亚也成为了世界上货物最贵的地方。但是电子商务的发展和渗透可以有效地解决这一问题。 2、电子商务能够触及到传统进口市场无法满足的人群,包括女性(中亚地区居民多为穆斯林,女性有许多禁忌)、小企业和乡村企业。 据悉,哈萨克斯坦的B2C网站chocolife.me和吉尔吉斯斯坦B2B网

站prodsklad.kg比较活跃,也在一定程度上刺激了中亚电商的发展。另外,由于(威胁)在中亚地区,信用卡使用限制,网络支付存在很大风险,货到付款是目前主要的支付方式。 西亚及中东 一句话概况:中东市场有两个鲜明的形象:(机会)一是易被忽视的“金矿”,二是跨境电商伺机爆发。 从告了解到,中东地区电商市场从2011年的70亿美元增长至2012年的90亿美元,有29%的涨幅,2013年这一规模达到112亿美元,2014年达到128亿美元,2015年突破150亿美元。 中东地区各国电商渗透率为:阿联酋46%,沙特阿拉伯25%,黎巴嫩9%,埃及8%,科威特35%。根据Research and Markets的报告,从B2C电商销售额来看,阿联酋是中东地区最大的B2C电商市场,沙特阿拉伯是中东地区第二大B2C电商市场。 在电商基础设施上,物流方面,中东地区的公司只能通过公路传送货物而不是飞机,且目前没有公认的法律规定可参考。支付方面,大约80%的网上购物是货到付款,15%是信用卡和借记卡支付。 由于联系亚、欧、非三大洲,沟通大西洋、印度洋,中东地区自古以来就凭借十分重要的地理位置成为国际贸易要塞。这一地区给人的总体感觉是,居民有钱(石油输出国),但物资缺乏,所以人们跨境网购的热情非常高。 在这一背景下,Amazon、eBay等全球电商早早的就进入中东市场,而近年来,中国的电商企业在此也颇为活跃。根据中国跨境出口电商商

土耳其的经典谚语

土耳其的经典谚语 土耳其在中世纪曾一度是世界上最强大的国家之一,她的强大绝非偶然,而是和这个民族的血液与文化有关的,下面整理了土耳其谚语,欢迎大家阅读。 土耳其谚语(热门)一小勺盐能调好一大碗汤 犬吠,而驮队继续向前。 与朋友吃喝,可是别跟他们做生意。 邻人的鸡仔如鹅;邻人的妻子娇如娥。 通红的果子招石头。 上帝也给每只笨鸟儿准备了一个矮树枝。 上帝每关上一扇门,就打开另外一千扇。 如果技能可以看会的话,那么每条狗都能开肉铺了。 “说”是耕种,“听”才是收获。 能力的话没有学校的。 土耳其谚语(最新)长远来看,吝啬鬼和慷慨人花出去的还是一样多。 饥饿的肚子是不长耳朵的。 未至溪流前,莫把裤腿卷。 每个“糟糕”背后还有个“更糟糕”。 闲散如酸醋,会软化精神的钙质;勤奋如火酒,能燃烧起智慧的

火焰。 科学的敌人,不比朋友少。 不学习的人总以后悔而告终。 真理不需要喋喋不休的誓言。 不劳动的玩乐,就象没有放盐的面包。 谎言的船开不远。 土耳其谚语翻译İnsanoğlu bir varmış bir yokmuş, dünya gelip geçicidir. 人如匆匆过客,天地亦非永恒。 Hırsızlık bir ekmekten, kahpelik bir öpmekten. 盗物一次便是贼,卖淫一回即为娼。 Ağustosta gölge kovan, zemheride karnı ovar. 夏天干活不卖力,冬天就要饿肚皮。 Akı karası geçitte belli olur. (或者) Ak göt, kara göt geçit başında belli olur. 是骡子是马,拉出来遛遛。 Duvarı nem, insanı gam yıkar. 潮湿摧墙倒,忧伤催人老。

大学英语精读 第二册第一、二课 课文翻译

Unit1 The Dinner Party 关于男人是否比女人更勇敢的一场激烈争论以一种颇为出人意料的方式解决了 The dinner party 晚宴 1. I first heard this tale in India, where is told as if true—though any naturalist would know it couldn’t be. Later someone told me that the story appeared in a magazine shortly before the First World War. That magazine story, and the person who wrote it, I have never been able to track down. 我最初听到这个故事是在印度,那儿的人们今天讲起它来仍好像确有其事似的——尽管任何一位博物学家都知道这不可能是真的。后来有人告诉我,在第一次世界大战之前不久,一家杂志曾刊登过这个故事。但登在杂志上的那篇故事以及写那篇故事的人,我却一直未能找到。 2.The country is India.A colonial official and his wife are giving a large dinner party. They are seated with their guests—officers and their wives, and a visiting American naturalist—in their spacious dining room, which has a bare marble floor, open rafters and wide glass doors opening onto a veranda. 故事发生在印度。某殖民地官员和他的夫人正举行盛大的晚宴。筵席设在他们家宽敞的餐室里,室内大理石地板上没有铺地毯;屋顶明椽裸露;宽大的玻璃门外便是走廊。跟他们一起就坐的客人有军官和他们的夫人,另外还有一位来访的美国博物学家。 3. A spirited discussion springs up between a young girl who says that women have outgrown the jumping-on-a-chair-at-the-sight-of-a-mouse era and a major who says that they haven't. 席间,一位年轻的女士同一位少校展开了热烈的讨论。年轻的女士认为,妇女已经有所进步,不再像过去那样一见到老鼠就吓得跳到椅子上;少校则不以为然。 4. "A woman's reaction in any crisis, "the major says, "is to scream. And while a man may feel like it, he has that ounce more of control than a woman has. And that last ounce is what really counts." 他说:“一遇到危急情况,女人的反应便是尖叫。而男人虽然也可能想叫,但比起女人来,自制力却略胜一筹。这多出来的一点自制力正是真正起作用的东西。” 5. The American does not join in the argument but watches the other guests. As he looks, he sees a strange expression come over the face of the hostess. She is straight ahead, her muscles contracting slightly. She motions to the native boy standing behind her chair and whispers something to him. The boy's eyes widen:he quickly leaves the room. 那个美国人没有参加这场争论,他只是注视着在座的其他客人。在他这样观察时,他发现女主人的脸上显出一种奇异的表情。她两眼盯着正前方,脸部肌肉在微微抽搐。她向站在座椅后面的印度男仆做了个手势,对他耳语了几句。男仆两眼睁得大大的,迅速地离开了餐室。 6. Of the guests, none except the American notices this or sees the boy place a bowl of milk on the veranda just outside the open doors. 在座的客人中除了那位美国人以外谁也没注意到这一幕,也没有看到那个男仆把一碗牛奶放在紧靠门边的走廊上。 7. The American comes to with a start. In India, milk in a bowl means only one thing—bait for a snake. He realizes there must be a cobra in the room. He looks up at the rafters—the likeliest place—but they are bare. Three corners of the room are empty, and in the fourth the servants are

土耳其礼品及家用电器展览会展后分析报告

展会名称:第20届2009土耳其国际家庭用品、礼品及家用电器展览会 (Zuchex) 展览时间:2009年10月15日至18日 展览地点:土耳其,伊斯坦布尔,TüYAP展览中心, 展馆名称:1号 --10号展馆 网站:https://www.360docs.net/doc/966346959.html, 展会综述: 这届2009年秋季展,有超过1000间来自22个国家和 地区的企业参展,展厅规模达到10个展厅之雄伟规 模。该展是世界第四,欧洲第三的家庭用品展览会, 是第20届举办。每年分春秋两届举行,展出面积达到 60,000平方米。2008年秋季展已经有16个国家和地 区的550多家企业参展,专业买家超过45,000人次, 来自全球100个国家和地区,主要来自欧洲、美国、 中东及亚洲等地。土耳其每年官方消费品部分采购进 口额为10亿欧元,近几年越来越多的西欧买家发现该 展是个不错的产品采购资源,这部分买家的数量正呈 显著增长的趋势。由于土耳其毗邻欧亚非三大陆的特 殊地理环境,使该展成为辐射欧亚非三大陆的枢纽型 展览会。2008年春季开始该展首次设立国际展区,是 有意进军欧亚市场的出口型贸易和生产企业的绝佳良 机。 参展范围: 家用小电器,各类礼品,家居装饰品,家纺用品,餐厨用品,玻璃制品,塑料制品,陶瓷制品,灯具及照明设备。 买家分布:欧洲,南北美,亚洲,中东,非洲 新产品推荐:

藤制家居用品 Portable Grill Standard Drum Oven 背景为画的tableware 花纹典雅高尚的Dinnerware

展会情况及市场展望: 开展4天,我司展位一直是同列展位买家人流最旺,受关注度最高之处.虽然同行https://www.360docs.net/doc/966346959.html,紧靠旁边,他们非常羡慕我司展位前驻足登记询盘的买家数量之多,并表示他们参该展数次,第一次见有如此的买家人流量。这次来我司登记的买家多数为各公司的President或者Import Manager,占买家比率77.83%,其余问贸易公司和采购商的Product Develop Manager。 在历史过程中,作为连接亚欧大陆的灵魂之都,伊 斯坦布尔成为丝绸之路之间的东部和西部的世界 过渡点,至今仍保持这一传统的生存延续。该枢纽 型展览会的举办,吸引了不少来自欧洲,中东,亚 洲,非洲和南北美洲的买家,对2010年这些地区 的家居和家电用品发展趋势起了至关重要的作用。环球市场展会专业展会数据分析

在中东做生意

如何在中东做生意 2010-05-11 11:10:39| 分类:关于中东.中国商阅读376 评论0 字号:大中小订阅 -- 关于中东市场 中东指巴林、塞浦路斯、埃及、伊朗、伊拉克、以色列、约旦、科威特、黎巴嫩、阿曼、卡塔尔、沙特阿拉伯、叙利亚、阿联酋、也门等 15 个国家。 一、市场概述 中东市场通常指环绕波斯湾和阿拉海的9个国家和周边阿拉伯国家,总人口达到5-7亿,人均年收入从阿联酋,科威特等的3-4万美元到伊朗,伊拉克,也门等国家的年人均收入5-6千美元不等,这些阿拉伯国家的轻工,日用,电子,服装基本外要依赖进口,产品的价格要求为中低要求,档次不是非常高.同时这些部分地区,几乎可以讲是一种完全纯消费的特点。中国产品在全世界以物美价廉著称,中国作为世界第一轻工,电器,服装等产品大国,以绝对优势占据整个中东市场。尽管欧美市场是目前世界上最发达、最成熟的市场,但这个市场的竞争十分激烈,商品趋于饱和,加上高进口关税及贸易壁垒的限制,后来者或中小实力的企业很难进入。纵览当今世界,唯有中东和非洲市场是一个充满商机、前景良好的热点市场。 说起中东,就不能不提阿联酋和迪拜。迪拜是阿联酋7个酋长国中的一个,位于阿拉伯半岛的东端,处于"五海三洲"中点的重要战略位置,是东西方的交通要道和贸易枢纽。每天通过迪拜转口的集装箱达到几万个,为中东第一大港口,在迪拜你可以找到全世界120国家的商人,他们常年穿梭于中东与本国,从事商品贸易。据不完全统计,阿联酋年贸易额约800亿美元,由于自然资源的匮乏,除了石油和天然气外的其它产品,从工业用原料、设备到民用生活物品均依赖进口,故阿政府一直实行开放的自由贸易政策,没有贸易壁垒,无外汇管制及其管理机构,从得到授权的银行可以无限制地获得外汇,没有征收公司或企业的利润税和营业税的规定,没有所得税、增值税、消费税和中间环节的各种税收,利润可以自由汇出。除烟、酒等极个别商品外,其它大部分商品只是象征性征收1%-4%的关税。外贸在阿联酋经济中占有重要位置。阿各类商贸公司14108家,其中外国公司680家,国有中国公司37家,自由贸易区4个。1995年,加入世界贸易组织。进口主要有粮食、机械和消费品。阿转口贸易比重较大,同100多个国家和地区有贸易关系,与40多个国家签订了双边贸易协议和避免双重征税协议。我国主要出口阿联酋的主要产品为:轻工,纺织,机电,和家用电器,由于国内厂家对阿市场的重视不够,使得出口产品的质量有所下降,直接影响到价格,目前所有出口产品的纯利润从1997年的45%,下降到17%,部分产品下降到10%左右,在2001年出口中东,仅仅到迪拜港口的货物容值达到25亿美元。 往来中东80%以上的货物要经过迪拜进行中转,同时辐射到非洲大部分国家,辐射人口达到13亿。在这里云集了非洲近30多个国家的客商,常年在这里采购日用,轻工,电器,服装,等货物。通常进口交易额度的75%转口非洲市场,20%转口周边海湾国家,5%直接在阿联酋消费。迪拜素有“中东的香港”之美誉,是阿联酋的金融经济中心,以其自由宽松的经济政策、得天独厚的地理位置、完善齐备的基础设施,迅速成为中东地区的交通枢纽和最大的货物集散地。通过迪拜,货物可转销到海湾地区、俄罗斯、东欧、非洲、地中海。目前我国产品以种类繁多、档次适中、规格齐全、价格合理颇受青睐,进一步开发潜力极大,尤其是机电五金、汽摩配件、纺织服装、轻工工艺极具竞争优势。通过迪拜,货物可转销到海湾地区、东欧、非洲、南亚。海湾地区以盛产石油天然气闻名于世界,石油天然气的储量占世界65%以上,沙漠面积占陆地面积80%以上。整个海湾地区除了石油天然气工业比较发达以外,几乎没有其它的工业。全部的工业产 品依赖进口。

土耳其概况

土耳其概况 【国名】土耳其共和国(Republic of Turkey)。 【面积】78.36万平方公里,其中97%位于亚洲的小亚细亚半岛,3%位于欧洲的巴尔干半岛。 【人口】7256万,土耳其族占80%以上,库尔德族约占15%;城市人口为4970多万,占总人口的70.5%。土耳其语为国语。99%的居民信奉伊斯兰教,其中85%属逊尼派,其余为什叶派(阿拉维派);少数人信仰基督教和犹太教。 【首都】安卡拉(Ankara),人口447万,年最高气温31℃,最低气温-4℃。 【政体】共和制 【货币】土耳其里拉(Turkish Lira)。 【国家元首】总统阿卜杜拉·居尔(Abdullah Gül), 2007年8月28日由议会选出,当日就任。 【主要节日】新年:1月1日;国家主权和儿童日:4月23日;青年和体育节:5月19日;胜利日:8月30日;共和国成立日:10月29日。 【简况】地跨亚、欧两洲,邻格鲁吉亚、亚美尼亚、阿塞拜疆、伊朗、伊拉克、叙利亚、希腊和保加利亚,濒地中海、爱琴海、马尔马拉海和黑海。海岸线长7200公里,陆地边境线长2648公里。南部沿海地区属亚热带地中海式气候,内陆为大陆型气候。

土耳其人史称突厥,8世纪起由阿尔泰山一带迁入小亚细亚,13世纪末建立奥斯曼帝国,16世纪达到鼎盛期,20世纪初沦为英、法、德等国的半殖民地。1919年,凯末尔领导民族解放战争反抗侵略并取得胜利,1923年10月29日建立土耳其共和国,凯末尔当选首任总统。 【政治】2002年11月,正义与发展党在土第22届议会选举中获胜,实现单独执政,结束了土自1987年以来多党联合执政的局面。正义与发展党上台后,积极推进政治、经济改革,统筹社会协调发展,取得明显成效。2007年7月,正发党以46.6%的得票率赢得大选,继续单独执政。 2011年6月,土举行第24届议会选举,正发党以49.9%的得票率赢得胜利,实现第三次连续单独执政。 【宪法】土立法体系效仿欧洲模式。现行宪法于1982年11月7日生效,是土第3部宪法。宪法规定:土为民族、民主、政教分离和实行法制的国家。 【议会】全称为土耳其大国民议会,是土最高立法机构。共设550个议席,议员根据各省人口比例经选举产生,任期4年。实行普遍直接选举制,18岁以上公民享有选举权。只有超过全国选票10%的政党才可拥有议会席位。本届议会成立于2011年6月29日,是土第24届议会。目前议会议席分布情况是:正义与发展党327席,共和人民党135席,民族行动党52席,和平民主党29席,无党派7席。 【政府】又称部长会议。本届政府是土第61届政府,成立于2011年7月13日,系正义与发展党单独执政,法定任期4年。 【行政区划】土耳其行政区划等级为省、县、乡、村。全国共分为

翻译答案大学英语精读第二册

Unit1 1.她砰地关上门,一声不吭地走了,他们之间那场争执就此结束。 Their argument ended when she slammed the door and left without a word. 2. 出席晚宴的客人对那个美国人威严的语气感到有点意外。 The guests at the dinner party were slightly surprised at the commanding tone of the American. 3. 约翰尼已长大成熟,不再害怕独自呆在家里了。 Johnny has outgrown the fear of staying at home alone. 4. 当全部乘客都向出口处(exit) 走去时,他却独自留在座位上,好像不愿意离开这架飞机似的。 While all the other passengers made for the exit, he alone remained in his seat as if unwilling to leave the plane. 5. 这封信必须交给威尔逊博士本人。 The letter is to be handed to Dr. Wilson himself. 6. 南希虽然很想参加辩论,但腼腆得不敢开口。 While she felt like joining in the argument, Nancy was too shy to open her mouth. 7. 你觉得什么时候最有可能在家里找到他? What do you think is the likeliest time to find him at home? 8. 猎人一看见有只狐狸从树丛中出现并向他设下(lay) 的陷阱(trap) 方向跑去,脸上顿时闪出了兴奋的表情。 The hunter’s face lit up with excitement as soon as he saw a fox emerge from among the bushes and run in the direction of / make for the trap he had laid. Unit2 1) 会上有人建议任命一个十一人委员会来制定新章程。 It was suggested at the meeting that a committee of eleven be appointed to make a new constitution. 2) 这些青年科学家通过现场观察,获得了研究工作所需的第一手资料。 By making on-the-spot observations, the young scientists obtained first-hand information they needed in their research work. 3) 他很可能会因视力不好而被拒收入伍。 It is very likely that he will be rejected by the army because of his bad eyesight. 4) 委员会成员在新机场最佳选址(location) 这一问题上持有不同意见。 The committee members have conflicting opinions as to the best location of the new airport. 5) 亨利创作的艺术品在许多方面比他兄弟的要好。 Henry's works of art are superior in many respects to those of his brother's. 6) 我们产品质量的稳步提高在很大程度上是由于设备有所改进。 The steady rise in the quality of our products owes much to the improvement of our equipment. 7) 吉姆本想按照自己的判断行事,但他没有这样做,因为作为军人他得服从命令。 Jim would have preferred to act on his own judgment, but he didn't because as a soldier he had to obey the order. 8) 如果让我来决定我们是要一个没有自行车的城市呢,还是要一个没有汽车的城市,我会毫不犹豫地选择后者。 Were it left to me to decide whether we should have a city without bikes or one without cars, I should not hesitate a moment to prefer the latter. Unit3

与土耳其人做生意应该注意什么

与土耳其做生意,需要注意什么? 与土耳其做生意,需要注意什么? 与土耳其商人做生意,要特别注意以下有些问题,以将商业风险降低到最低。 ?出口商品质量要保证 中土双边经贸往来中出现的问题有相当部分是源于商品质量。土商很善于经商,善于讨价 还价,但不能因为土商压价而提供不合格的出口商品。 ?要作到单单相符、单证相符 任何不符点都有可能被用作拒收、拒付的理由。即使是信用证付款也不例外。在土耳其银 行开立信用证,需100%的资产抵押,因此,一般情况下,只要土商赎单,收汇就有保证。 ?选择较保险的收汇方式 付款方式是贸易环节中的最重要的一环。为了收汇安全,以下几种方式可供参考: 1、信用证。 有实力、有信誉、效益好的公司都能应对方要求开立信用证。远期信用证亦可接受,但开 证行要信誉好的银行。 2、预付+电汇。 许多土耳其商人一般不愿开立信用证,因为开证费用高,须100%的押金,1.5%的手续费。资金占压严重,商业操作中多有不便。在这种情况下,可考虑: (1)、预付10%,FOB港口交运后通知对方电汇。 (2)、预付10%,FOB港口交运时电汇时付40%,到货时电汇50%。 3、预付+托收方式。 (1)、预付10%,其余采取银行托收方式,提单打空白提单。有风险,但出口商可转移拒收 风险,在国内保险公司投保收汇险,保险费约为货值的3%。 (2)、预付10%,其余采取银行托收方式,提单打空白提单。除现金外,有银行保兑的汇票 和银行保兑的商业支票是较安全的。 ?土耳其海关有三条规定比较重要,希望各有关公司引起注意。 如出口商不谨慎,将会处于非常被动地位: 1、货物到港后,如未收到收货人的正式“拒绝收货通知”,海关不允许出口商将货物拉 回或转运。 2、货物放在海关四个月后,海关将作无货主处理,有权拍卖此货物; 3、货物被拍卖时,原进口商为第一购买人。

大学英语精读第二册课文翻译

大学英语精读第二册课文翻译 (全) UNTH 2-1 It is humorous essay. 这是一篇幽默的文章。 But after reading it you will surely find that the author is most serious in writing it. 但是读过之后你将会发现作者写这篇文章的时候是很严肃的。 Ts There Life on Earth? 地球上有生命吗? Art Buchwald 阿特.布奇沃德 There was great excitement on the planet of Venus this week. 金星上本周异常热闹。 For the first time Venusian scientists managed to land a satellite on the plant Earth, 那里的科学家首次成功地将一颗卫星送上了地球, and is has been sending back signals as well as photographs ever since. 从此卫星便一直不断地发回信号和照片。 The satellite was directed into an area know as Manhattan 卫星被发射到一个叫曼哈顿的地区(named after the great Venusian astronomer Prof. (它是用金星上伟大的天文学家曼哈顿教授的名字命名的, Manhattan, who first discovered it with his telescope 20,000 light years ago). 两万光年前是他首次用望远镜发现了该地区)。 Because of excellent weather conditions and extremely strong signals, 由于良好的天气条件以及高质量的信号, Venusian scientists were able to get valuable information 使得金星上的科学家们能够获得宝贵资料as to the feasibility of a manned flying saucer landing on Earth. 有关载人飞碟能否在地球上着陆。 A press conference was held at the Venus Institute of Technology. 一次有关记者招待会在金星理工学院举行。 "We have come to the conclusion, “我们得出以下结论,based on last week's satellite landing," Prof. Zog said, 根据一周进行的卫星地球着陆,”佐格教授说,"that there is no life on Earth." 地球上不存在生命现象。”

新航路开辟原因之土耳其阻断商路说新论要点

本科毕业论文(设计) 题目:新航路开辟原因之土耳其阻断商路说新论 学生:孔祥宝学号: 201040410108 学院:社会发展学院专业:历史学 入学时间: 2010 年 9 月 11 日 指导教师:洪共福职称:教授 完成日期: 2014 年 4 月 27 日

诚信承诺 我谨在此承诺:本人所写的毕业论文《新航路开辟原因之土耳其阻断商路说新论》均系本人独立完成,没有抄袭行为,凡涉及其他作者的观点和材料,均作了注释,若有不实,后果由本人承担。 承诺人(签名): 年月日

新航路开辟原因之土耳其阻断商路说新论 摘要:西欧的地理大发现的原因很复杂,但奥斯曼土耳其帝国的崛起并不是其中之一,该帝国并未阻碍传统东西商路,地中海东部贸易在中世纪后期一直掌握在以威尼斯为首的意大利各城市国家手中。由于传统的宗教偏见以及帝国对欧洲强大的军事威胁,奥斯曼帝国常遭到抹黑,所谓土耳其阻断商路说也就此出现。 关键词:地中海;贸易;葡萄牙;土耳其;威尼斯 New View Of Turkey Blocked The Route View Of The Reason Of Opening Up New Routs Abstract: The cause of the great geographical discovery of western Europeans are complex, but the rise of the Ottoman Empire is not one of them, the empire did not block the traditional trade routes. In the late middle ages, the trade in the eastern Mediterranean has been controlled by countries headed by Venice Italy cities. Because of the religious prejudice and the powerful military threat of the Ottoman Empire to Europe, the empire has been fequently smeared, the so-called block trade routes said Turkey also appeared. Keywords: Mediterranean; trade; Portugal; Turkey; Venice

相关文档
最新文档