FUZZY LOGIC CONTROL SYSTEM FOR DEVELOPING EXPERT SEA TRANSPORTATION

Transcription

1 FUZZY LOGIC CONTROL SYSTEM FOR DEVELOPING EXPERT SEA TRANSPORTATION Auli Siti Aisjh, A A Msroeri 2 Engineering Physics Dept., Industril Fculty of Technology, ITS Jurusn Teknik Fisik, ITS, Gedung E Lnti II, 60 2 Mrine System Engineering Dept., Fculty of Mrine Engineering Jurusn Teknik Sistem Perkpln, ITS, 60 ulis@ep.its.c.id ABSTRACT For chieving integrted modern se trnsporttion, which is consist of rectifying of shipping in ccording to trffic demnd, regulte of shipping mngement, sfe in se nd lso expnding the industril of designing of ship. One of these strtegies is designing ship monitoring nd control system for chieving sfe in se trnsporttion. This pper propose designing expert control system which is integrting monitoring system for course, position, trcking nd ship mneuvering. The controller system is consist of modules controller, there re : (i) course controller, (ii) position controller, (iii) speed controller, (iv) control of voiding collision, nd (v) controller for rejection se disturnce. In the erly reserch hve done designing module controller for course keeping nd trcking. System designing se on fuzzy logicl, in term if... then... The module of fuzzy logic controller (FLC) is composed some rules. These rules re express the est rule, which is chieved from the input nd output nother controller tht is LQG/LTR (Liner Qudrtic Regultor / Liner Trnsfer Recovery) controller which is roust. Composing the rules re sed from Sugeno - Tkgi methods, in then... ction from lest squre estimtion. The good performnce of FLC design is showed in mny simultions, like in wve disturnce when the significnt high is 3 meters, nd the setting heding is 30 o. The response controller giving settling time 279,34 seconds. Wheres in complex disturnce (significnt wve height is 3 meters), when setting is fulfilling the dot point in erly (0,0) meter nd trget is (6000, 3465) meters, the pth error is 7,308 meters. The sme strtegy is done t the controller proceed for the course keeping in 30 o (fulfilling the point (0,0) nd (6000, 3465) meters, the response of controller giving the pth error is 4,8 meters. Keywords: FLC, Se trnsporttion, course keeping, mneuvering, monitoring, trcking. INTRODUCTION Se trnsporttion needs of improving sfe in se nd efficiency of trnsporttion, these re most importnt. The progrm cn e done for incresing sfety in se nd reducing se ccidence, eside the regultion spect, the coordintion for reducing ccident, in writer ide is designing control nd monitoring system in se trnsporttion. The integrted monitoring nd control system in ship, is needed some unit controller, s control of ship speed, control of steering mchine for course keeping nd trck keeping, nd lso control for fulfilling the trjectory nd ship position. The monitoring system hve een doing from DGPS (Differentil Glol Positioning System), this component giving the informtion such s: position nd speed ship in few time, ccurte, chep nd in wherever, no depend on wether seson. The condition is needed for good controller is : (i) Hving stility chrcteristics, roustness performnce nd giving response s desired (ii) Adptive for the chnging the ship prmeters nd lso from disturnce environment (iii) Possiility implementtion in computer nd other supporting hrdwre. The wekness of former controller ws usully clled conventionl system, in which this system ws coming from mthemticl models. The mthemticl models is not cple ccommodte of ll system condition is controlled. In this mtter, the controller is do sed on feedck informtion of sensors (compss, gyrocompss), the specifiction controller hve not een doing in wide rnge, especilly in mesurement vrile is tend to zero or tend to infinite. This condition is occur when heding error, or pth of error is lest or very ig. But, this mtter is surpss with controller tht hs een doing se on dt se mngement in rel condition (the position, heding, speed, ship trjectory), nd dt is processed in numericlly for construct the rules. The rules re formed from rel dt, expertise, or intuition 67

2 68 The 5th Interntionl Conference on Informtion & Communiction Technology nd Systems of expertise. The output of controller is decision or ction of controller such s the position, speed, pth is recommended. 2. AN AUTOPILOT CONTROL SYSTEM An utopilot system in ship is occurring when compss nd gyrocompss is used in mneuvering ship control. Some of invention the models or control method in ship mneuvering, s figure elow, tht is clssifying in four methods : conventionl, dptive, modern nd expert methods. 2. Control System in Ship Mneuvering A control system is used for moving the ship steering s like s desire performnce, nd mny invention in this designing hve een trying in computer scle nd lso in minil. For exmple, is like trcking controller in un predictle se condition [Tee, Keng Peng, nd Ge, Shuzhi Sm, 2006, Velgic, J., et.l 2000]. This invention is conducted from former resercher in expert utopilot. Control systems t ship mneuvering sed on the technique designing, is distinguishle in four methods (Aisjh, 2007). The conventionl method is develop methods cuse finding n electricl gyroscope (Hopkins, 980). And then emerge mgnetic gyro compss which is sensitive in mgnetic noise. This founded is support close loop controller tht is nmed Metl Mike (Speery, 9). Minorsky (970) ws nlyzing the PID controller, this invention clled the utopilot, wheres this designing control system ws SISO (Single Input Single Output). The input of controller is deflection of gyro compss nd output is deflection of rudder. The former resercher in Liner Steering (Dvidson nd Schiff, 946, Nomoto, 957) nd Non Liner Steering (Akowitz, 964, Norrin, 970), hve een given ide in SISO Minorsky controllers. The conventionl designing ws done y some reserchers simply meet mny weknesses. It ws not fully ccommodte the existence of high frequency disturnce, lthough it hs een dding nd filter designed (Lozowicki nd Tino, 2000). The environment disturnces cuse the chnging of prmeters controlled system, nd on this reson so pper n dptive controller. In the dptive control system, mthemticl model like the one degrded in modified, so the dptive controller ws occurred. In this methods, the mthemtics models ws modified from the conventionl method nd ecoming n eqution with ccommodting the environment influence. This modifiction mthemticl model is expressed in so mny term, like MRAC (Model Reference Adptive Control) y Amorengen, V. (970), Blnke, (980) nd ARMA (Auto Regressive Moving Averge) y Nejim (998). The proposed methods y Amorengen (970) showed non liner rudder y Amorengen, V (984) nd he proposed commnd genertor for otined certinty model prmeters. While in Nejim structure showed the cpility of controller when the chnging of service ship speed. Figure : The developing system control design in ship mneuvering [Aisjh, A.S, 2007]. The invention of computer, cuse hppen the growth in designing of modern control systems. Where in designing this modern control systems, mthemticl models of mneuvering dynmics is expressed in the stte spce eqution. Modern technique ws proposed y Bertin (980) with strtegy ILQ (Inverse Liner Qudrtic) nd Kijim (2003), H2 y Donh (998), LQG (Liner Qudrtic Gussin) y Bertin (998), H~ y Strnd (998) nd lso Consegliere (2002), MPC (Model Predictive Control) nd LQR (Liner Qudrtic Regultor) y Whl et l (998). 2.2 Fuzzy Logic Controller in Ship Mneuvering Fuzzy logic theorem ws found y Zdeh in 965. This theorem imitte logic of humn common

3 C26 - Fuzzy Logic Control System For Developing Expert Se Trnsporttion - Auli Siti Aisjh 69 sense in decision mking t mtters, nd simply cn e pplied in controller t vrious households products for exmple wshing mchines, room refrigerting mchine, rice cooker nd others. The ppliction of fuzzy logic is not only growth in usge in this field, ut lso t some industril in Jpn which hve put it from the eginning 980 Th. And so do in mneuvering system which hve een proposed y Noguchi nd Mizoguchi (998) from Ishikw Jim - Hrim Hevy Industries, where mneuvering is mjored for sfe pth of ship. The position vrile nd sfe positions of ship expressed s fuzzy vrile. In the simultion is shown of ship mneuver performnce with fuzzy logic cn void existence of other ship. Vukic et l (998) designed of fuzzy logic control t mneuvering system y two wys tht re using 2 fuzzy inputs nd 3 fuzzy inputs. This mechnism developed y Aisjh, et l (2004), input of the FLC is 3 tht is yw nd error yw divide in 7 res of fuzzy memership function. The FLC worked sed on the output of PI controller (Proportionl - Integrl). In the simultion pplied t Mriner clss ship with first order of Nomoto. Aility of Vukic FLC s y 3 inputs it ws etter in so mny condition compred with Aisjh, et l (2004). Bsed on tht performnce the result hereinfter developed y Aisjh, et l (2005, 2005, 2005c, 2006) in designing of roust controller, through model of system in stte spce eqution. This wy pply to get otin detil nlysis from internl chrcter of good ship dynmics controller in or without disturnces. Lee, et l (2004) developed FAM (Fuzzy Associte Memories) of FLC which compiled sed on expert informtion nd experiment t rel condition. Fuzzy Controller is pplied in mesuring of heding, yw rte, reltive speed s to desire ship position (loction nd heding) in yield output control on the expected course. Input controller is reltive ship position offset to expected, heding errors, yw rte, projection of vector distnce s from center grvity to expected position, projection of reltive speed to the expected position. The developing of FAM in FLC sed on experiment result s hve done y Lee et l (2004), would ffect to the cpility of control systems, when ship operting in other condition. This prolem is nticipted y Aisjh, et l (2005c, 2006, 2006c), developing of FAM sed on output of roustness control systems tht is LQG/LTR controller. The LGQ/LTR controller is cple to overcome the externl disturnces. A fuzzy control system s is more simple in rule se ws proposed y Vukic, et l (998), s extending from FLC of Velgic, et l (998). This FLC uses 49 rules for output. The input of controller is yw error nd yw rte. The memership function of error yw nd yw rte re tringulr while output of signl of controller (commnd rudder) is trpezoid function. And then Omerdick, et l (2000) developed Vukic, et l (998) sme in their rule se, wheres the inference is Mmdni method s nd dding nd filter. The oject of Omerdick, et l (2000) in Mriner clss ship, the length is meters nd service speed s is 5 knots. The simultion implemented y entering the environment disturnces, the wves in Pierson - Moskowitz spectrum, modl frequency is rd/sec, nd se current is st order Guss Mrkov. The Notch filter is use for shfting wves. The performnce of control systems is settling time - Ts time is less thn 50 seconds, t the set point 0 o nd 30 o. While in disturnces, the response of Vukic FLC s is fluctutions. Development of rules using Sugeno-Tkgi lgorithms, which re expressed in mthemtic eqution s function of ll input vriles. This rule sed is coming from lest squre estimtion from the reltion of input nd output LQG/LTR s reference control. The input of LQG/LTR controller is yw error nd yw rte, while the output is signl commnd rudder. The mechnism like this ws designed y writer, nd simultion giving good performnce for developing module controller in monitoring nd control trnsporttion system. 3. Fuzzy Logic Controller FLC for Developing Expert Se Controller Trnsporttion. 3. The Development Se Trnsporttion The se models trnsporttion is compose with other system such s monitoring nd control system. This pper is proposes designing expert monitoring nd control tht is integrted one nd other. The expert system is using fuzzy logic. Designing Monitoring system In monitoring system, is system is done s monitor for position, speed, course nd the trjectory of ship, in the specify shipping re. The informtion is coming from monitor sttion with GPS s nvigtion system using stellite. GPS receiver is

4 70 The 5th Interntionl Conference on Informtion & Communiction Technology nd Systems otined signl from stellites in erth oriting. The mechnism of system designed is: GPS will give informtion out position nd the time. From two dt cn clculte the speed nd heding of ship, nd then continuing informtion to Monitor Sttion. The dt re storge in Monitor Sttion nd then for processing the controller. Designing Expert Controller The second susystem is fuzzy logic controller, the output this system is informtion out the heding, position, trjectory nd speed of ship. These informtion s recommended for collision voiding with other ship. This informtion is send to Controller Sttion nd cn ccess y user in ship. A. Informtion from sensors (heding / yw ngle, yw rte, position, ship speed) B. GPS Informtion (position of other ship / other foreign ody). C. Prmeters for developing Fuzzy Logic Controller.. Fuzzy Sets Clssifying The mechnism for clssifying dnger re (possiility collision) is done using grouping horizontl zone s suitle with ship heding ( o ). This grouping is like infollowing tle. Tle : Description of grouping horizontl zone Figure 2: Digrm lock of Monitoring nd Control System Bse on GPS informtion cn tken to e ccount for distnce (d) etween ships or to other ship. The vlue of d in linguistic term, there re: very close, close nd fr, or in other term, such s: in front, ehind, right nd left. The output of fuzzy I (figure 2) is decision out possiility ccident. In the logic vlue re : the Front Zone Accident (FZA), Bck Zone Accident (BZA), Left Zone ccident (LZA), Right Zone Accident (RZA), nd relted with confidence fctor : Not possile (NP), Possile (P) nd most possile (MP). The confidence fctor is in numericl vlue. The occurrence ccident is tken ccount in suitle rules, for exmple in the following rule: R: If d i is (LD (k) ) Then c j is (LC (k) ) () Figure 3: Fuzzy Control system rchitecture The prmeters re influence in designing control system:. The rchitecture of control system design. 2. Prmeters in ship dynmic : environmentl disturnces (ocen) The rchitecture of system will e design s drwing in following figure. In the figure 3, is shown t Fuzzy Logic Control rchitecture in this designing, there re prmeters re needed in this controller: Where: k is mount of rules, d i is distnce (is ccounted from GPS informtion), LD: linguistic vrile in the D set: (Very Close, Close, nd Fr), c i : direction of ccident (the ngle of heding collision, from compss mesurement), nd LC is C set vrile = is (Not Possily, Possily nd Most possily). And then to grouping dnger zone (possily occur collision), d is distnce to other ship in fuzzy vriles, nd weighting vlue of ccident collision is determined from result verifiction dt.

5 C26 - Fuzzy Logic Control System For Developing Expert Se Trnsporttion - Auli Siti Aisjh 7. A module voiding collision This module control is processes for voiding collision with other ship / other foreign vessels, is nvigtion free of collision. The gol of the controller design is trget course; this course is new course of ship. The rule is needed in this module for reducing heding error until to zero, which is the difference of trget nd ctul heding. Vriles Input module is: (i). Possiility ccident (LC) in linguistic vrile (NP not possile, P possile, MP most possile). (ii). Error of yw (Heding error) is expressed in linguistic vrile (LB Left Big, L left, LS left smll, Ze zero, RS right smll, R right, RB right ig). The output of module re two kinds, there re: (i). The chnging of heding / course (dψ) is expressed in linguistic vrile : (LF left fst, L left, LS left slow, Ze zero, RS right slow, R right, RF right fst). (ii). Surge velocity (u) is expressed in linguistic vrile (S slow, N norml F fst). The exmple of voiding collision rule is expressed: If c j is LC (k) And ψ is Lψ (k) Then dψ is LDψ (k) nd u is Ldu (k) (2) Where : k is mount of rules, c j is kind of j ccident, ψ is heding error, u is surge velocity, LC, Lψ, LDψ nd Ldu re expressed in linguistic vriles of c j, ψ, Dψ nd du. The k th rule in mthemticl expressed is reltion fuzzy R(k) to C x ψ, in domin fuzzy memership function : µ R (k) (c j, ψ) = min[µ LC (k) (c j ), µ Lψ (k) (ψ)] (3) The whole of rule se re expressed in union from ll of individul rule: K ( k ) µ R (c j, ψ,dψ, u) = µ ψ (4) k= R ( c, The output of nvigtion is expressed in term: µ * Nv * µ AND ( c j, ψ ), ( dψ, u) = mx min c j, ψ µ R ( c j, ψ, dψ, u) j ) (5) * Where: µ ( j,ψ ) input nd output? µ (, ψ, dψ u) AND c is comintion from R c j, c. A Control module ship dynmics In this module re composed two modules: there speed control module nd heding control module. Figure 4: Su system speed controller Input speed control module re (i) distnce to other ship (Zero, Close, Fr, very fr), (ii) Surge velocity (Slow, Norml, Fst), (iii) Surge velocity trget (Slow, Norml, Fst) nd (iv) Heding error ((Fst_Astern, Slow_Astern, Ded, Slow_Ahed, Fst_Ahed). One of those rules for speed controller is expressed in this term: If (distnce is fr) And (Surge_velocity is Norml) And (Trget_Surge_velocity is Norml) And (Heding_error is Norml) Then (RPM_of_propeller is Slow_Ahed) (6) d. Heding or Course Controller In su system heding controller, the input for module is heding error (the difference of desire heding to ctul heding / yw ngle). The environment fctor s disturnce (wve, se current nd wind) will influence heding error. The third disturnce fctor will e modified heding error, the chnging of yw rte nd distnce to trget point. The third inputs module re expressed in linguistic vriles : Heding error (NB - Negtive Big, N Negtive, Z Zero, P Positive, PB Positive Big), Yw rte (Ne - Negtive, No - Norml, Po - Positive), nd distnce (Z Zero, C Close, F - Fr). Output of heding controller is two kinds, there re: the voltge of thruster nd rudder ngle, re expressed in linguistic vrile (B Negtive Big, N Negtive, Z Zero, P Positive, PB Positive Big). And rule of two inputs in this module is shown y:

6 72 The 5th Interntionl Conference on Informtion & Communiction Technology nd Systems If (Heding_Error is Positive) And (Yw rte is Norml) And (distnce is Fr) Then (Thruster_Voltge is Positive) And (Rudder_Angle is Positive). (7) function of universe discourse is nlogous to level of proility mrginl from the vrile (Aisjh, et l, 2005d). Mny reserches in designing fuzzy control systems use tringulr function, ecuse simple esy to represent in numericl vrile (Kosko, 997). In Fig.7, F j i is fuzzy set, c i is rel prmeter, y k is output of systems in rule R (l), M is mount of fuzzy rules. From those rules, in the prt of IF in the form of fuzzy set, while prt of THEN is crisp vlue, is liner comintion of input vrile. Sugeno - Tkgi fuzzy control systems is shown in Figure 7. The x is input vrile, y is output vrile tht is result of Sugeno - Tkgi inference, w is ville weighting fctor from itertion of simultion. Figure 5. Structure of digrm lock Monitoring nd Control system for designing expert trnsporttion L () : IF x is F nd nd x n is F n THEN y = c 0 + c x + +c n x n w, y e. Control for reducing disturnce The disturnces re commonly from environment will e considered: wind, ocen current nd wves. In generl the third disturnces will influence heding of ship. Anlyzing those disturnces is derived in vlue nd ngle of coming from every disturnce [Aisjh, A.S, 2007]. In the ssumption tht wves is wind generted, so in the future module design just on two disturnces, there re wind nd ocen current. In these control modules, there re two position fuzzy controller, trget to X o nd Y o. Inputs of controller re ship position, trget position, Euler ngle, wind speed, wind ngle, velocity current nd current ngle. Projection of current nd wind will influence of dynmic nd surge velocity nd lso drg effect. The output of this controller is rpm propeller nd thruster voltge. 3.2 Designing Course nd Pth Control Module. Mechnism of FLC designing The mechnism in fuzzy logic control systems yielding decision (signl controlled) through some steps, which re fuzzifiction, mechnism in inference engine, nd defuzzifiction (Jmshidi, 993. In defuzzifiction is function to chnge crisp vrile ecomes fuzzy vrile y expressing in the memership function. This memership L M) : IF x is F M nd nd x n is F n M THEN y M = c 0 M + c M x + +c n M x n Weighting verge w m,y m Figure 6. Configurtion of Sugeno-Tkgi Fuzzy Logic Systems (Wng, 994). Model reference (LQG/LTR) is pplied to derive nd evlute of FLC performnce. Input of LQG/LTR controller s reference controller is yw error nd yw rte is pplied to otin unknown prmeters c ij nd c ij 2 which for rule se t FLC. At erly model reference is implemented eforehnd precede KLF, to otin the oth vlue c ij through lest squre estimtion, nd then FLC is implemented seprtely without LQG/LTR. Structure of FLC controller device consist of two units, in first unit sed on yw error (e) nd yw rte (r) input, second FLC unit working sed on normliztion of pth error. FLC structure in this reserch is different from other structure when it is used Mmdni inference function. At the time Mmdni inference is pplied, some roustness prmeters of controllers re not met. When simultion is done in some condition like s such s circle mneuver, the fluctutions is pper in complex environment condition. The error of rdius circle

7 C26 - Fuzzy Logic Control System For Developing Expert Se Trnsporttion - Auli Siti Aisjh 73 mneuvering is ±20% nd tends to not convergent (Aisjh et l, 2006d). To determine rule se of fuzzy controller which pplied is s follows: u = δ d + du + du 2 (8) in which 2 du = c e k + c r k (9) ij ij ( ) ( ) ij δ d is commnd signl rudder, c ij is gin signl controller tht is sed on chnge of error yw, c ij 2 is gin signl controller tht is sed on yw rte, du 2 is output of controller tht is sed on normliztion pth error η. To otin the gin in eqution of (9), it is done using estimtion lest squre. Input of 2 nd FLC unit is normliztion of pth error d, nd output of controller is du 2. When d is negtive hence du2 is negtive, nd so do on the contrry if d is positive, the vlue of du 2 is positive. The ville rule for this logic cn e expressed s form (0). ( η. ) du 2 = tnh s (0) S is sensitivity of FLC in which its input is η. In Figure of (8) is showing digrm lock of controller structures which is designed in this reserch. Mechnism of FLC tht is input is yw error (e) nd yw rte (r). In the course keeping simultion, output of FLC signl control is du feeding to steering mchine nd steering mchine is moving ship s ccording to heding setting. If simultion is done y setting heding from position ojects, hence FLC works with input yw error ( e), yw rte (r) nd pth error nd output of FLC signl controller is du + du 2. This signl is moving steering mchine nd then the rudder will move ships towrds the expected position. Wy point desire (x d, y d) FLC Bsed LQG/LTR Course desire Ψ d Disturnce Offset from desire pth Ψ r η Fuzzy Autopilot Offset Pth Autopilot du = c e + c 2 r du2 = tnh (sη) δ c Steering Mchine δ Ship Dynmics - Ψ d + LQG/LTR Controller Ship Dynmics Comps Giro Comps GPS x,y Ψ r Figure 7. Digrm lock of Sugeno Tkgi FLC with LQG/LTR s model reference for heding nd trcking control

8 Tle 2 Nottion usge in gin FLC controller The vlue of c ij, i = 5 nd j =..5 re otined from lest squre estimtion from output of LQG/LTR controller. stedy stte error (ess). These prmeters give stility chrcter of controller. Though those prmeter re not one kindness of controller performnce. To get stility prmeter like those, y the wy of through giving certin testing signl nd compring the result of the response to this signl testing. With test signl cn e nlyzed mthemticlly nd experimentl from control systems. Tle 3 Rnge of error yw, yw rte, pth of error nd output of controller. Figure 8. Heding response of Mriner clss ship when setting point is 30 o in UN disturnce condition 4. SIMULATION 4. Course Keeping Simultion At ccomplishment of course keeping (heding is constnt), controller will work if d ~ 0, here just 2 nd FLC unit is ctive. The gol of fuzzy controller is ring ship to the desire ngle of yw / heding the expected, in this cse when Δψ = ψ d ψ = 0. st unit of FLC is ctive with input e nd r, 2 nd unit of FLC is off, nd then ction of controller is du which is signl of commnd rudder δ c = du. At course keeping simultion is done y giving signl testing in the form of step function s n input of control systems. The first simultion, set point of heding is 30 o nd done t UN disturnce condition. The result of simultion like seen t Figure of 9 for Mriner clss nd Figure 0 for Tnker ( dwt). Seen t oth, the settling time (Ts) when set point heding is 30 o in LQG/LTR controller is 4.3 second nd 7.8 second in FLC. While Ts of Tnker ( dwt) is second (FLC) nd 40.9 second (LQG/LTR). This vlue is etter thn result of LQR controller ( Aisjh, et l, 2006). The performnce of controller in time domin, in generl mrked in prmeters: mximum overshoot (mov), settling time (Ts), rise time (Rs), Figure 9. Heding response of Tnker ( dwt) when setting point is 30 o in UN disturnce condition A first system order is expressed in the trnsfer function s following: C( s) K = () R ( s ) + Ts Which is K is gin of system, T: time constnt of system, C: output of system nd R: input of system (Ogt, 992). The response of control systems in some vessel types of level could e determined of time constnts system. The wy for finding this time constnt y when it reched 63.2 % from heding trget, this time is time constnt of system (Ogt, 992). For Mriner clss ship, the length is 6.9 meters, nd service speed is 7.72 m/sec nd Tnker (90000 dwt) is meters, 8.24 m/sec, the time constnt of oth type vessels re 07,89 seconds nd seconds. 4.2 Trck keeping Simultion At trck keeping simultion (pth ccomplishing), the gol of fuzzy logic control systems is ring ship from ctul position towrds to pth desired. 74

9 C26 - Fuzzy Logic Control System For Developing Expert Se Trnsporttion - Auli Siti Aisjh 75 If the ship in left side from the pth expected tht is the vlue of η > 0. This mens required signl commnd rudder is positive yielding trnsformtion of positive direction. Signl of du 2 will hve + nd will oost up signl commnd rudder, nd then cuse ship direction (ngle of yw) is igger thn which expected for few moment. At this impulse of ship towrds to the expected orit, so tht d will go down towrds 0. Mov = 3.4 o Stedy o Mov = o o Stedy o Muv = c Figure 0. The response of Mriner clss ship, in setting heding is 30 o in wve disturnce wind generted, () Hs = m () Hs = 2 m (c) Hs = 3 min the opposite, when the vlue of η < 0, in which the ship in the right side of desired pth, with symmetriclly in the rule t η > 0, so the vlue of du 2 is negtive nd reduce to signl commnd rudder, so e reducing the course of ship If ship in desired pth the vlue of η = 0, in this cse du 2 = 0, fuzzy utopilot will ctive in such yw error nd yw rte. This mechnism is clled course keeping controller. The input of control system is dot series of desired position (x i,y i ). In simultion in which is input is (0,0) meters nd (3465, 6000) meters in where this point s liner trcking is suitle with heding is -30 o. In elow figure is shown position of ctul nd desired pth ship, when in complex disturnces. The vlue of significnt height wve re,2 nd 3 meters, velocity of se current is 3 m/sec. (0,0) desire ctul (3465,6000) βc=30 o, γw=30 o Figure 2 Trjectory of Mriner Clss in complex disturnces, setting point position is initil is (0,0) meters nd trget (3465, 6000) meters, Hs = 2 meters. desire ctul desire ctul (3465,6000) Ψd = 30 (0,0) βc=30 o, γw=30 o Figure Trjectory of Mriner Clss in complex disturnces, setting point position is initil is (0,0) meters nd trget (3465, 6000) meters, Hs = meters. βc=30 o, γw=30 o Figure 3 Trjectory of Mriner Clss in complex disturnces, setting heding is -30 o Hs = meters. In their two different mechnism in course control nd trcking control, in which trjectory of ship moving s shown in fourth ove figures. If the setting of controller is heding trget, the st unit of FLC is ctive, wheres setting of controller is

10 76 The 5th Interntionl Conference on Informtion & Communiction Technology nd Systems position of desired pth, st nd 2 nd units of FLC re ctive. The different of those mechnism proceeds error of the ctul positions to desired trget. In figure of nd 3, zoom of nd loction, the distinct of pth when setting in controller is heding -30 o, result of error igger thn in which is setting controller is point trget. The error when just st unit of FLC is ctive nd the oth re ctive is shown in tle of 4. desire ctul Ψd = 30 βc=30 o, γw=30 o Figure 3 Trjectory of Mriner Clss in complex disturnces, setting heding is -30 o, Hs = 2 meters. Tle 4 Error performnce of FLC result t the heding nd trcking set point. 5. THE CONCLUSION From result of simultion, re result conclusions The developing n expert control system in se trnsporttion is is done for fulfilling trjectory, position nd voiding collision. The Sugeno Tkgi FLC controller giving good performnce t turning, this is mrked y gin of Nomoto from simultion result is tend to sme with result from the clcultion. FLC controller with Sugeno- Tkgi consequence function giving good performnce when turning nd trcking in wve disturnce. The Sugeno Tkgi FLC is follow LQG/LTR roust controller, this is could e seen in trjectory response in turning nd lso trcking. REFERENCES [] Aisjh, A.S., Soegiono, Msroeri, AA., Djtmiko, E.B., Wsis nd Sutntr, I.N., (2004), Liner Trcking Se Vehicle sed on Fuzzy, Proceeding Seminr Ntionl Psc Srjn ITS. [2] Aisjh, A.S., Soegiono, Msroeri, AA., Djtmiko, E.B., Wsis, Sutntr, I.N nd Bud, K., (2005), The extended of Trcking Control Se Vehicle sed on Fuzzy Logic, Proceed. Seminr Ntionl FTI ITS. [3] Aisjh, A.S., Soegiono, Msroeri, A.A., Djtmiko, E.B.,Wsis, Sutntr I.N. nd Bud, K., (2005), A Study of Extended Fuzzy Logic Control for Ship Mneuvering Bsed on LQG/LTR Control, Proceeding Seminr Ntionl FTI ITS. [4] Aisjh, A.S., Msroeri, A.A., (2005c), Extended Fuzzy Logic Control for Ship Mneuvering Bsed on LQG/LTR Control, Interntionl Seminr ISME Jpn. [5] Aisjh, A.S., Soegiono, Msroeri, AA., Djtmiko, E.B., nd Wsis, (2006), Roust Control Systems in Ship Mneuvering from Wve, Se Current nd Wind disturnces, Journl Kelutn FTK ITS. [6] Aisjh, A.S., Soegiono, Msroeri, AA., Djtmiko, EB., nd Wsis, (2006), The roustness of Fuzzy Logic Controller in Ship Mneuvering from Stochstic Disturnces, Seminr Psc Srjn ITS. [7] Aisjh, A.S., nd Msroeri, AA., (2006c), Fuzzy Logic Control of Type Sugeno Tkgi with The Model Reference of LQG/LTR t Mneuvering Ship Controller, Interntionl Seminr JSPS. [9] Aisjh, A.S., Soegiono, Msroeri, A.A., Djtmiko, E.B., Wsis nd Sutntr, I.N., (2006d), Stility Are of Ship Mneuvering using Fuzzy Logic Control, Seminr Ntionl Teori dn Apliksi Tekn. Kelutn VI, FTK ITS. [0] Aisjh, A.S., Soegiono, Msroeri, A.A., Djtmiko, E.B., Wsis, dn Sutntr, I.N., (2006e), Smrt Controller in Ship Mneuvering, Jurnl T. Kelutn, FTK ITS. [] Aisjh, A.S., Soegiono, Msroeri, A.A., Djtmiko, E.B., Wsis, D.A, dn Sutntr, I.N, (2006f), The Impct of Se disturnces in Ship Mneuvering Controller, Seminr Nsionl Teknik Fisik FTI ITS. [2] Aisjh, A.S., Soegiono, Msroeri, A.A., Djtmiko, E.B., Wsis, D.A, dn Sutntr, I.N, (2007), The LQG/LTR Controller s Reference of Fuzzy Logic Controller in Mriner Clss Vessel., Jurnl GEMATEK STIKOM Sy. [3] Aisjh, A.S., Soegiono, Msroeri, AA., Djtmiko, E.B., Wsis, D.A, dn Sutntr, I.N, (2007), FLC Sugeno Tkgi in

11 C26 - Fuzzy Logic Control System For Developing Expert Se Trnsporttion - Auli Siti Aisjh 77 Mneuvering of Mriner Clss Vessel, Jurnl Industri, Vol. 6 No. 2, [4] Berge, S. K., Ohtsu, K. dn Fossen, T. I., (998), Non Liner control of Ships Minimizing The Position Trckings Errors, IFAC Conference CAM S 98, Fukuok Jpn. [5] Bertin, D., (998), Trck Keeping Controller for Precission Mnoeuvring Autopilot, IFAC Conference CAMS 98, Fukok, Jpn. [6] Breivik, M. dn Fossen, T.I., (2004), Non Liner Roust Mneuvering nd Positioning Control os Ships in extreem sitution, [7]Bretscneider, C.L., (969), Wve nd Wind Lods. Section 2 of Hndook of Ocen nd Underwter Engineering., Mc. Grw Hill, New York. [8] Consegliere, A. dn Lopez, M.J., (2000), H Controller Tuning Method for Ship Autopilot, 5th IFAC Conference on Mnoeuverring nd Control of Mrine Crft, MCMC 2000, Alorg, Denmrk. [9] Fossen, T. I., (994), Guidnce nd control of ocen vehicle,, John Willy & Son. [20] Kosko, B., (997), Fuzzy Engineering, Prentice Hll. [2] Lee, S. M. dn Kyung, K., (2004), A Fuzzy Logic for Autonomus Nvigtion for Mrine Vehicles Stisfying Coldreg Guidelines, Interntionl Journl Control nd System, Vol 2 No. 2. [22] Lozowicki, A.dn Tino, A., (2000), On the Design of High Precission Ship Trck Keeping System, 5th IFAC Conference on Mneuvering nd Control of Mrine Crft, MCMC 2000, Alorg, Denmrk. [23] Minorsky, N., (922), Directionl Stility of Automtic Steered Bodies, Journl Americ of Nvl Enegineers, No. 34 Vol 2. [24] Moyno, E., Lopez, E., nd Velsco, F.J., Trck Keeping Autopilot, Proceeeding 5 th Interntionl Conggres on Mritime Technologicl Innovtions nd Reserch, 2 23 Novemer 2007, Brcelon. [25] Omerdick, E., dn Roerts, G.N., (2004), A fuzzy trck keeping utopilot for ship steering, Journl of Mrine Engineering nd Technology, Vol. A2. [26] Perez, T. dn Blnke, M., (998), Mtehemticl Ship Modelling for Control Appliction, Technicl Report Dept. Of Electricl nd Computer Engineering, The University of NewCstle, Austrli, Section of Automtion t Oersted DTU, Technicl University of Denmrk, Lyngy Denmrk. [27] Trintfyllou, M. S. Dn Hover, F.S., (2004), Mnoeuvering nd Control Mrine Vehicles, Dept. Of Ocen Engineering, Msschusett Institute of Technology, [28] Velgic, J., Vukic, Z. dn Omerdick, E., (2002), Adptive Fuzzy Ship Autopilot for Trck Keeping, www. Elsevier.com. [29] Velgic, J., Vukic, Z. dn Omerdick, E., (2000), Adptive Fuzzy Ship Autopilot for Trck Keeping,5th Conference on Mneuvering nd Control of Mrine Crft, MCMC [30] Vukic, Z., Omerdic, E. dn Kuljc, L., (998), Improved Fuzzy Autopilot for Trcking Keeping, IFAC Conference CAMS 98, Fukuok Jpn.

12 78 The 5th Interntionl Conference on Informtion & Communiction Technology nd Systems [This pge intentionlly left lnk]