87Vol.2. N.1,Fall2008, control engineers magazine, iran

JournalofControl,2008,V.2,N.1 1387 2 1

1 !" # $ %& '()* +,-./0 0$ 456 , ANFIS 9, :" $ 'PSO

'& ';!?(@

20 B. '$- ' CD$ ( E':5 FE. G! , 5 '(0 .

26 ':E! I( ' J$ D0K , B. +"L! 5%$ M; 5 :!0 6NO ' PE

40 . , EE (C$ '( $ 9, /.$ . $ ':5 0 6R S & = E(,

49 >& UV

JournalofControl,2008,V.2,N.1 1387 2 1

& ' :$ N W.

1387 2 1 =X@ 19-1

Journal of Control, 2008, Iranian Society of Instrument & Control Engineers http://www.isice.ir 1387S I , +.; JN +( ?

!"# $%&' ( )*%+ ,-. /01 1 2! 34 - /ANFIS 5-2 6# (PSO

'& ';! !?(@ l

!" # $ %& ' ()* +,-. /0 0$ 456 ...

2 '& ';! !?(@

Journal of Control, 2008, V.2, N.1 1387 2 1

oN . o 'o oNN 'o o( o!" #m ?oP PM-10 , J ECO . 6, W.m \S #m 'o NN

?P 3g

m J E , ppm1 .

X ' ?P ( !" #m CP ) J( IN h( #@ #m N$.( o$ d& q)P #m N J(

;o 'o

!" # $ %& ' ()* +,-. /0 0$ 456 ...

'& ';! !?(@ 3


', C5 +$ (" #@ o +o$ o o N! 'N +:" : 0 ( #@ #1 X. F.

;' )1 (, )2 ( oV( Fo; , o@ 'o. , 'o) o (o 5%$ #@ ( DN W. F.$ F '. '

@ '. : V( F; G o \S , oE oN X. I(N o . oC o*H +o 5o%$ +Io o oI

(C J(

!" # $ %& ' ()* +,-. /0 0$ 456 ...

4 '& ';! !?(@


W. o456 Wo LN !C J( b(N ,

!" # $ %& ' ()* +,-. /0 0$ 456 ...

'& ';! !?(@ 5


W. . \o DN P$ r , E, ( 0. J( +

!" # $ %& ' ()* +,-. /0 0$ 456 ...

6 '& ';! !?(@


#x$ JP$ , #.= 6ijw );N:, ( o o,0 () : o F(:\ J( : \S)15 (F( F WS :

)8 ( ( )( )

ijijl

ijl

yx t

w t

)9 ( 00

( ) ( )ij ijij

ij i

y t y tw

x u

() ';N:, F4$ 'F( W!6 N : )10 ( EV

V

)11 ( 2

2E E O NetV O VNet

2

2, ,E O NetD O O HO VNet

)12( )13 ( ( )V D O O H

)14( 2 2, ( )V H D O O ( W!6 ',, () ';N:, F4$ 'F :

)15 ( EWW

)16 ( 2 1

2 1E E O Net H NetW O H WNet Net

)17( ( )E D O O V H XW

)18( 2 1W V H X X 1 2 V H )19 (

)20 ( new oldnew old

V V VW W W

! "

+" ooVW , ',, , Noo oo() 'oo;N:, oo& oo oo,HX , +ijx N S G. 456 . 3 .6 . ,# -C $D - ]16-23 /87[

o o o( 'o n (C a$ X DNN $ N 0 (IE W : , FNo oE( J(o E

o456 ?o& FNo ; #& ( F W=$ 456 ' : o( Go F(,o W=$ (| /.$ W.

. 456 (TDL 'o FNo o Fo Wo6 o. 'o N zo0 'no 9: ( n _o(P$ 'o oP( Fo ao

+" +o qoNO o JoP$ o456 ?& q), F) o ( Wo=$ IoN +" o%& o o 9:o" : o6 R , S 9:"

W. +EN 456 FE. : C$ IN J( ! 9:".

456 0. FN o456 %& ' 'oP o ?oS 'o o &$ +EN 456 (P '

ooR (;(oo(N +oo" oo oo( , ':ooE! ,oo= Woo5 ooNL( \& , 9:".

\(, $ o 'o( >o$ WS IN N ' joN 'oFIR o W(I 'IIR :6 + F ' , +: I

o( o( o$ WN /o.$ oN o \o(, \o

!" # $ %& ' ()* +,-. /0 0$ 456 ...

'& ';! !?(@ 7


W5 ,0 , XV () . \ N +" \ )14 ( W.

o%& ' )(zG #o&

1z

o(1

1

)1(1

zz

oS

\ B. (: #.= )17 ((n #@

)21 (

1

1 0 01

1 1 0

1 1 0

( ) ( ) ( ) ( )1 (1 )

( ) (1 ) ( 1) ( 1)( ) (1 ) ( 1) ( 1)

zx t G z x t x tz

x t x t x tx t x t x t

Wo o* o o (| \! 1 +" 'C W6 W; W. _!V.

ooC Woo. oo. \ooS oo 'ooL( Foo; ooN oo! oo Woo.10 ## o o( o5 ( 5 +" 'o;DS o o

=X@ W. W. , 6, ( \0Z , W.21 ## o ( 5 + =Xo@ o* Wo. +" 'o;DS o Z Wo.

G20 ## W.. 'o5 o$ >! ( z0 wP : $ 5FIR ,

5 $S z0 : wPIIR ]16[ . . #.= ! ko0 d= @6 : , ' IN '

+*, W. EE #@ ' o. o \o! J E. EE F F(:.

3 .6 .1 . -C $D

?R'; )15( ,)17( F( 0 0 1 1

0

( ) ( ) ( ) ( )

( )

ij ij ij ij ij ijk ijk

k

ijl ijll

y t w x t w x t w x t

w x t

)22(

, 1( ) (1 ) ( 1) ( 1)

, 0ijl ij ijl ij ij lx t x t x t

l

$)23(

L,0l W.iijl ux . o N IN #.= C 'oMLP Wo. + #o@ \oS

V !TDL : o( o F( #.= &N IN '0 +" $ E N J( ;$ ( , F F

: h$ ( J5 Sk ( : ;$ G. ( 'o ho$ Jo( : ao Jg W. X. +" 456 ?& JP$ 'o

+:, ( ' +:, + CN 'oMLP ': 456 +, .

o F(N +:, ( ;$ N ( hS, ) \o o o$ o)15 ( o oP$ E o 'o (1k o( : ,ioP o \oS +:,

5 F(o0 IN 456 ?& W; . o@ oV o o6NL( 9:" P( F(: .

3 .6 .2 . 56#

F : 9:" W; \S N IN ( #oP jo ':o. X. D0 F

)24 ( 21 1( ( ) ( )) ( )2 2i i ii i

E d n o n e t

#x$ JP$ , #.= 6ijw ,0 () : J( : \S F(:

\ )15( ,)17( F( F WS :

)25( ( )( )

ijijl

ijl

yx t

w t

D $ Jg,)22 ( F(]16[ :

)26 (0 0

( )( )( )

k kijlij

ijl ijl ijlij ijl l

x ty tw w t%

+"

)27 ( , 1, 1

( )( )

(1 ) ( 1) ( 1)

( ) ( 1)

ijlijl

ij

ij ijl ij ij l

ij l ijl

x tt

t t

x t x t

%

% %

)oo #oo.= oo0$t ,0$loo Joo( ooK oo oo#@0ijl% . D)27 ( oD 'o ?

!" # $ %& ' ()* +,-. /0 0$ 456 ...

8 '& ';! !?(@


E L( #.= ;( . Wo. om \oS o( ; N s$ LNL*TDL . , Fo WoS vo0 o1

o( \($ \S 6 5 F S TDL o oP( Wo.h W!6 ( , $ o WoEN 'o$TDL . #o.=BP o '

TDL o;$ Wo. %6 \S W!;. ' N /, : IN W. J( NTDL 9:" N .

4 .ANFIS

'ANFIS o. 1993 'o +(o

!" # $ %& ' ()* +,-. /0 0$ 456 ...

'& ';! !?(@ 9


%& MLP . o ,i& Jg o aoS o o!" o N: '. -Bi1 W. N IN.

4 .1 . G&' (G G 2! -C $D - 21 (6G

ANFIS J( \ ?D () b ')22 ( ]53:[

, () :?P$ h$ () , ',, () : ?oP$ ho$ ( () J( 2 (N 5 456 6, ( +& , . +o& o

, o o. Io \oS : (o ' o Jo. ?oP$ ho$ ,. GN( 9:" ' +& aoC oV (n

0N . a, () :JNS () : o 0 JNS ' +& () J(

N . ?oP$ +Io vo| /o.$ +NS $" #S +I WES J( R ?P$ h$ ',, W. L( (" ) D)31 .((

a. () :+.I!N () : o o$ o +NoS o o$" #S () J( I!N L( JNS $" ' #S . $" #S #@ J(

FEC$ JNS a$ $" ';$S \ j +NS . a;* () :':5 K ().: : W. JNS : a ,0 () J(

;(,, : D0 h$ \ J( (")if ( I!N $" #S v| .

F () :,0 () :,0 o,0 jo : () J( ' \@6 F () ' ) D)33.((

\ )5 ( W!6 MISO ' W. 5 4N n ',, \o >o$ $ ',, 1k 2k ... ,nk ho$

?P$ . j J( 6 +$ R o J( ' +NS+" WN :

1 2 nR k k k )30(

i jj

jAx ',, ?P$ +I P j- ?P$ h$ ajijA-

',, ?P$ h$ & : aj- , #.= LN" F _(P$ a \ )5 (W. (: #@ :

1 21 2

1 2i i inn

i nAA Aw x x x )31( +"

1 1

2 2

1, 2 , , ,

1, 2, , ,,1, 2, ,n n

i k

i k

i k

&

&

&

)32(

1 Mackey-Glass 2 Member Ship Function

,i jjA

',, ?P$ Cj- ?P$ h$ aji - a.

D )31 (iw +NS v| k. $ ,0i- , W. aF( W(;N :

1

1

R

i ii

R

ii

w fO

w

)33(

+"

n

l

il

ili Ribxaf

1,,2,1,

)34(

4 .2. 56# (-2&' ( (6

DN' ( 9:o" 'o o Xo. 9, J(o$ +( ';, .G ' $

!" # $ %& ' ()* +,-. /0 0$ 456 ...

10 '& ';! !?(@


'o J!,EKF o No ':o5 'o]58[ #o@W. 5 . b(N , E ':. D0 +, 9, J( !

W. ! . ,o ';o, : 'E $)C : ( 9:" ANFIS E ]59[ j 9, J(*

o o0 o;N" oE(C o , oN 5P aC ' 9:" +( B. ;, J( $N : o. ,0 aC V ' D0K w6 $

N.. $& D (;, & \! , +( E, q

F : + W. . , >.N . o E, (;,; ' O (;, +& N F #P J( ' ':.

S X. !$ V ' W. J( F; N N o; o= #o@ o !$ V o ':o. \o. !, ';P (. , '

io o o Jog , o. o o4N o \o= oE lX$ J(aL;N7 . !]67 [ ]68 [ ]68 [ 'o( d=

(L ,PSO !, N R 'P * /V 'w5 ]67 [ W.

!" # $ %& ' ()* +,-. /0 0$ 456 ...

'& ';! !?(@ 11


3. 'PSO Po.$ J!o o5 , aC V 9:" 5( !$ V '

! !C J( o o 'oD0 : o ao. , a, ';, 1 oD$ , (I h$ +& ?PSO ;F(.

#=|$ ( 'P V 'o%V F(oL! o PSO o-F(: F

!" # $ %& ' ()* +,-. /0 0$ 456 ...

12 '& ';! !?(@


7. S )?D$ ( q)P1 o$ #oP5 o6 ( v0 ( k0 C6 : ,".

8.C6 +()i' #oo@ oo oo Woo. 'oo& : 5oo%$ oo oo( LNoo

5%$ , W0(~ (0, )i iU c' h(:$N.)L&(%& %& v| LN ) @ VEN PSO ,o= oE( m W&. maxV*

=!. 6 .)I( ( TDL - /

ANFIS : ' No: ';(o. ) : Wo. #o& o

o >o$ No: >oE6 o o '("5 : #

!" # $ %& ' ()* +,-. /0 0$ 456 ...

'& ';! !?(@ 13


NL" E .$ D0 WEN o +

!" # $ %& ' ()* +,-. /0 0$ 456 ...

14 '& ';! !?(@


nonlinear behaviour, Atmospheric Environment 34, pp. 2134, 2000. [30] Jehng-Jung Kao and Shang-Shuang Huang, Forecasts using neural network versus BoxJenkins methodology for ambient air quality monitoring data, Journal of the Air and Waste Management Association, Vol. 50, pp. 219226, 2000. [31] W.G. Cobourn, L. Dolcine, M. French and M.C. Hubbard, A comparison of nonlinear neural network models for ground-level ozone forecasting, Journal of the Air and Waste Management Association, Vol. 50, pp. 19992009, 2000. [32] A.L. Dutot, J. Rynkiewicz, Fredy E. Steiner and J. Rude, A 24-h forecast of ozone peaks and exceedance levels using neural classifiers and weather predictions, Environmental Modelling and Software, Vol. 22, pp. 12611269, 2007. [33] M.W. Gardner and S.R. Dorling, Artificial neural networks (the multilayer perceptron). A review of applications in the atmospheric sciences, Atmospheric Environment, Vol. 32, pp. 26272636, 1998. [34] A. Comrie, Comparing neural networks and regression models for ozone forecasting, Journal of the Air and Waste Management Association, Vol. 47, pp. 653663, 1997. [35] G. Spellman, An application of artificial neural networks to the prediction of surface ozone concentrations in the United Kingdom, Applied Geography, Vol. 19, pp. 123136, 1999. [36] U. Schlink, S. Dorling, E. Pelikan, G. Nunnari, G. Cawley, H. Junninen, A. Greig, R. Foxall, K. Eben, T. Chatterton, J. Vondracek, M. Richter, M. Dostal, L. Bertucco, M. Kolehmainen and M. Doyle, A rigorous inter-comparison of ground-level ozone predictions, Atmospheric Environment, Vol. 37, pp. 32373253, 2003. [37] U. Schlink, O. Herbarth, M. Richter, S. Dorling, G. Nunnari, G. Cawleyd and E. Pelikan, Statistical models to assess the health effects and to forecast ground-level ozone, Environmental Modelling and Software, Vol. 21, pp. 547558, 2006. [38] Sameer Sharma, S.V. Barai and A.K. Dikshit, Studies of air quality predictors based on neural networks, International Journal of Environment and Pollution, Vol. 19, pp. 442453, 2003. [39] B.M. Fernandez de Castro, J.M. Prada Sanchez, W. Gonzalez Manteiga and M. Febrero Bande, Prediction of SO2levels using neural networks, Journal of the Air and Waste Management Association, Vol. 53, pp. 532539, 2003. [40] A.B. Chelani, C.V.C. Rao, K.M. Phadke and M.Z. Hasan, Prediction of sulphur dioxide concentrations using artificial neural networks, Environmental Modelling and Software, Vol. 17, pp. 161168, 2002. [41] M.W. Gardner and S.R. Dorling, Neural network modelling and prediction of hourly NOx and NO2concentrations in urban air in London, Atmospheric Environment, Vol. 33, pp. 171176, 1999. [42] Faizal A. Hasham, Warren B. Kindzierski and Stephen J. Stanley, Modeling of hourly NOx concentrations using artificial neural networks, Journal of Environmental Engineering and Science, Vol. 3, pp. 111119, 2004. [43] C. Cappa, D. Anfossi, M.M. Grosa and P. Natale, Short-term prediction of urban NO2 by means of artificial neural network, International Journal of Environment and Pollution, Vol. 15, pp. 483496, 2005.

[44] M. Kolehmainen, H. Martikainen and J. Ruuskanen, Neural networks and periodic components in air quality forecasting, Atmospheric Environment, Vol. 35, pp. 815825, 2001. [45] Harri Niska, T. Hiltunen, A. Karppinen, J. Ruuskanen and M. Kolehmainen, Evolving the neural network model for forecasting air pollution time series, Engineering Applications of Artificial Intelligence, Vol.17, pp. 159167, 2004. [46] S.M. Shiva Nagendra and Mukesh Khare, Artificial neural network based line source models for vehicular exhaust emission predictions of an urban roadway, Transportation Research Part D, Vol. 9, pp. 199208, 2004. [47] J. Hooyberghs, C. Mensink, G. Dumont, F. Fierens and O. Brasseur, A neural network forecast for daily average PM10 concentrations in Belgium, Atmospheric Environment, Vol. 39, pp. 32793289, 2005. [48] J.B. Ordieres, E.P. Vergara, R.S. Capuz and R.E. Salazar, Neural network prediction model for fine particulate matter (PM2.5) on the USMexico border in El Paso (Texas) and Ciudad Juarez (Chihuahua), Environmental Modelling and Software, Vol. 20, pp. 547559, 2005. [49] G. Grivas and A. Chaloulakou, Artificial neural network models for prediction of PM10 hourly concentrations, in the Greater Area of Athens, Greece, Atmospheric Environment, Vol. 40, pp. 12161229, 2006. [50] A. Zolghadri and F. Cazaurang, Adaptive nonlinear state-space modelling for the prediction of daily mean PM10 concentrations, Environmental Modelling and Software, Vol. 21, pp. 885894, 2006. [51] T. Slini, A. Kaprara, K. Karatzas and N. Moussiopoulos, PM10 forecasting for Thessaloniki, Greece, Environmental Modelling and Software, Vol. 21, pp. 559565, 2006. [52] W. Wang, W. Lu, Xiekang Wang and Andrew Y.T. Leung, Prediction of maximum daily ozone level using combined neural networks and statistical characteristics, Environment International, Vol. 29, pp. 555562, 2003. [53] J-S.R. Jang ANFIS: Adaptive-Network-Based Fuzzy Inference System. IEEE Trans. on Systems, Man and Cybernetics, Vol. 23, pp. 665-685, 1993. [54] P. Werbos, beyond regression: New tools for prediction and analysis in the behavioral sciences, PhD thesis, Harvard University, 1974. [55] J. S. R. Jang, C.-T. Sun. and E. Mlizutani, Neurofuzzy and soft computing: a computational approach to learning and machine intelligence, Prentice Hall, 1996. [56] S. Shah, F. Palmeieri, and M. Datum, Optimal filtering algorithms for fast learning in feedforward neural networks Neural Networks, Vol. 5, 779787, 1992. [57] S. Singhal and L. Wu, Training multilayer perceptrons with extended kalman filter algorithm, In Davis S. Touretzky, editor, Advances in neural network information processing systems I, pp. 133-140. Morgan Kaufman Publishers, 1989. [58] D. Simon, Training fuzzy systems with the extended Kalman filter, Fuzzy Sets and Systems, Vol 132, 189 199, 2002. [59] M. S. Chen, A Comparative Study of Learning methods in Tuning Parameters of Fuzzy Membership Functions, IEEE Conf. on Sys. Man and Cybernetics, 1999. [60] D. B. Parker, Optimal algorithm for adaptive networks: Second order back propagation, second order direct

!" # $ %& ' ()* +,-. /0 0$ 456 ...

'& ';! !?(@ 15


propagation, and second order Hebbian learning, In proc. Of IEEE International Conference o n Neural Networks, Pages 593-600, 1987. [61] A. Chatterjee , K. Watanabe, An optimized Takagi-Sugeno type neuro-fuzzy system for modeling robot manipulators, Neural Comput & Applic , Vol. 15, No. 1, 55 61, 2006. [62] C. Karakuzu, Fuzzy controller training using particle swarm optimization for nonlinear system control, ISA Transactions Vol. 47, 229239, 2008. [63] S. E. Fahlman, Faster-learning variations on back propagation: an empirical study, Proc. Of the 1988 Connectionist Models Summer School, Pages 38-51, Carnegic University, 1988. [64] R. L. Watrous, Learning algorithm for connectionist network: applied gradient methods of nonlinear optimization, In Proc. Of IEEE International Conference on Neural Networks, pages 619-627, 1991. [65] R. Babuka, H. Verbruggen, Neuro-fuzzy methods for nonlinear system identification, Annual Reviews in Control, Elsevier, pages: 73-85, 2003. [66] C. F. Juang, A TSK-type recurrent fuzzy network for dynamic systems processing by neural network and genetic algorithm, IEEE Trans. Fuzzy Systems, Vol. 10,No. 2, pp. 155-170, April, 2002. [67] C. Juang, Y. Liou, TSK-Type Recurrent Fuzzy Network Design by the Hybrid of Genetic Algorithm and Particle Swarm Optimization, IEEE International Conference on Systems, Man and Cybernetics, 2004. [68] Clerc M., Kennedy J., The Particle Swarm Explosion, Stability, And Convergence in a Multidimensional Complex Space, IEEE Trans. on Evolutionary Computation, Vol. 6, No. 1, Feb 2002. [69] Kennedy, J., & Eberhart, R. C., Particle swarm optimization. In Proceedings of the IEEE international conference on neural networks, 1995. [70] Heppner, H., & Grenander, U., A stochastic non-linear model for coordinated bird flocks, AAAS, 1990. [71] Eberhart, R. C., & Kennedy, J., A new optimizer using particle swarm theory. In Proceedings of the sixth international symposium on micro machine and human science, 1995. [72] Eberhart, R. C., Simpson, P. K., & Dobbins, R. W., Computational intelligence PC tools. Boston: Academic Press, 1996. [73] M. Aliyari. Sh, M. Teshnehlab, A. K. Sedigh, A Novel Training Algorithm in ANFIS Structure, American Control Conf., 2006. [74] M. Aliyari. Sh, M. Teshnehlab, A. K. Sedigh, A Novel Hybrid Learning Algorithm for Tuning ANFIS Parameters using Adaptive Weighted PSO, IEEE Int. Fuzzy Sys. Conf., 2007. [75] M. Aliyari. Sh, M. Teshnehlab, A. K. Sedigh, Identification using ANFIS with intelligent hybrid stable learning algorithm approaches, Neural Comput. & Applic, 2008. [76] Fujita, S.; Tamura, H., Development of a Short-term Prediction Model for Predicting Photochemical Oxidants in a Local Area, SICE-ICASE, 2006. [77] Castro, J.R.; Castillo, O.; Melin, P.; Rodriguez-Diaz, A., A hybrid learning algorithm for Interval Type-2 Fuzzy Neural

Networks in time series prediction for the case of air pollution, NAFIPS 2008. [78] Tanaka, K.; Sano, M.; Watanabe, H., Modeling and control of carbon monoxide concentration using a neuro-fuzzy technique, Fuzzy Systems, IEEE Transactions on Volume 3, Issue 3, Aug. 1995 Page(s): 271 279. [79] Sungshin Kim; Jaeyong Kim; Chong-Bum Lee; Min-Young Kim, Fuzzy decision support system to the prediction of ozone concentrations, ISIE 2001. [80] Morabito, F.C.; Versaci, M., Prediction and estimation of atmospheric pollutant levels by soft computing approach, IJCNN 2001. [81] Tanaka, K.; Sano, M.; Watanabe, H., Identification and analysis of fuzzy model for air pollution-an approach to self-learning control of CO concentration, Industrial Electronics, Control, Instrumentation, and Automation, 1992. [82] Negnevitsky, M.; Kelareva, G., Air quality prediction using a neuro-fuzzy system, Fuzzy Systems, 2001. [83] Yilmaz Yildirim, Mahmut Bayramoglu, Adaptive neuro-fuzzy based modelling for prediction of air pollution daily levels in city of Zonguldak Chemosphere, Vol. 63, pp. 1575-1582, June 2006. [84] Francesco Carlo Morabito, Mario Versaci, Fuzzy neural identification and forecasting techniques to process experimental urban air pollution data Neural Networks, Volume 16, , pp. 93-506, April-May 2003. [85] M. Aliyari. Sh, M. Teshnehlab, A. Khaki. Sedigh, Climate short term prediction of air pollution using multi layer perceptron neural networks, Third Regional & First National Conf. on Climate Change, 2003. [86] M. Aliyari. Sh, M. Teshnehlab, A. Khaki. Sedigh, Short term prediction of air pollution using TDNN, Conf on Intelligent Systems CIS2004. [87] M. Aliyari. Sh, M. Teshnehlab, A. Khaki. Sedigh, Short term prediction of air pollution using MLP & Gamma neural networks , UKACC2004.

8 . -%L -

\1 :6 ' x$ \=$ N:#x$ Co

!" # $ %& ' ()* +,-. /0 0$ 456 ...

16 '& ';! !?(@


\2: #x$ :6 ' x$ \=$ N PM-10

\3: E P N L Co

\4: LE P N PM-10

)(n #@ ':. (S = WEN ((

\5: N! 'N N Co

\6: N! 'N N PM-10

\7: . 0$ /0 (

\8: , 0$ /0 >$MLP

\9: Co MLP1

!" # $ %& ' ()* +,-. /0 0$ 456 ...

'& ';! !?(@ 17


\10: PM-10 MLP1

\11: Co MLP2

)' NN 6,, W&. FN ' NN ppm , 3mg(

\12: PM-10 MLP2

)' NN 6,, W&. FN ' NN ppm , 3mg(

\13: . '$ ' 0.

\14: 456 0.

\15: 456 G. (

\16: 5 (FIR .

!" # $ %& ' ()* +,-. /0 0$ 456 ...

18 '& ';! !?(@


\17: 456

\18: Co TDL

)' NN 6,, W&. FN ' NN ppm , 3mg(

\19: PM-10 TDL

)' NN 6,, W&. FN ' NN ppm , 3mg(

\20: Co

)' NN 6,, W&. FN ' NN ppm , 3mg(

\21: PM-10

) NN' NN 6,, W&. FN ' ppm , 3mg(

\22 : ( 'NANFIS

\23: Co ANFIS(PSO+EKF)

)' NN 6,, W&. FN ' NN ppm , 3mg(

\24: PM-10 ANFIS(PSO+EKF)

)N' NN 6,, W&. FN ' N ppm , 3mg(

!" # $ %& ' ()* +,-. /0 0$ 456 ...

'& ';! !?(@ 19


,1 :JE ' ;, E(C Co AR MLP1 MLP2 TDL Gamma

RMSE 3287/2 4960/0 4446/0 4159/0 4003/0NMSE 8168/0 1194/0 0959/0 0786/0 0729/0

MA 0454/1 3860/0 3278/0 3006/0 2898/0

,1) :(JE ' ;, E(C

Co ANFIS (GD+RLS) ANFIS

(PSO+GD) ANFIS

(PSO+RLS) ANFIS

(PSO+EKF)

RMSE 4048/0 4236/0 08159/0 0734/0NMSE 0743/0 0825/0 0051/0 0034/0

MA 2961/0 3143/0 0983/0 0882/0

,2 :m ';, E(C?P # PM10 AR MLP1 MLP2 TDL Gamma RMSE 6211/26 5388/14 9112/16 5487/13 5472/13NMSE 5764/0 2128/0 2880/0 1915/0 1915/0

MA 3643/20 0029/10 0472/12 5925/9 5920/9

,2 ) :(?P #m ';, E(C

PM10 ANFIS (GD+RLS) ANFIS

(PSO+GD) ANFIS

(PSO+RLS) ANFIS

(PSO+EKF)

RMSE 9738/12 5879/14 9365/4 8831/3 NMSE 1215/0 2643/0 0537/0 0273/0

MA 7631/8 6875/11 5673/5 4365/4

1387 2 1=X@ 25-20

I S ICE


2! 9M )J(6 NJ O (. ( G!, 5 '(0 .

+;$ L

E CD$ (':5 FE. B. '$- ' G!, 5 '(0 .


#o&iR o! Wo!6 F(oL! o oE(C o@5 F(L! 0m 0 'E . , @5 F(L!

(E +" o$ V$ @5 %C N #&iR ' N ao o 'oN Jog , Wo. oN +" +. (I + N

W. 0m , +" ( 'P]5 .[ o o@5 J(o$ o.=! (I F + B. Wo. o%C o$ . ooE* 'o. o( + S ', o No: 'o

E '-. %C +. ' N: , o o >o(C$ +o$ 0 ," Wo. Wo. o o #oR$ '0p$ : .

o o5 o4N _o@ o( o! ' aN ';(:. W. . ! o( 'N L voVN oE +o.5 'o o

o o! +" o s _@ , E o4 oCN 'o , o . , WN $ .! W5 o , Fo; !poE o( IoN o

W. $ . F(L! )P o( : oE +.5 ,i& (E '

F4$ $ IN ;! ', F6 Jo aIE J( Wo. o 5$ aIN . o o o o (oN" :

#,X IN %C , , W. x qi o aINo ; (E (" $ ( qi ( 5$ .

( o aN ':5 FE. ( 5$ o o o 0 _V /( o . 9, Jo( /o.$

$& +$2 ; W& IN (" , (Io5 ho >. (

3 . L (6 NJ FE. o. o o ':o5 \oS : (o \o& NoEN o$, oN B

. , $ \S & : 'E N . o IN , , WPDS a& ( #o&iR 'o. o( +o$

1 Queue Size2 Throughput

FoE. Jo* oN kVo. o0 5 : WoS (o +

CD$ (E':5 FE. B. '$- ' G!, 5 '(0 .


3.1 .(6 NJ

No k. 9, : D ':5 FE. 1 o o5 o;W. . #& N k. 9, and , 'o Fo Bo. o

\oCO Io 9, . + ':X( \&2 o #o@ (no . FoE. + (: #@ N k. :

)))(,max(min( ZW for all z >$ J( , ',, W(w& h$ &)(ZW h$

Woo. oo,0 W(oow& . Ioo5 aooN oo oo oo4N oo ':oo5 FooE.MATLAB ':. , 6R.

3.2 .)?' K w& h$I(. ';(,, s W( _@ ,0 , ! 0p$ ,

No voVN o W(w& h$ L ! (I . ho$ Jo oD\ ,0 , ;(,, W(w&)2 (W. ,".

\2- ,0 , ;(,, W(w& h$ J D

\ )3 ( ,0 , ;(,, W(w& h$ $ (oN o5 N.

1 Mamdani Inference Method 2 Centroid Defuzzifier

)_!(

)v(

)k(

\3- )_! ( _@ I(. W(w& h$)v (0p$ W(w& h$)k ( (I W(w& h$ !

3.3 .(6 %'R '3 o kVo. 0 5 ':5 FE. X. &S

W.]2 .[, &S J()1 ( ," oN . o;X( ,o Jo( (N ! 0p$ +I ;N. , _@ I(. . FoE. o,0 , \0 W.;! (I ':5 N.

4 .N)2Q" $SIL

4.1 . !2 TU +& , ;(:. o NN ']2 [ ,]6 [ Xo.

\o o o5 oS (o:" o oN)4 ( ,)5 ( Fo.$ N.

3 Rule Base

,1- 5$ ' ':5 &S 0p$

_@ I(. 0* * /. I

0I

0* F 0 F 0 F F /.

* F 0 F F /. /. /. F F /. /. (: I F /. /. (: (: 0I (: 0 (: 0

0(:

0(:

0(:

E CD$ (':5 FE. B. '$- ' G!, 5 '(0 .


\4- (:" 8 N

\5- (:" 14 N

';! N {& , 1.5MBPS h(:$ 0p$ ,

!1 1.5 W. 5 4N NO . I(o. F(I 'o Jo F _@50 vE= W( . Fo ':o. o +o: \o30 oNO

.! 0p$ .= ' E : o Xo. E*

CD$ (E':5 FE. B. '$- ' G!, 5 '(0 .


Wo. o ;, C : (:" , IN +: J( , N h(. CN , h(:$ ':5 FE. WS J( N..

,2 _V ';, ' " W. $& +$1

E I(. )W(( _@ $& +$$(;(%) $& +$

0.8_@+0.20p$(%) $& +$

0.5_@+0.50p$(%) ':5 $& +$(%)

800 59.16 83.91 96.34 94.22 900 51.78 76.63 82.64 90.21

1000 47.91 71.66 76.54 82.21 ,3 _V ';, ' " W. $& +$2

E I(. )W(( _@ $& +$(;$(%) $& +$

0.8_@+0.20p$(%) $& +$

0.5_@+0.50p$(%) ':5 $& +$(%)

900 55.56 81.82 95.55 95.16 950 56.46 82.93 91.79

1000 53.33 77.68 87.97 88.21 ,4 _V ';, ' %C +. +: /. 1

E I(. )W(( +. +: /.(;$ _@)NO( +. +: /.

0.8_@+0.20p$)NO( +. +: /.

0.5_@+0.50p$)NO( +. +: /.

':5)NO( 800 1.67 1.32 0.91 0.76 900 1.58 1.26 1.14 0.86

1000 1.75 1.47 1.33 1.03

,5 _V ';, ' %C +. +: /.2

E I(. )W(( +. +: /.(;$ _@)NO( +. +: /.

0.8_@+0.20p$)NO( +. +: /.

0.5_@+0.50p$)NO( +. +: /.

':5)NO( 900 2.01 1.34 0.83 0.78 950 1.92 1.02 0.79

1000 1.97 1.37 1.09 0.81

5 .( < (E !C J( ( : X. '$- '

W5 S . ':5 FE. . /(o o4N o ':o5 FoE. _V +o$ ;o, (o. o oE(C o _oV D. , '

., $) $& +

E CD$ (':5 FE. B. '$- ' G!, 5 '(0 .


[2] A. Pasupuleti, A. V. Mathew, N. Shenoy and S. A. Dianat, "A Fuzzy System for Adaptive Network Routing", Proceedings of SPIE Vol. #4740, SPIE's 16th Annual International Symposium on Aerospace/Defense Sensing, Simulation, and Controls, Orlando, Florida USA. 1-5 April 2002

[3] J. Boyan and M. L. Littman, "Packet Routing in Dynamically Changing Networks: A Reinforcement Approach", Advances in Neural Information Processing Systems, volume 7, pages 671-378, 1994.

[4] Bruce S. Davie, Larry L. Peterson, Computer Networks: A Systems Approach Morgan Kaufmann,second edition, June 2000.

[5] D. W. Glazer and C. Tropper, "A New Metric for Dynamic Routing Algorithm", IEEE Transactions on Communications, Vol. 38, No. 3, March 1990.

[6] M. Kara, H. Karabelli & N. Duru, "Fuzzy Based Routing In Packet Switching Networks", International XII, Turkish Symposium on Artificial Intelligence and Neural Networks TAINN 2003.

[7] Chiang Kang Tan, "The Use of Fuzzy Metric in QoS Based OSPF Network",

http://www.ee.ucl.ac.uk/~lsacks/tcomsmsc/projects/pastproj/ck_tan.pdf, 2000/2001.

1387 2 1=X@ 39-26


(6J"% !I)! (! Y2 M1Z - O ,#Q"

!& [ %

:!0 61 NO ' PE 2

. 5%$ '$ : X. W. + 4 J .V$ ( J ,0 'L ! ( , 5%$ M;L ' (I h$ $ 5%$ M;

+" & , 6R ' $N^$ FE. IN , ,0 5 L! W!6 5 L! 55( FP$ J!W!6 LV$ +P (C E.V$ E(C 5%$ ; E C6 FE. .L J!

W. 5%$ M; ! ') ] J #N.N D ( >DS ( .

' ' ^,: 5( FP$ J! 5 +$ + ' $N^ $ ,0 , W!6 5 L!5%$ M;

Abstract: This paper presents nonlinear modeling, simulation, and controller design for wind energy conversion system in Dizbad power plant. The wind power plant scheme consists of a three-phase induction generator that connected to the wind turbine via a fixed ratio gearbox. A static VAR compensator was connected at the generator terminals to regulate its voltage. The induction generator was connected to the utility through a double transmission line. The mechanical power input and terminal voltage output was controlled using blade pitch angle and firing angle of thyristor. An optimal stochastic control is designed for the system based on the optimal stochastic estimator. The performance of the proposed scheme is compared with both output feedback and state feedback controllers with extended Kalman filter. These controllers are applied to the nonlinear system and their performances are recognized under gust and different type of disturbances. The responses of the proposed optimal stochastic controller exhibited a good damping and fast recovery under these disturbances.

Keywords: State and Output Feedback Controllers, Wind Turbo Generator, VAR Compensator, Extended Kalman Filter, Stochastic Optimal Control.

________________________________________________________________

':E! I( ' J$ D0K , B. +"L! 5%$ M; 5 :!0 6 NO ' PE 27


1 .%9 oD0K FoE. +: W&. '(nx$ W W&

\o qio ',,

o FE. W0( K o . Jo( \o6 'o NoN 'o 9,W. ;

28 ':E! I( ' J$ D0K , B. +"L! 5%$ M; 5 :!0 6 ' PENO


.'+

/21 2VARCPT p

R

mm

(3)

o(,: o W&o. \& . ' J$ N #S (M W&. , s$ $ ' (,: . +$ >(| #x$ (: \

W. +(| #x$)(. _V :5 '(,: '

2.3. )9" 2

':E! I( ' J$ D0K , B. +"L! 5%$ M; 5 :!0 6 NO ' PE 29


dc

qccb

ds

qs

b

b

ds

qs

ii

XVV

VV

++

+0

0

(16)

M(,: \ W( E($ '90 . Mo(,: (I5 o +o( , o +$ d& $" o o . : Xo. o

.= (: #@ (C! +( @& (5 \=$.

eq

ss

LL X

VVX

i

/

%/%/ )](2sin[)(2 (17)

+" ooeqX , ooE($ ooP GNoo% Woo. oo$" Moo(,: . 'o= 'o. \ENX( #)Pqd, o \o Wo!6 'ow5

'Li N (: #@ +,. h L. :

ds

qs

Leq

b

dL

qL

b

b

dL

qL

VV

Xii

ii +

++0

0

(18)

5.3. 9$ S W!6 'w5 D0,I5 \o@6 o(:.

Eduzz 12 (21)

_(P$ (: #@ +" .

30 ':E! I( ' J$ D0K , B. +"L! 5%$ M; 5 :!0 6 ' PENO


1

2

3 3

0,

0,

00

,

,

111113

211113

112

11111313

12

111113

4

4

EB

DA

dtVVdtPP

Xz

tsrefgref

$ ,I5 'o 9, o o Wo!6 o5 L! ) \@6 (: \ >DS ( 6R .

zKu s (22)

G($o oE oC6 'o >oDS vVN sK o o \o@6 . o ,0y N (: #@ .

11313616 zCy (23)

+" y MD #@ (24) _(P$.

tsref

grefrssg

dtVV

dtPPIVPy

3

3

])(

,)(,,,[ + (24)

(: \ a( +

':E! I( ' J$ D0K , B. +"L! 5%$ M; 5 :!0 6 NO ' PE 31


JV$ z o Wo!6 J(o$ (IN , J(; W. $@ W!6z %$ , 0N 'I(N w6 5 . Wo!6 o

W. (: #@ !, '. 00 )0(,)0( zzpP

(26) W. (: #@ JV$ 'D0 GN( P G($ + :, :

PtzHRtzPHGQGtzPAPtzAP TTT ),(),(),(),( 1 (27) ( J DA ,H ) ';E($J^( o _o(P$ o(: #o@ o

W. :

66"

66!

z)t,z(h)t,z(H

z)t,u,z(f)t,z(A

(28)

'oD0 GNo( G($o +o UVo

32 ':E! I( ' J$ D0K , B. +"L! 5%$ M; 5 :!0 6 ' PENO


)()( 1 tSBRtK T (35) +" :

N: +C G($)(tS \o o $( M!P \6 : \o@6 o(: :

QSBSBRSASAS TT 1 (36)

M(I h$ , J( $ )( 0tj Mo; Wo!6 #@ )( 0* tj (":

3

T

t

TT dtGQSGtracetxtStxEtj0

21)]()()([

21)( 0000

* (37)

M; L. N )(* tu (: #@ W. (" : )()()( 1* txtSBRtu T (38)

'T . , #.= F6 W; I 5 :N 'F] W!6 WO : +$ L! &K Xo. o(: \

:

8

8 SBRKT1 (39)

+" 8S ,= \6 P JP W ,W. $( M! .)0( S )a`< "4

{5 W!6 J( Wo!6 o$ 'o)(tx \oS qoCS:N N ' . :oN o +" 'o o 'o)(tz o o(: Mo!P

: vHxtz )( (40)

+" ),0( vRNv 5 :N I(N o o o. 'o . Jo( L. W!6)(tu JV$ ;$)( tx :oN ', : 'o

ooo oooE, ooo \ooo@6 oooooo . ooo(I hooo$ ':ooo. ooo; ooo]}),([{)( 000 ttxJEtj v, 9, : X. ,4 >oC& ,

(& , JV$ (5 ' $( E($ !P . ,. (: #@ JV$ #)P : [11]

HPRPHGQGPAAPP vTTT 1)( 0tt 7

00 )( PtP (41) 1)( v

T RPHL (42)

)( xHzLBuxAx (43) +" P , D0 GN( G($)(tL , J! 5 ;)( tx

W!6 JV$W. . ; ; /( J( )(tK & ' )(tx JV$ L. )( tx , .

0 (: #@ D0 5 +NS: )()()( txtKtu (44)

#)P s )(tK #)P +(36),(35) 0.

4 Sweep method

':o (I h$ (I h$ +" 'nL( , l5 #)P : X. W. %6 \S (: \ ; JV$ , ; .

(45))(21

)(21)]()()([

21)(

0

0

0000*

3

3

T

tv

T

T

t

TT

dtKPRKtrace

dtGQSGtracetxtStxEtj

F4$ : , M W. m a:)LQ PDS \@6W. .W& (I (I5 LN a, M WPDS a& 9DS N. , )2( \= >DS +

':E! I( ' J$ D0K , B. +"L! 5%$ M; 5 :!0 6 NO ' PE 33


5 ; 'sK )>DS ( ', : " W. ( +$;V. E(C ,

34 ':E! I( ' J$ D0K , B. +"L! 5%$ M; 5 :!0 6 ' PENO


,)2 (C6 FE. ' >DS C6 FE. : , W!6 5 E5%$ ; 5 c)- )!9 NJ% J 94

!& c)- )!9NJ % J 94

"4 c)- )!9 NJ6 94

2356106.18 j 4.28884.23 j 4.28884.23 j

2356106.18 j 4.28884.23 j 4.28884.23 j

172848.17 j 4.213428.27 j 4.213428.27 j

172848.17 j 4.213428.27 j 4.213428.27 j

58.31324.26 j 08.37504.22 j 08.37504.22 j

58.31324.26 j 08.37504.22 j 08.37504.22 j

75.1969.3 j 37730 j 37734.0 j

75.1969.3 j 37730 j 37734.0 j

2456 59.914.37 j 59.914.37 j

46.61 59.914.37 j 59.914.37 j

22.7 40 42.0

586.0 5.5 0

442.0 8. 0

':E! I( ' J$ D0K , B. +"L! 5%$ M; 5 :!0 6 NO ' PE 35


\)7 (+: >E6 W&. #x$ N

\)8 (+: >E6 ' $N^ W&. #x$ N

\)9 ( +: >E6 ' $N^ W&. #x$ N)a,: (

\)10 (+: >E6 $N^ +$ , $. ^!, #x$ N

\)11 (+: >E6 $N^ +$ , ^!, #x$ N

) a,:(

36 ':E! I( ' J$ D0K , B. +"L! 5%$ M; 5 :!0 6 ' PENO


\)12 ( :5 M(,: , $" M(,: #x$ N FE. ' +: >E6 D0 K

\)13 (#x$ N FE. ' :5 M(,: , $" M(,: +: >E6 D0 K) a,:(

; , J! 5 & + ' 5%$ FoE. \ I(N !" E(C (: ' I(N !" D0K FE. J '

o5 o o o 5( FP$ J! JV$ >oDS o( o Wo!6) h[8] ( o; Wo!6 o5 FE. + , 5o%$ #o@

W. 5 .,0 k \ ',, , FE. ' oE C6 FE. ' ' \ )14 ( $)19 ( Wo. o +DS ( 5 o vD '( , ( ' . 'oR o

,6 F] W!6 #N.N +$ ' 25 % o $o. ^o!, 'o ,60 % , , #N.N 85 % Fo0 F] W!6 #N.N

W . (,: W&. ' Jg ,o6 F] W!6 #N.N $ '50 % Wo. 5( . ,o6 Fo] Wo!6 #No.N $o. +o( 'o40 %W. 5( .

',, #,X$ W. (: E . G($o + S W& N:,R L. ')(tu ',, #N.N Wo. o5( E . o'R $" M(,: ' 88 % o , , #N.N

97 %W F0 F] W!6 #N.N . Jo$ :5 M(,: ' Jg n . ( ,i&1 W. ; E , NO . F] W!6

( MDCN 6 #N.N IN 80 %W. 5( .

\)14 ( +: >E6 ' $N^ W&. #x$ N

\)15 (+: >E6 $. ^!, #x$ N

\)16 ( +: >E6 $N^ +$ #x$ N

':E! I( ' J$ D0K , B. +"L! 5%$ M; 5 :!0 6 NO ' PE 37


\)17 ( +: >E6 $N^ $. +( #x$ N

\)18 ( +: >E6 D0 K FE. $" M(,: #x$ N

\)19 ( >E6 D0 K FE. ' :5 M(,: #x$ N

+:

10 .< ( ':oE! !C J( oD0K , og FoE. o( : oCS oEN

J$ >- \o oI( 'o ,N $N^ o$ 'o J o$N^(C! !CN W5 +$ $. +:0 , CN /0 aN

D0K W!6 'w5 #)P j 13 o \oDS ( W!6 5 ! .

JoV$ o oE(C 5o%$ o; o o oE oC6 FE. .( J! Jo o o #No.N o oD ( >DS

! ') ]L W. 5%$ M; +o* W. N J( ( IN , ' L! vVN W. 5%$ ' X! ' P!D FE.

W. 5%$ ; '$.

KL ]1[ $ o= . (=D PoE o0 oC o( GNo5 ^o!, o

ox 0o. 9, o : \CE I( o $ * ,N. l .; GNX J

38 ':E! I( ' J$ D0K , B. +"L! 5%$ M; 5 :!0 6 ' PENO


e: 5 'x WE!:

qsds 00 , . E6pu

qrdr 00 , ' = '. $ E6pu

qsds ii , ' = '. $. +(qd,>E6pu

qrdr ii , ' = '. $ +(qd,>E6pu

qcdc ii , ' = '. +:0 +(qd,>E6pu

qLdL ii , ' = '. E($ +(qd,>E6pu

qTLdTL ii , ' = '. CN /0 +(qd,>E6pu

ss VI , >E6 $. +( , ^!,pu

refP >E6 (C! $N^ N +$pu

gP C! $N^ +$>E6 (pu

rs RR , >E6 $., $ W,Cpu

TLTL XR , >E6 CN /0 GN , W,Cpu

CL XX . GN , GN,NSVC >E6pu

qsds VV , ' = '. $. ^!,qd ,>E6pu

qbdb VV , B ^!, '= '. W(;Nqd ,>E6pu

refV $. N ^!,>E6pu

b+ N W&. (314rad/s)

r+ ' $N^ $ W&. (rad/s)

\($ ' ( (C(pu) ( +$ :KVA660 FE. a$ ' ( ^!, :V690 FE. a$ '

+$ +(SVC) : puX L 0.4 ,puX C 8.3

CN sD0 : W,C :puRTL 015.0 .GN :puX TL 15.0

I( ' J$ 5 #%V(| 1:56

' $N^ 5 #%V( 91/0 GRooox GNooo

(pu) 88/2 +,oo. W&oo.

(RPM) 1500

$. W,C(pu) 0063/0 $ W,C(pu) 0059/0 $. GN(pu) 0489/0 $ GN(pu) 0058/0

':E! I( ' J$ D0K , B. +"L! 5%$ M; 5 :!0 6 NO ' PE 39


,

TL

TLbb

TL

b

bTL

TLb

TL

b

bTCR

b

bTCR

b

cbcbbMcbMcb

cbcbbMcbrrcb

m

m

A

b

A

dsMb

A

qsMb

A

drMb

A

qrMb

drssrrb

mbMrrb

qrmbssssbMrrb

bMssbrrssb

b

bMssb

bssssb

XR

X

XR

X

X

X

XXD

XXD

XX

XXD

XXD

XX

THDH

XDH

XDH

XDH

XD

XRD

XRD

XRD

XRD

XRD

XRD

XRD

XR

A

++

+

+++

++

++

+++++

+++++

++9+9+9+9+

9+

+++

9++++

+++

+

++

++

00000000

00000000

000000000

000000000

000000

000000

00000022222

000000)(0

000000)(0

0000000

0000000

0

0

0000

00

00

,

0000

0

00000

02

00000000

%

%

-+

g

f

TH

B

mb

1387 2 1=X@ 48-40


4_ 6 2!1( 5-2 0 )!( 9) - JJ

& E(, S 1 = 2 + ; L J .]N" :$

6R \6 & #@ 6 + +" +$' (. L;E' KD0 N I.

^,' ' :$$' (' C$( EE$$ ' (' C$( .:5 ' $ .

Abstract: In this paper, an optimal Takagi-Sugeno fuzzy control is designed with a novel method called combined discrete and continuous action reinforcement learning algorithm (CDCARLA). The proposed method is implemented for a nonlinear system that is Cart-Pole system. Simulation results show that the proposed method has significant performance. The advantage of CDCARLA is that it does not need system dynamics as well as any other information of power system. It can be said that, this method will consider nonlinear features of power system. It is shown that CDCARLA method can be considered as one of the automatic design technique for designing of controller parameters.

Keywords: Continuous Ation Rinforcement Larning Atomata, Continuous Action Reinforcement Learning Automata, Takagi-Sugeno Fuzzy Control.

_________________________________________________________________

1.%9

;!. ' 0. ;E' :5 ' ' N,5 .;E' KD0 ,gN z%0 0 .

( . F; jN :;E' :5 ':5 ' $ jN W. . .. * ;E' :5 ' $ .( 0 . E;' KD0 :5 i0 N' N '

6R :5 ' $ .(. #)P W=$ FE aP \ D . ]1[ Tanaka ,Wang ,]2[ Sugeno ,

Tanaka 6R 9, :5 ' $ : X. .D0 #)P . N ] W=$ FE . ]3[ $]6[ .(

C=$;, #C' =$ W. " .(@ ( .= ;, J$ >]| , : .' D0 \ . W!6 #)P FE

W=$' , a,C +," W.(W. .;,' N I6R :5 ' $ 5 S X. .

N . ]7[ W;(:.' .:5 : BP uN," FE' N 6R L! /.$ (N^ FW. X. . ( ;

. , EE (C$ '( $ 9, /.$ . $ ':5 0 6R = E(, S & 41


F

:

x

l2

$$6R > PID 6R , 5 (W. X. . 9,( h , J' (; C J5x %$ : F' *

, >.' j,(6R 9, .i , N6R , #)P ( ' . W=$ FE

a:) . 9, n!' n' 6R $$N >. WE . L!' 6R ]( , 6 W. 6 , \ !C J

: X.(' C$( P$ >]| >. : EE J( : X. G-. ,(' C$( ; .( : +" C JP$ J

W. "( .(L! J' 6R . W=$ FE( ( . P N , LNN :' , #)P (;' N +" .

. BP uN," FE( 5,P :(. J;E' KD0 RK WPD0 Ni +& $ d& +" z0 ' .

] ;,' 6R 5 4N .;,' XV ' . W. ] BP uN," FE . 9,;

42 . , EE (C$ '( $ 9, /.$ . $ ':5 0 6R = E(, S &


\3 (w& h$(x WC5 W&. )21MF ,22MF(

\4 (w& h$(x W,: ( )31MF ,32MF(

\ 1) \5( w& h$(x W,: W&. ( ' )41MF ,42MF(

D $ )2 (:5 &S ' x b \%$ F1 .( J' 6R \( 80 x%$ P$ FN C J .

4. Q"( !C( 4_ !1 % 26(

9, ;

. , EE (C$ '( $ 9, /.$ . $ ':5 0 6R = E(, S & 43


\6 ($ (' C$( EE

,1- :' ,' x%$ F : x$ ,6# 1 ]45-,55-[2 ]35-,45-[3 ]25-,35-[4 ]15-,25-[5 ]5-,15-[6 ]5,5-[7 ] 15,5[8 ] 25,15[9 ] 35,25[

10 ] 45,35[11 ]55,45[

3.5 .a a : 6 ,!2-# %g - !i 6)!( 9) JJ

' ( , : ' x!L* h$ 65,,2,1)( infi WEN .!, C!L* h$

R 6( vVN W0 (P:

)3( 6"

6!

other

nnnnf lili0

,,2,11)()0(

iln : P$' P$( _' ,iW. a .$ (' C$( EE' n x , \ )6 (W. +. LN j .D0 h' PS,: , W(W. .(P:

33

' )7 ( ,)8 ( W. +. ( . )6(

essMxJJ

~

essx D0' ] W!6 PS , WM ( :N W & 5 i I1000M W..

3.3.5 . ,!2-# %9 Q)-2 6 ( I) ' x$ N"!L* h$ # +," W. >. 6

.C$ L(, : W' I( (: ( $ F: 1- I( ' C$ ) EN(N X@ W 2- I( ' ] ENC$ J( W(N 3- C$(I W( ' l5 6 , J , X@ J( .

LE EN WX@ C (I (S .( >. h$ :.C$ L( D W)7 ( .

)7(

66"

66!

#

>>

$

min

minmin

1

0

)(

JJ

JJJJJ

JJJJ

J meanmean

mean

mean

-

meanJ ,minJ $$ >LN , JI F( ' S W. . D E #@)7( D )8 ( .

)8( ?@A

"!

?@A

"!

min

,0max,1min)(JJ

JJJmean

mean-

4.3.5 .g"Q+ K 4 : ' $ 6 vVN :!L* h WEN 6

+ v0 (x$ vVN J ( ( P J( vVN J v06 P \6 +" vVN' ( < .(x$ J . : X.C$ L( W). N W=$ FE ( aN .

x$!L* h$ 6 D : X. )9 ( & . )9( ),()()()( )()1( inQJnfnf kiiki -%

h$),( inQ. ]N L : I$ i #@ , W. a)10 (P$( _ . )10( 11,...,2,1,2),(

2)(

ininQ in ' !L* h$ #)6 j N" 6 ( i% :

D)11 (P$ J .

)11(

11

1

)( ),()()(

1

n

ki

i

inQJnf -%

5.3.5 .!& jk r , : , W.' D)13 (x , : vVN .

)13( "!

>#>

)()1(

)1(1jCrjCj

CrIndex

ii

ii

. , EE (C$ '( $ 9, /.$ . $ ':5 0 6R = E(, S & 45


6.3.5 lk)g - 6 6 Y ; 6 ) N, $ ( x ,o o :o J o Io5 ( ( , o5

. 6( : ( .;' )9 ( $)13 (x$ N, ho$ #!L* x , b 6 +

46 . , EE (C$ '( $ 9, /.$ . $ ':5 0 6R = E(, S &


$$$ ' (' C$( N .$$' (' C$( \ W. EE)6 (W. +( ( . \' )14 ( $)17 (x$ N,!L* # | . 6( >

6R a, 6 : G +( F:)3a(

\15 (x$ N,!L* # 6| .(FIN >

\16 (x$ N,!L* # 6| .( > a, , WE

;(| C J(:5 >' $ C .' W.

. , EE (C$ '( $ 9, /.$ . $ ':5 0 6R = E(, S & 47


\18 (N(D0 x$I h$ #(

7 .2

48 . , EE (C$ '( $ 9, /.$ . $ ':5 0 6R = E(, S &


[2] K. Tanaka, M. Sugeno, Stability analysis and Design of Fuzzy Control Systems, Fuzzy sets and systems, vol. 45, pp. 135-156, 1992.

[3] G. Ren, Z. Hung-Xiu, Analytical Design of Takagi-Sugeno Fuzzy Control Systems, American Control Conference, pp. 1733-1738, 2005.

[4] M. Akar, . zgner, Decentralized Parallel Distributed Compensator Design for Takagi-Sugeno Fuzzy Systems, Proceeding of 38th Conference on Decision & Control, pp. 4834-4839, 1999. [5] A. Jadbabaie, M. Jamshidi, A. Titlie, Guaranteed-Cost Design of Continuous-Time Takagi-Sugeno Fuzzy Controllers via Linear Matrix Inequalities, IEEE, pp.268-273, 1998.

[6] K. Tanaka, T. Hori, H. Wang, New Robust and Optimal Designs for Takagi-Sugeno Fuzzy Control Systems, International Conference on Control Application, pp. 415-420, 1999. [7] M.Y. Shieh, C.W. Huang, T.H. Li, A GA-based Sugeno-Type Logic Controller for Cart-Pole System, IEEE, pp. 1028-1032. [8] M.N. Howell, M.C. Best, On-line PID tuning for engine idle-speed control using continuous action reinforcement learning automata, Control Engineering Practice 8 147-154, Elsevier Science Ltd., 2000.

[9] M.N. Howell, T.J. Gordon, Continuous action reinforcement learning automata and their application to adaptive digital lter design, Engineering Applications of Artificial Intelligence 14 54956, Elsevier Science Ltd., 2001.

1387 2 1 =X@ 55-49


L) kC L 2J ' m ]

,E0 ='1X@ N ' 2 N J03

1 L%N 0 $O .;!. =! J( : ,` 0 '

E. ; ( !E >& UV.W. 5 S . . ]6 [ ' FE.

W. vVN l5 !E R E +P D0K . ]7 [ ( : #4N & W; E. ( !E : '" ( 0.

(, (C W. X. x * FE. ( ` . ]8 [( !E E. X@ +N ( !E . ( (' FE. CN

' W. s$ x * . aN. ,]9 [(, B.' ;= JP$ ' , W.E6 (4N ' FE. ( >.

( >& UV

50 0N J 'X@ N ',E0 = >& UVE6 ' w5 ?C=$' #% +" W!6),,,(: DCBAG (N .w5' $;E(' \($ h$

CC6 ( , . ( 8RH +

0N J 'X@ N ',E0 = >& UV&f w5 (nN" >& ' 2L ' FE.)2 (: (' 121 , C&ww F 21 yy . D

P f &w5 (nN" >'2L (: & ?P W.

}..,:{ 2322 C&H& weafGwGLfD & _(P$D PR D ; . ( : J(I

>&f & D a N (nN"N$ W. W;(LV. P F; . J( ' a.(n" N."' +:,

& D _(P$ F . T)`2 . (N {5W \($ h$ G($qq w5 '

8RH LN"

2* sup Wf

Df&

I:4

(n" N."' & +:,D W. . N:, G($ , W. m \S W _(P$ 2 + J(

E w5 (nN" >& J(I .= $ :.' 2L ( EN5 : (n aN V! ., (: F! ) + .= W;'

*:4 ( .

N"2 . {52G X@' ,' =+j N , (I$' #%NMG 12 +" N F h$ (-

0 ,M 0 h$ ( E .N:, G($ Jg W 8RH N ' vVN013 G

WMG . LN":: 44 >*

+"

1

3

2

WMG:4

)3(

$o . wS ?D35-13 ]12 [(N" : N F h$ (- 0 n! W.

112

88

8 NNMMMG

52 0N J 'X@ N ',E0 = >& UV! N' ( 0. W!6 N ! L(, +" B W; N L.

W. >& jS, UV

0N J 'X@ N ',E0 = >& UV

54 0N J 'X@ N ',E0 = >& UV& UV

0N J 'X@ N ',E0 = >& UV

1387 2 1 =X@ 64-56


ID J( #@ _N! FCEK 9, : X. IN . ';. & + +. WS >. "

1 Ott2 Grebogi3 Yorke

IN! Z+E " FE. CD$ - !- J* /(D& = (| 'S J5 4N

' '6, a5 N ?5 n! l5 FE. . W!6 #)P D0 ;$ " FE.

IN! Z+E FE. +&- !- 5P (: D , J* .

66"

66!

zxyz

yxzxyxyx

38

)129()3528())(1025(

%%%

%

)1(

. ': l5 " FE]1,0[&% Wo. " 5 ' . Jo(S, FE.)8.0,0[&% N IN! & FE. o . oS,

8.0% ooo ooo! ooo& FoooE. ooo \(ooo$ ooo oooS, , ooo]1,8.0(&% N0 J* & FE. .

( ' _(P$ a b c ,d(: /, ?D :

66

"

66

!

38

1293528

1025

%%

%%

dcba

)2(

IN! Z+E " FE. W!6 #)P +$- !- #o% J* #)P)3 (N E(N: .

6"

6!

dzxyzcyxzbxy

xyax

)(

)3(

. : 'E #)oP : Xo. o o;N" o( 5 (;EIN! Z+E FE.- !- _@$ J* b /,

)2 ( ,)3 ( Wo. oD o \oS o ( +P . o5 o +oP _@$ IN! #)P ( W5 ( jS, aL; #). "

]15 ,16 .[ J( b o& ) m(4 ! FE. . a5 IN o o o& #o6 o >.

0 . N=$ N . : o( ';

58 IN! Z+E " FE. CD$ - !- J*(D& = !S , a , .

N"1: IN! Z+E FE.- !- +" W!6 #)P J* D)6 ( 4N W. F( . ';!L. vVN FE. J(

u1 ,u2 #%

"!

gzuu

2

1 0 )7(

+" g D ?D

xyxg 2 )8(

:, oP$ soCN o& o* o hoS, 5o%$ oDCN ( D FE.)4 ( ( N D .

$o: oD o o o +NoS )7( Lo. , o !2u o FoE. a, Wo!6 o!P D0K V n6 X,

;& . ! ,gN C )(1 cbg o FoE. a, , , !P$ ';!6 ':. Z+E + 0

F5 . o( FoE. 'o + JPN \! J!,?S C1gW. aPN . o FoE. 'ox : Xo. o +

'; P( gz,y,x, o 'ox w ,g~ >o$ /, )9 ( ,)10 ( _(P$ F.

xyw )9(

ggg 1~ )10 (

F( 4N (: W!6 #)P FE. G-..

"!

),,,(~),,,(

gzyxhgggzyxfw

)11(

oD o o _o@$ FE. ' ,)11 ( o(: _No! ho$ 'o(N vVN F(N.

22 ~21)~,( gwgwV )12(

h$ N: ?

IN! Z+E " FE. CD$ - !- J* /(D& = . . F4$ =N J( ;2k +o Wo #o@ X ) oX

60 IN! Z+E " FE. CD$ - !- J*(D& =

IN! Z+E " FE. CD$ - !- J* /(D& = $$ IN)7 ( $)12 (N +

62 IN! Z+E " FE. CD$ - !- J*(D& =

IN! Z+E " FE. CD$ - !- J* /(D& = !S n! l5 6 ,( #O . a,oC , ';

64 IN! Z+E " FE. CD$ - !- J*(D& =

1387 2 1 =X@ 73-65


(-6 !' ,! - O` W4 ( %)%L 5-2 S2. ( D .

1 Chirikjian

# J( ' WP@ X. +$+.$ [2,3] ' E ' X.[4]6 [6,5]

: , WXN ' !! .:[7]L /= (. #& [8] \ .:E '[5] # [9] ( J #& [10]

##0 .N (w5[11] (w5 ' LE( J0. ,[12] .

DCN +. 5@ ) ':" N ',: : E R J6 N JP E ', W6 ( , UV

N^ F(L! /.$ ) ':" N ',: & + ; , BP $. \6 ' ( 9, ]66 (X@ \! _. N: 5


, ,

N^ F(L! /.$ ) ':" N ',: & + ; , BP $. \6 ' ( 9, ] (X@ \! _. N: 5 67


) D iZ +, =iR a I $ \%X J EN !)iC ( ,iFN,' m' )D5( .

\2 :N( +! ' ,

N^ F(L! /.$ ) ':" N ',: & + ; , BP $. \6 ' ( 9, ]68 (X@ \! _. N: 5


\ 4 :N(P|, j, # W6 W

3.3. $2 !' ,! (i 9, V J % + ; , # W6 + uN

'^N # \@X , ,. D10 N , ^N + ' .:($@ $ #6 i0 ; \@X ,n N

jN ;N"; I . ':. ; \@X #6 i0 #@ N^ F(L! /.$ WE( S a$ +&

0 N^ F(L! + >. N ':. ; E , . ' 'w5 U%V$ +& W=$ 'L( 9, : \! J

X. \%X. \%X 5 4N \%X ' (w5 4 J( '

+, S ;$N: w5' W. JP$ . +, (,: .= ' +. ' # uNi W6 W; N: \%X :

5 4N UV. WS . ) ' ':" '

>. WS $ ( (I5 W P$ WE((" W. . $ .= =N L( ' 9, 9, J( #,X$ ;$ N: h

vVN (5 , . \6 ., \6 N: h$ .= ' S ( 6 ( \6 , (N ," S a$ (

N ' FEC$ N ," .h$ : D : , , N15 D : a, , ,16 (" W. . )51( )()( xhxF )16(

j

ji xgfxF

1max )()(

D 15)(xh D :1 .= . W. UV

69

Jou

I .

#

N^ F( 9

urnal of Control, 200

I5 aN " W.

0 ;N"

' _V ( U)\6( +(

9/0 15

2/0 NO

6/0 NO: :

':"

6 WU4 X)< + " #

W. ':. 2 4 `e-

L! & .1 ':" 6

, # #%V#%V$ '

. @5 ;N"

JP$ 6R\6 J( $ ,D J(

\6 ' ( 9,

1

' \6 a$ :. (, .=

}

Journal of Co

w5

; W!6, F.g , J

\ .( ' E

N^ F(L! /.

ontrol, 2008, V.2, N.1

JP$ +.

# W6 W+)J* /0 (.

N '^N + #

,

71

Jou

N^ F( 1

urnal of Control, 200

, \@X WPS 9,WSDF ,b (=N

6 ; +, # W

(L! /.$ ) '

08, V.2, N.1

1 ':" ) +@ #@

+ ; #OkgM 1 ,

N^ F(L! /.$ ) ':" N ',: & + ; , BP $. \6 ' ( 9, ]72 (X@ \! _. N: 5


7.4. M1 2 5-2 J)9 -WSDF (I5 $ # . @5 +I + D0

( (I5 ':" . ';, D0 +I . : \@6 b(N n)9, : K WSDF (W. >D J( '( IN .I : ( '(

! R ( , ':" (I5 D0 (I5 a& 9, J( . S \S ' v vVN ., W. +" >D J( \!

: ;N" @5 N vVN ( v N^ F(L! /.$ P C ) # . +& 1 N. '

.(':" N # /( x$ >D J( , \%X jN N x$. \ $ a -12 'E i0 W. UV

N^ F(L! /.$ ) ':" N ',: & + ; , BP $. \6 ' ( 9, ] (X@ \! _. N: 5 73


[5] Sastry S and Kohn M, Milli Robotics for remote minimally invasive surgery, Robotics and Automation systems, 1997. [6] Paljug R , Ohm T and Hayati S, The JPL serpentine robot : a 12 DOF system for inspection ,IEEE International Conference on Robotics and Automation, ,1995. [7] Paap k .l, Christaller T and Kirchner F, A robot snake to inspect broken building , IEEE Int,Conf on Intelligent Robots and systems, 2000. [8] Wolf A, Brown H.B, Casciola R, Costa A, Shammas E and Choset H , A mobile hyper redundant mechanism for search and rescue tasks ,International Conference on Intelligent Robots and Systems, Las Vegas , Nevada, 2003.

[9] Gravange A and Woodfine R, mine-sniffing robotic snakes and eels, IEEE Int Conf on Robotics and Automation, 2000. [10] Suthakorn J, Chirikjian G S, A new inverse kinematic algorithm for binary manipulators with many actuators, Advanced Robotics, Vol. 15, No. 2, pp. 225 244 2001 [11] Fromherz M P J, Hogg Tad, Shang Y, Jackson W B, Towards constraint-based actuation allocation for hyper-redundant manipulators, Published in IJCAI-01Workshop on Modelling and Solving Problems with Constraints, Aug. 2001. [12] Deb.k , An introduction to genetic

Documents

87Vol.2. N.1,Fall2008, control engineers magazine, iran