Indonesi an  Journa of El ect ri cal Engineer ing  an d  Comp ut er  Scie nce   Vo l.   23 ,  No.   3 Septem ber   20 21 pp.  1385 ~ 139 7   IS S N: 25 02 - 4752, DO I: 10 .11 591/ijeecs .v 23 .i 3 . pp 1385 - 139 7          1385       Journ al h om e page http: // ij eecs.i aesc or e.c om   Enh ancement   in   pn eumatic   posi tion ing   syst em   usin g   nonli ne ar   ga in   constr ained   model   p re di ctive   contr oll er :   expe riment al   va lid ation       Siti   Fa tima h   S ulaima n 1 ,   M.   F.   Ra h ma t 2 ,   A hma d   A th if   F au dz i 3 ,   Kh air uddin   Osm an 4 ,   N.   H.   S un ar 5   1 ,4 Advanc ed   Sen sors   and   Embed ded   Contro l   Res ea rch   Group,   Ce ntre   for   T el e comm unic at ion   Rese arc h   and   Innov ation   (CeT RI),   Facu lty   of   E lectr oni c s   and   Com pute r   E ngine er ing,   Univ ersit i   Te kn ika l   Malay s ia   Mel ak a   2 ,3,5 School   of   Ele ct ri ca l   Eng ine er i ng,   Univer si ti   Teknologi   Ma lay si a   2, 3 ,5 Cent re   for   Ar ti ficia l   Int el l ige n ce   and   Robo ti cs   (CAIRO ),   Unive rsiti   Te knolog i   Malay s ia       Art ic le   In f o     ABSTR A CT   Art ic le   hist or y:   Re cei ved   M ay   9 ,   2021   Re vised   Ju l   1 8 ,   2021   Accepte d   J ul   2 9 ,   2021       The   issues   of   in ac cur ate   posit io ning   control   h av e   m ade   an   industri al   use   of   pneumati c   a ct u at or   r emains   re stric t ed   to   c ert a in   app li c at ions   onl y .   Non - complia nc e   wit h   s y stem   li m it s   and   prope rl y   c ontrol   the   op erati ng   s y st em   may   al so   degr a de   the   per form a nce   of   pneumatic   positi oning   s y stems .   Thi s   stud y   proposed   a   new   appr oac h   to   enha nce   pn eumati c   positi o ning   s y stem   while   consid eri n g   the   constr ai nt s   of   sy st em.   Fir stl y ,   a   m at h ema ti c al   m odel   tha t   r epr ese nt e d   the   pneuma ti c   s y s te m   wa s   det ermined   by   s y s te m   ide nti f icati on   ap proa ch.   Second l y ,   m odel   p red i c ti ve   cont ro ll er   ( MP C)   was   deve lop ed   as   a   primar y   cont r oll er   to   control   the   pneuma ti c   positi onin g   s y stem,   which   t ook   int o   a cc ou nt   the   constra in ts   of   the   s y s tem .   Next,   to   enha nc e   the   per f orm anc e   of   the   over all   s y stem,   nonli ne ar   gai n   f unct ion   was   inc orpora te d   wi thi n   the   MPC   al gorit hm .   Fin all y ,   th e   per form anc es   wer e   compare d   with   othe r   con trol   m et hods   such   as   constra in ed   MPC   (CMP C),   proporti ona l - inte gra l   (PI),   and   p red ictive   fun ctio nal   con trol   with   observe r   (PF C - O).   The   va li dation   bas ed   on   real - ti m e   experim ent al   result s   for   100   mm   positi oning   control   rev eale d   th at   the   inc orpor at i on   of   nonli nea r   gai n   within   the   MPC   al gorithm   improved   21 . 03%   and   2. 69%   of   the   spee d   res ponse   give n   by   CMP C   and   PFC - O,   and   red u ce d   100%   of   th e   over shoot   give n   by   CMP C   and   PI   cont rolle r;   thus,   providin g   fast   and   a cc u rat e   pn eumati c   positi oning   cont rol   s y st em.     Ke yw or d s :   Pn e um atic   act uator   Po sit io n   c on tr ol   Pr e dicti ve   co nt ro l   Syst e m   identific at ion   Transi ent   re spon s e   This   is   an   open   acc ess   arti cl e   un der   the   CC   BY - SA   l ic ense .     Corres pond in g   Aut h or :   Sit i   Fatim ah   Su la i m an   Ce ntre   for   Tel e com m un ic at ion   R esea rch   an d   I nnovat ion   (C eTR I)   Faculty   of   Ele c tr onic s   an d   C om pu te r   En gin e erin g   Un i v ersit i   Te knikal   Ma la ysi a   Me la ka   76100   D ur ia n   Tu nggal,   Me la ka,   Ma la ysi a   Em a il :   sitifatim ahsu la im an @u te m .ed u. m y       1.   INTROD U CTION     The   pne um at ic   syst e m   is   of te n   ass ociat ed   with   the   issues   of   nonlinea riti es   and   un ce rtai nt ie s   su ch   as   com pr essibil it y   of   ai r   a nd   fr i ct ion   e ff ect   sur rou nd i ng   the   s yst e m   [1 ] [ 2].   Ther e f or e,   it   is   a   c halle ng i ng   ta sk   to   con t ro l   the   syst e m   for   a   certa in   desire d   perf or m ance.   C ont ro ll in g   the   syst e m ’s   posit ion   al so   bec om es   m or e   com plica te d   with   the   nee d   to   sim ultaneou sl y   consi de r   the   accu racy   of   s yst e m   and   res pons e   tim e.   V ario us   con t ro ll ers   we re   re porte dly   pro po se d   to   c on t ro l   t he   pos it ion   of   pne um at ic   syst e m ;   su c h   as   propor ti onal - integral - de riva ti ve   (P I D )   [ 3],   po le - p la cem e nt   [ 4],   H   loop   sh api ng   co ntr oller   [ 5],   ada pt ive   con t ro ll er   [6 ] ,   Evaluation Warning : The document was created with Spire.PDF for Python.
                          IS S N :   2502 - 4752   Ind on esi a J  E le c Eng &  Co m Sci,   Vo l.   23 , N o.   3 Se ptem ber   20 21 13 85   -   139 7   1386   fu zzy   l og ic   [7 ] ,   an d   predict iv e   co ntr ol   [8 ] [ 9].   H owev er,   m os t   of   these   r eported   co ntr ol le rs   we re   i ncapable   to   si m ultaneou sly   delive r   high - s peed   res ponse   with   acc ur acy .   It   is   al s o   c ru ci al   to   a ddress   t he   c onstrai nts   of   the   syst e m   in   the   desig n   of   c on t ro ll er   for   the   act ual   app li cat ion s   of   t he   syst e m .   The   non - com pliance   with   the   pr esc ribe d   c on strai nts   m ay   ca us e   dam age   to   the   pne um atic   syst e m   and   its   com po ne nts   a nd   af fects   the   overall   perform ance   of   the   c ontrol   s yst e m   i tse lf.   Hen ce ,   this   st ud y   reg a r ded   the   sig nificanc e   of   co ns ide ri ng   t he   const raints   of   the   pn e um at ic   s yst e m   du rin g   t he   dev el op m en t   of   the   c ontr oller.     T his   stu dy   propose d   the   m od el   pre dicti ve   con tr oller   ( MPC )   as   the   con tr ol   strat egy   for   the   po sit io ning   of   the   syst e m .   M PC   can   co ns id ers   the   co ns trai nts,   deals   with   m ul ti var ia ble   con t ro l   issue ,   an d   al s o   su it able   f or   si ngle - in put   an d   s ing le - outp ut   (SISO)   syst em   a nd   m ulti - input   and   m ulti - ou t pu t   (MIMO )   s yst e m   [10 ] - [ 12] ,   w hi ch   e xp la in   why   it   is   hig hly   f avou rab le   in   the   process   i ndust ries.   MPC   is   al so   re ported   use d   to   con t ro l   rob ots   [13],  [ 14] .   Fi gure   1   sho ws   t he   basic   st ru ct ure   of   MPC .         R e f e r e n c e   t r a j e c t o r y M o d e l + - O p t i m i z e r P a s t   i n p u t s   a n d   o u t p u t s F u t u r e   i n p u t s C o s t   f u n c t i o n C o n s t r a i n t s F u t u r e   e r r o r s P r e d i c t e d   o u t p u t s     Figure   1.   Ba sic   struct ur e   of   M PC   [ 11 ]       MPC   in   Fig ur e   1   us es   a   proce ss   m od el   to   pre dict   the   f uture   ou t pu t   base d   on   the   previ ou s   and   cu rr e nt   values   by   so lvin g   the   opti m al   fu ture   co ntr ol   act ion s   at   each   sa m pli ng   ti m e   instant   [1 1].   Th e   optim iz er   consi ders   the   c os t   f unct ion   an d   the   co ns trai nt s   of   the   syst e m   to   cal culat e   the   co ntr ol   act ion s .   In   pa rtic ular,   t he   add e d   co ns trai nts   in   MPC   pr e ven t   the   wi nd - up   ph e nom eno n.   W it h   the   use   of   MPC   as   a   s trat egy   to   con tr ol   the   po sit io n   of   the   pneum at ic   syst e m   us ed   in   this   stu dy,   t he   c onstrai nts   in   the   pne um at ic   syst e m   can   be   consi der e d   [ 15] .     The   em plo ym e nt   of   MPC   as   a   con tr ol   strat egy   to   con t ro l   the   posit ion   of   pn e um at ic   syste m   us ed   in   this   stud y   has   been   e valuate d   in   pr e vious   stud ie s   [ 16] [ 17].   In   orde r   to   ve rify   the   eff ect iveness   of   the   strat egy   in   ha ndli ng   syst e m s   with   co nst raints,   c on st ra ints   we re   ap plied   to   t he   in put   of   t he   syst em   (co nt ro l   si gn al   to   the   valves ).   From   the   st ud y,   they   fou nd   t hat   c onstrai ned   MPC   su ccess fu ll y   pr oduce d   acc ur at e   tracki ng   com par e d   to   unc onstrai ne d   MPC   [ 16 ] ,   [ 17 ] .   H oweve r,   they   f ound   th at   the   i nclusi on   of   co ns trai nts   in   the   MPC   al gorithm   reduce d   the   syst e m   resp onse   and   ag gr e ssivene ss   of   the   pn e um at ic   syst e m   since   it   req ui res   m or e   com pu ta ti on al   effor t   to   opti m ise   the   cost   functi on   (c om par ed   to   the   unc on st raine d   cas e)   [ 16 ] [ 17 ] .   These   fin dings   w ere   co ns ist ent   with   sev eral   pr e vious   st ud ie s   [11],   wh ic h   dem on strat ed   that   the   i nclus ion   of   const raints   in   t he   MPC   al gori thm   did   reduce   the   res pons e   t i m e   in   the   syst e m   ou tp ut.   T hus,   t he   init ia l   us e   of   const raine d   M PC   was   i nca pa ble   to   pro vid e   accu rate   an d   tim e ly   po sit io ning   re spo ns e   of   pneum at ic   syst e m   us e d   in   this   stu dy   [16],  [ 17 ] .   The   us e   of   li ne ar   co ntro ll er   to   con tr ol   a   syst em ,   especial ly   the   nonline ar   s yst e m ,   is   ty pical l y   al s o   incapa ble   to   si m ultaneou sl y   deliver   high - sp ee d   res ponse   with   accurac y   [1 8].   Des pit e   that,   this   stu dy   co nsi der e d   that   t he   syst e m   can   be   i m pr ov e d   th rou gh   certai n   m od ific at ion s.   This   st ud y   fo c us e d   on   t he   a ppr oac h   a dopte d   by   Sera j i   [19],   co ns i der i ng   that   its   im ple m entat ion   is   eff ect ive   a nd   st raig htfor ward.   Sera j i   [19]   i m pr ove d   the   co nv entional   P ID   c on t ro ll er   by   co m bin ing   a   nonl inear   gain   in   casca de   and   a   li near   fixe d - gain   P ID   con tr oller   in   orde r   to   co ntr ol   the   robo ti c   ar m .   Con seq ue ntly ,   the   con t ro ll er   is   a bl e   to   adap t   its   r esp on se   base d   on   t he   pe rfo rm ance   of   the   cl ose d - l oop   c on t r ol   syst e m .   When   the   error   of   the   co ntr olled   var ia bl e   is   la rg e,   the   gain   am plifie s   the   err or   to   ge ner at e   a   la rg e   cor recti ve   act ion   to   rap i dly   dr i ve   t he   syst em   ou t pu t,   res ulti ng   in   high - s pee d   r esp on se .   Wh e n   the   er ror   di m inishes,   the   gain   is   autom at ic ally   reduce d   to   pre ven t   e xcessi ve   os ci ll at ion s   a nd   la rg e   over s hoots   in   the   r esp on se ,   re su lt ing   in   accurate   res pons e   with   ze ro   s te ady - sta te   er r or.   In   oth e r   w ords ,   the   co ntr ol le r   posses ses   t he   a dv a nta ge   of   high   init ia l   gain   ( w hich   deliver s   hi gh - sp ee d   respon s e)   an d   low   gain   (that   pre ve nts   an   os ci ll at or y   beh a viou r   in   the   syst e m   resp on se   gi ven   its   au tom a ti c   nonlin ear   gain   a dj us tm ent).   In   an ot her   stu dy,   Ra hm at   et   al.   [20]   f ound   Evaluation Warning : The document was created with Spire.PDF for Python.
Ind on esi a J  E le c Eng &  Co m Sci     IS S N:  25 02 - 4752       En hance men t   in   pn e umatic   posit ion i ng   syst em   us in g   nonli near   ga i n   c on s traine d   ( Siti   Fatim ah   Sulai man )   1387   that   the   c om bi nation   of   a   no nlinear   gai n   a nd   the   PID   c ontrolle r   re du ces   the   ov e rs hoot   a nd   pro duces   ac cur at e   trackin g   in   the   pneum at ic   posit ion in g   syst em   resp on se ,   w hich   was   c orr oborat ed   by   S al i m   et   al.   [21 ] .   The   stud y   al s o   de m on strat ed   the   en hancem ent   of   pne um atic   po sit io ning   syst e m   in   the   tra nsi ent   res ponse   of   the   syst e m   wh ere   the   nonlinea r   ga in   functi on   an d   PID   co ntr oller   wer e   c om bi ned   to   c on t ro l   the   cy li nd er   str ok e   of   pn e um at ic   syste m   with   a   pay load   of   m axim u m   weigh t   up   to   28   kg   at   the   en d   of   the   pn e um at ic   cylinder   stroke.   T he   ef f ect iveness   of   t he   pro pose d   appr oach   by   Se raj i   [ 19]   in   reducin g   the   ove r sh oot   an d   pr oducin g   bette r   trac king   in   the   syst em   per f or m ance   wa s   pro ven   us i ng   var i ou s   a ppli cat ion s,   s uc h   as   in   r obotics,   m il li ng   syst e m s,   and   w ast ewater   treat m ent   process   [ 22 ] - [ 24 ] .     Give n   s uch   co ns ide rati ons,   t his   stu dy   inc orporated   a   no nlinear   gai n   in   t he   co nv e ntio nal   co ns trai ne d   MPC   (CMPC )   as   a   ne w   c ontr ol   strat egy   to   im pr ove   sp ee d   re spo nse   an d   acc ur ac y   of   the   pne um atic   po sit io ning   sy stem .   Since   this   stu dy   involves   with   re al - tim e   i m p leme ntati on ,   the refor e ,   this   s t ud y   al s o   consi der e d   t he   us e   of   ob se r ve r   in   t he   desig n   of   c ontr ol   str at egy.   Esse ntial ly ,   ob se rv e r   is   us e d   to   est im a te   the   internal   sta te   va riable   of   the   r eal   syst e m   [2 5] .   This   st ud y   is   di vid e d   int o   6   sect io ns .   T he   bac kgr ound   of   the   stud y   is   discu s sed   in   sec tion   1.   Sect io n   2   prov i des   the   in form ation   ab ou t   the   pneum at ic   syst e m   e m plo yed   in   this   stud y   by   descr i bing   the   syst e m ’s   com po nen ts   an d   exp la inin g   its   op e rati on.   S ect ion   3   ex pla ins   the   exp e rim ental   s et up   a nd   the   proces s   of   m odel li ng   the   syst e m   us ing   a   syst e m   identific at ion   te ch nique.   The   proce dures   in   desig ning   the   pro po se d   co ntr ol   strat egy   to   perform   the   con tr ol   ta sk   are   pr ese nted   in   sect ion   4.   Sect ion   5   pr es ents   an d   discu s ses   the   exp e ri m ental   resu lt s   of   the   propose d   co ntr oller.   T he   eff ect i ven es s   of   the   pro po se d   c ontr ol   strat egy   in   c om par ison   to   t he   oth e r   c on tr ollers   is   al s o   dem on strat ed   in   sect i on   5   an d   la stl y,   sect i on   6   c oncl ud e s   the   overal l   fin dings   of   the   stu dy .       2.   PNEU M ATI C   S YS TE M   D ESCRIPT IO N   The   pne um at ic   syst e m   us ed   in   this   stu dy   is   sh ow n   in   Fi gure   2.   It   was   e quip ped   with   opti cal   senso r   (A E DR - 8300 ) ,   la ser   st ripe   r od,   press ur e   se ns or   ( KOG ANEI:   PS U - EM - S) ) ,   valves   ( K OGA NEI :   EB10ES 1 - PS - 6W) ,   an d   pro gr am m able syst e m  o c hip   (P S oC)   c ontr ol   boar d.   The   pn eum atic   syst em   us ed   in   t his   stud y   is   a   double - act in g   ty pe   cy li nd e r   ( K OGA NEI - HA:   t winport   cy li nd ers )   with   16   mm   rod   di a m et er   and   200   mm   rod   st roke   le ngth.   T he   sc hem at ic   diagr am   of   the   syst em   is   il l us trat ed   in   Fig ur e   3.             Figure   2.   The   pn e um at ic   syste m   and   its   m ain   c om po ne nts       Tw valve w hich  w ere  at ta ched   at   the  e nd  of  the  cy li nder,  we re  em plo ye to  co ntr ol  the  inlet   and  ou tl et   ai of   t he   cy li nd er.   I t his  stu dy,  the  e xtensi on   a nd  r et racti on   of   t he   cy li nd er  st rok are  m anipu la te by  the  duty   cy cl of   a   pulse - widt h   m od ulato (PWM si gn al   t dr i ve  th val ve s.  T he  P WM  m od el   in  this  s tud y   has  pr eci si on  of   8 - bit.  Wh e the  P W m od el   recei ves  a   po sit ive  sig nal   fr om   the  plant  or   co ntr oller,  it   will   conve rt  the  sig nal  into  e qu i va le nt  P W sig na and   se nd   t ha sign al   to  the  valve  to  pe rfor m   extensio n.  If   the  P W M m od e l r ecei ves  neg at iv e   sig nal, the  m od el   will  send t he  si gn al  t o valve  to  p e rfo r m  r et racti on     S t r i p e   c o d e   O p t i c a l   e n c o d e r P r e s s u r e   s e n s o r V a l v e s P S o C   m i c r o c o n t r o l l e r   b o a r d G u i d e   r o d Evaluation Warning : The document was created with Spire.PDF for Python.
                          IS S N :   2502 - 4752   Ind on esi a J  E le c Eng &  Co m Sci,   Vo l.   23 , N o.   3 Se ptem ber   20 21 13 85   -   139 7   1388       Figure   3.   The   pn e um at ic   syste m   schem atic   diag ram       3.   MO DEL   I DENTIFI CA TI O N   Figure   4   sho ws   the   ex per im ental   set - up,   as   a   par t   of   the   pro cess   to   obta in   t he   m at he m atic al   m od el   of   the   syst e m .   In   this   stud y,   a   m at hem atical   mo del   of   the   pne um atic   syst e m   was   ide ntifie d   by   m eans   of   syst e m   identific at ion.   The   platf orm   f or   this   st ud y   was   MAT LA B/ Si m ulink ,   w hich   was   eq ui pp e d   in   t he   co m pu te r.   The   no m inal   pr essu re   us ed   w as   0.6   MPa   a nd   a   natio nal  in strum ent  (N I )   (P CI/P XI - 62 21)   was   us e d   f or   data   acqu isi ti on   (DAQ)   syst em .         D A Q   s y s t e m P n e u m a t i c   s y s t e m P e r s o n a l   c o m p u t e r   ( e q u i p p e d   w i t h   M A T L A B   s o f t w a r e ) A i r   c o m p r e s s o r   s y s t e m     Figure   4.   The   s et - up   of   e xperi m ent       A   total   of   2000   m easur em ents   of   i nput   an d   ou t pu t   data   w ere   colle ct ed   at   sa m pling   ti m e   ( )   of   10   ms   du ri ng   e xp erim ent.   The   input   data   c on t ai ns   20 00   data   po i nts   of   co nt inu ous   ste p   si gn al   a pp li ed   to   the   valves ,   w hile   the   ou t pu t   da ta   con sist   of   2000   m easurem ents   of   cy li nd e r   stroke   po sit io n   sig na l.   The   Au t oReg ressiv e   with   eX og e nous   i nput   ( ARX)   par am et ric   m od el   was   cho se n   for   this   stud y   since   it   sat isfie s   the   crit eria   f or   syst e m   identific at ion .   T he   id entifi ed   discret e   sta te - sp ace   m od el   based   ARX   m od el   st ru ct ure   util iz ed   thr oughout   this   stu dy   is   represe nted   by   ( 1).       = [ 0 1 0 0 0 1 0 . 1284 0 . 9976 1 . 8690 ] , = [ 0 0 1 ] , = [ 0 . 0016 0 0 ] ,   = [ 0 ]   (1)     C h a m b e r   1 C h a m b e r   2 V a l v e   1 V a l v e   2 P r e s s u r e   s e n s o r C y l i n d e r S u p p l y   p r e s s u r e   ( 0 . 6   M P a ) E x h a u s t O p t i c a l   e n c o d e r E x t e n d R e t r a c t S t r o k e I 2 C C o u n t e r P W M A D C P S o C Evaluation Warning : The document was created with Spire.PDF for Python.
Ind on esi a J  E le c Eng &  Co m Sci     IS S N:  25 02 - 4752       En hance men t   in   pn e umatic   posit ion i ng   syst em   us in g   nonli near   ga i n   c on s traine d   ( Siti   Fatim ah   Sulai man )   1389   The   ide ntifie d   m od el   fits   the   act ual   plant   m od el   at   a   value   of   ap pro xim a t el y   91 .09   %.   The   loss   of   8.91%   m ay   be   due   to   dea d - z one,   fr ic ti on  a nd   ai r   le a kag e .   in   the   pn e um at ic   sys tem   it sel f.   All   the   nonlinea riti es   in   this   stu dy   ar e   neg le ct e d   an d   the   syst em   util iz ed   is   assume d   to   be   a   li near   syst em .   As   the   m od el   in   ( 1)   pr ov i des   al l   the   pole s   insi de   the   un it   ci rcle   ( 0.1 887,   0.6 811   an d   0.999 2),   it   is   consi der e d   sta ble.         4.   CONTR OLL ER   DE SIG N   This   sect io n   di scusses   the   relevan t   desi gn   of   the   c onve ntio nal   co ns trai ned   MPC   (CMPC )   to   c ontr ol   the   posit io n   of   pn e um at ic   syste m   with   certai n   m od ific at ions   f or   en ha nced   perform ance.       4.1.     MP C   f or mulati on   MPC   is   a   ty pe   of   c on tr oller   t hat   is   desig ne d   base d   on   a   m at he m at ic a l   m od el   of   t he   plant.   In   this   stud y,   the   pne um atic   m od el   us e d   was   ta ke n   to   be   a   sta te - sp ace   m od el ,   and   the   pneum at i c   syst e m   us ed   in   this   stud y   was   ass um ed   as   a   SISO   syst e m .   The   de te rm inist ic   m od el   of   pneum at ic   syst e m   e m plo ye d   in   the   s tud y   with     inputs   a nd     outp uts   is   de scribe d   as   ( 2)   a nd   ( 3).       ( + 1 ) = ( ) + ( )   (2)     ( ) = ( ) + ( )   (3)     W he re   ,   ,   ,   a nd     ar e   syst em   m at ri ces   with   a ppropr ia te   dim ensi on s ,     is   t he   sta t e   va riabl e   vecto r   with   dim ension   ,     is   the   input   var ia ble   vecto r,   an d     is   the   process   outp ut   vect or.   In   this   stu dy,     a nd     =   1   sinc e   t he   s yst e m   is   a   S ISO   syst e m .   Ma trix     is   ass um ed   to   be   ze r o   in   orde r   to   dem on s trat e   that   there   is   no   direct   f eed   thr ough   the   in pu t,   ( )   an d   the   outp ut,   ( ) .   This   is   due   to   the   pr i nciple   of   r eced ing   horizo n   con t ro l   it sel f,   in   wh ic h   only   a   curre nt   inf or m at ion   of   the   plant   is   requir ed   for   pre dicti on   an d   c on t ro l .   The   m od el   in   (2)   w as   i m pr ov e d   in   order   to   incl ude   an   inte gr at or   in   the   desi gn.   The   dif fer e nt   on   bo t h   sides   of   (2)   yi el ded   the   inc rem ental   sta te - sp ace   in   ( 4),     ( + 1 ) = ( ) + ( )   (4)     W he re     ( + 1 ) = ( + 1 ) ( ) ,   ( ) = ( ) ( 1 ) ,   ( ) = ( ) ( 1 )       ( + 1 )   and   ( )   de note   the   differe nce   of   the   sta te   var ia ble,   a nd   ( )   de no t es   the   di ff e rence   of   t he   con t ro l   va riabl e.   A   new   sta te   var ia ble   vector   in   ( 5)   was   sel ect ed   in   order   to   c onnect   the   s ta te   var ia ble   ( )   to   the   outp ut   ( ) .       ( ) = [ ( ) ( ) ]   (5)     Con si der i ng   (4)   in   outp ut,   (3)   can   be   wr it te n   in   the   f or m   of   ( 6),       ( + 1 ) ( ) = ( ( + 1 ) ( ) )   (6)     Re arr a ng e   ( 6)   base d   on   ( 4)   gi ves     ( + 1 ) = ( ) +   ( ) + ( )   (7)     The   a ugm ente d   sta te - s pace   m od el   as   ind ic at ed   in   ( 8)   c an   be   ob ta ine d   by   pu tt in g   to get he r   (4)   a nd   ( 7).       ( + 1 ) =  ( ) + ( ) , ( ) =  ( )   (8)     W he re,     = [ 0 ] , = [ ] , = [ 0 1 ]       ,   and     is   the   a ugm ented   m od el ,   0   is   the   zer o   m at rix   with   dim ensio n   ×   and     is   a   un it   m at rix   wit h   dim ension   × .   Con side rin g   (1)   as   a   plant   m od el ,   the   a ug m ented   m od el   of   the   pn e um at ic   syste m   util iz ed   in   this   stu dy   is   re pr ese nted   as,     Evaluation Warning : The document was created with Spire.PDF for Python.
                          IS S N :   2502 - 4752   Ind on esi a J  E le c Eng &  Co m Sci,   Vo l.   23 , N o.   3 Se ptem ber   20 21 13 85   -   139 7   1390   = [ 0 1 0 0 0 0 1 0 0 . 1284 0 . 9976 1 . 869 0 0 0 . 0016 0 1 ] , = [ 0 0 1 0 ] , = [ 0 0 0 1 ]   (9)     T he   a ugm ente d   m od el   in   ( 9)   has   ei ge nv al ue s     at   1,   0.9 992,   0681 1,   a nd   0.188 7.   The   fir st     is   from   the   au gm entation   of   the   pl ant   m od el ,   w hile   the   la st   three     are   fro m   the   or i gin a l   pn e um at ic   plant.   Con se quently ,   the   au gm ented   sta te - sp ace   m od el   has   one   in te gr at or   em bedded   into   t he   a ugm ented   sta te - sp ace   m od el   to   su it   t he   c on t ro ll er   de sign   pur pose.   The   desig n   of   MPC   is   base d   on   opti m iz ing   the   di ff e ren ce   of   t he   con t ro l   sig nal   ( )   withi n   an   optim iz ation   window.   ( ) ,   or   so   ca ll ed   the   f uture   co ntr ol   tra j ec tory   is   denoted   by   ( 10),       Δ ( ) , Δ ( + 1 ) , , Δ ( + 1 )   (10)     W he re     is   the   con t ro l   horizo n,   w hich   is   use d   to   dicta te   the   nu m ber   of   par am et ers   in   order   to   ca ptur e   the   fu t ur e   co ntr ol   t raj ect or y.   Wh e n   the   sta te   var i able   vector   ( )   at   sam pling   tim e   instant     is   ass um ed   avail able   thr ough   m easur em ent,   the   c urre nt   plant   i nfor m at ion   is   prov i ded   by   the   sta te   ( ) .   W it h   giv en   in form at ion   ( ) ,   the   f uture   sta te   va riables   a re   de no te d   as   ( 11),      ( + 1 | ) , ( + 2 | ) , , ( + | ) , , ( + | )   (11)     W he re   ( + | )   is   the   predict ed   sta te   va riable   at   +   with   the   giv e n   c urr ent   pla nt   in for m at ion   ( )   an d     is   the   pr e dicti on   horizo n   or   t he   le ng t h   of   op tim iz at ion   window.   Gen e rall y,   .   In   t his   st udy,     an d     wer e   c hosen   to   be   3   an d   20,   resp ect ively .   T he   seq ue ntial ly   cal culat ed   f utu re   sta te   va ria ble   us in g   the   s et   of   fu t ur e   c ontr ol   par am et ers   is   e xpresse d     ( + 1 | ) =  ( ) + ( )       ( + 2 | ) = 2 ( ) +  ( ) + ( + 1 )       ( + | ) = ( ) + 1 ( ) + 2 ( + 1 ) +   + ( + 1 )   (12)     T he   s ubsti tuti on   of   ( 12)   i nto   outp ut   in   ( 8)   pro vid es   the   pr e di ct ed   outp ut   vari ables,   as   sho w n   in   ( 13)     ( + 1 | ) =  ( ) +  ( )       ( + 2 | ) =  2 ( ) +  ( ) +  ( + 1 )       ( + | ) =  ( ) +  1 ( ) +  2 ( + 1 ) +   +  ( + 1 )   (13)     Be sides,   (13 )   c an   al s o   be   wr it te n   in   a   c om pact   m at rix   form   as,      =  ( ) + ΦΔ   (14)     W he re     = [ CA C A 2  ] , Φ = [  0 0   0  1  2  ]       I n   t his   stu dy,   a   sta te   est i m a t or   or   an   ob se rv e r   syst em   w as   em plo ye d   in   the   desig n   of   c ontrol   stra te gy.   Assum ing   at   tim e   ,   the   in for m at ion   of   sta t e   var ia ble   ( )   wa s   not   m easur a ble   ( or   a vaila bl e),   an   obser ve r   syst e m   will   be   us e d   to   est im ate   the   sta te   var i able   ( )   f ro m   the   process   m easur em ent.   Hen ce,   the   f uture   sta te   var ia ble   in   ( 12)   w as   cal culat e d   us in g   t he   est im at ed   sta te   vari ables   as   in   ( 15 ),       ̂ ( + 1 ) = ̂ ( ) + ( ) +  ( ( ) ̂ ( ) )   (15)   Evaluation Warning : The document was created with Spire.PDF for Python.
Ind on esi a J  E le c Eng &  Co m Sci     IS S N:  25 02 - 4752       En hance men t   in   pn e umatic   posit ion i ng   syst em   us in g   nonli near   ga i n   c on s traine d   ( Siti   Fatim ah   Sulai man )   1391   W he re      is   the   gain   m at rix.   T he   po le - place m ent   was   em plo ye d   as   a   te c hn i qu e   to   cal culat e   the   s uitable      value.   The   pole s,   w hich   wer e   us e d   to   fi nd   the   s uitable      ar e   to   be   wer e   m ai ntained   in   the   un it   ci rcle   in   ord er   to   e nsure   t he   sta bili t y   of   t he   syst e m .   The   value   of   pole s   assi gned   in   t his   stu dy   is   dem on str at ed   in   Table   1.   In   this   stud y,   the   s equ e nce   of   Δ ( ) , Δ ( + 1 ) , , Δ ( + 1 )   in   (10)   was   a ppr oxim a te d   us in g   a   set   of   discrete - ti m e   Lag uerre   f un ct ion s .   The   set   of   discrete - ti m e   Lagu e rr e   functi ons   ex pr es sed   a   vecto r   f orm   descr ibe d   in   ( 16),   w hile   ( 17)   e xpresse d   its   di ff e ren ce   eq uatio n,      ( ) = [ 1 ( ) 2 ( ) ( ) ]   (16)     ( + 1 ) = ( )   (17)     W he re     is   t he   m at rix   with   di m ension   ×   an d   is   a   functi on   of   pa ram et ers     a nd   .   The   init ia l   conditi on   of   ( 16)   is   giv e n   in,      ( 0 ) = × [ 1 2 ( 1 ) 1 1 ]   (18)       is   the   scal in g   facto r   of   the   Lag uerre   netw ork   a nd   = 1 2 .   To   e nsure   the   sta bili ty   of   t he   netw ork,     m us t   be   within   0 < 1 .   In   this   st udy,   the   valu e   of     us ed   is   0.1.   This   stu dy   is   set   out   to   res pect   the   act uator   c on st r ai nts   w hile   br i ng i ng   a nd   m ain ta inin g   t he   po sit ion   of   the   cy li nd er   str ok e   as   cl os e   as   poss ible   to   the   desire d   pos it ion .   In   do i ng   so ,   the   m anip ul at ed   an d   c ontr olled   var ia bles   bei ng   co ns i de red   we re   t he   si gn al   to   the   val ves   a nd   t he   posit io n   of   the   cy li nder   str oke.   M PC   was   use d   in   this   stu dy   to   determ ine   the   fu tu re   adjustm ents   of   the   sign al   to   the   valve .   MP C   pr e dicte d   the   fu tu re   pla nt   ou t pu ts   a nd   pe rfor m ed   the   con t ro l   act ion s   acco r di ng ly   by   so lvi ng   the   optim al   fu tu re   co ntr ol   act ion s   (c os t   functi on   an d   c on st raints) .   Th e   cost   functi on     that   r eflect   the   c on t r ol   ob j ect ive   of   this   stu dy   is   de fine d   as,       = ( + | )  ( + | ) +  = 1   (19)     W he re     an d     ar e   the   weig htin g   m at rices   with   =   ( or   0 )   an d     wa s   ch os e n   to   be   0 . 1   (or   > 0 ) .     = = [ 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 ]   (20)     Wh e re,   = [ 0 0 0 1 ]       S ince   this   stu dy   aim s   to   br in g   an d   m ai ntain   the   pr e dicte d   ou t pu t   as   cl os e   as   po ssi ble   to   the   set - point   s ign al ,   the   sta te   var ia ble   ( + | )   in   ( 19)   m us t   be   re - de fin ed   in   orde r   to   include   the   set - point   sig nal   in   t he   c ost   functi on   e qu at i on.   Hen c e,   (21 )   is   t he   new   e quat ion   of   the   sta te   var ia ble   with   the   inclu sio n   of   set - po i nt   s ign al   ( )   within   the   opti m iz at ion   wind ow,      ( + | ) = [ ( + | ) ( + | ) ( ) ]   (21)     T he   opti m a l   so luti on   of   t he   pa ram et er   ve ct or     in   ( 19)   base d   on   m inim iz at ion   of   the   c os t   functi on   is   represe nted   as   (22),      = Ω 1  ( )   (22)     W he re   Ω = ( ) ( ) + = 1 , = ( ) = 1       U pon   ob ta ini ng   the   opti m al   s olu ti on   of   the   pa ram et er   vecto r   ,   the   c ontrol   la w   ca n   is   reali sed   as     Δ ( ) = ( 0 )   (23)     The   c ontrol   la w   in   ( 23)   ca n   al so   be   re prese nted   in   the   f orm   of   li near   sta te   feedbac k   c ontr ol   as   e xpres sed   in   (24),    Evaluation Warning : The document was created with Spire.PDF for Python.
                          IS S N :   2502 - 4752   Ind on esi a J  E le c Eng &  Co m Sci,   Vo l.   23 , N o.   3 Se ptem ber   20 21 13 85   -   139 7   1392   Δ ( ) = ( )   (24)     W he re     = ( 0 ) Ω 1 , ( ) = [ ( ) ( ) ]       H ence ,   the   co nt ro l   la w   to   be   a pp li ed   can   be   wr it te n   as     ( ) = ( ) + ( 1 )   (25)     ( )   is   the   error   si gn al   betwee n   the   predict e d   outp ut   ( )   an d   set - po i nt   signa l   ( ) .   W hen   an   ob se r ve r   is   us e d   in   the   des ign ,   al l   the   act ual   sta te   va riab le   ( )   will   be   re placed   with   the   obser ve d   sta te   va riable   ̂ ( ) .       4.2.     MP C   with   co nst r ain ts   on   in put   valv e s   Re stric ti on s   w ere   giv e n   to   the   co ntr ol   sig na l   to   the   syst em ’s   m anipu la te d   var ia ble   or   i nput   valve s   in   the   case   of   the   pn e um at ic   sys tem   us ed   in   t hi s   stud y.   In   t hi s   stud y,   a   co nt ro l   sig nal   was   def i ned   as   a   si gn al   expo rted   f ro m   the   con t ro ll er ,   wh ic h   will   infl uen ce   the   s yst e m   resp ons e   (i.e.   the   pos it ion   of   t he   cy li nd er   stroke) .   Th us ,   t his   sig nal   shoul d   be   co ntr olled   to   en sure   tha t   would   al ways   be   in   a   ra ng e   t hat   is   al lowed   by   the   syst e m .   If   the   m axi m u m   a llo wa ble   value   is   exceed ed ,   an   ov e rs hoot   m igh t   be   ge ne rated   in   the   syst e m   respo ns e.   T his   ph e no m enon   m ay   occu r   fr e qu e ntly   as   the   syst e m   is   i m p lem ented   in   real - tim e   env iro nm ent.   The   m axi m u m   am pli tud e   va lue   al lowe d   f or   t he   e xtensi on   an d   retract ion   of   the   cy l ind e r   str oke   duri ng   op e rati on   wer e   set   to   + 255   ( f or   val ve   1)   an d   - 255   (for   va lv e   2),   res pecti ve ly .   Hen ce ,   t he   sign al   from   the   MPC   to   the   syst em ’s   valves   wa s   co ns trai ne d   withi n   ± 255.   S uppo se   that   the   li m i ts   on   the   valve s   co ntr ol   sig na ls   are    = 255   and    = + 255 ,   and   ( ) = ( ) . 1 = 0   Con s eq ue ntly ,   the   ine qual it y   const raints   for   fu t ur e   ti m e   , = 1 , 2 ,   can   be   e xpresse d   as:     255  + ( 1 ) + 255   (26)     W he re ,       = [         1 ( ) 1 = 0 0 2 0 0 1 2 ( ) 1 = 0 0 0 1 0 2 ( ) 1 = 0 ]               W w he re   ( 1 )   is   the   previ ou s   co ntr ol   sig nal   a nd   0   is   a   row   vecto r   with   dim ension   as   in   ( 0 ) .       4.3.     The   desi gn   of   n on li ne ar   gain   c on s trai ned   m od el   pr edic tive   c ontr oller   This   st ud y   inc orp or at ed   the   nonlinea r   gai n   in   a   co ntr oller   a lgorit hm   kn ow n   as   the   co ns tr ai ned  m od el   pr e dicti ve  co nt ro ll er  (CMPC )   as   a   ne w   ap proac h   to   en ha nce   the   pe rfo rm ance   of   pn eum atic   po sit ion i ng   syst e m ,   especial ly   in   its   transient   respo ns e.   The   justi ficat ion s   of   inc orp orat ing   the   nonl inear   gai n   within   the   con t ro ll er   al gorithm   as   a   new   appr oach   f or   t his   stu dy   inclu ded   t he   f ollo w ing :   1)   it   is   ch al le ng in g   to   pr ov i de   good   transie nt   respon se   with   the   us e   of   li ne ar   co ntro ll er   (which   in   this   case,   the   co nventional   c on st r ai ned   MPC )   for   the   s yst e m   (in   ot her   w ords,   it   is   im po s sible   to   ac hi eve   high   s pee d   respo ns e   without   ov e rs hoot   with   the   us e   of   li ne ar   co ntr oller,   pa rtic ularly   in   r eal - tim e   env iron m ent)   due   to   the   existe nce   of   nonlinea riti es   an d   un ce rtai nties   in   the   syst e m   and   2)   the   inc lusio n   of   co nst raints   in   the   con t ro ll er   al go rithm   deg ra de d   the   respo ns e   ti m e   of   the   syst em ;   t hu s ,   m aking   the   syst e m   slow e r.     This   stu dy   al s o   co ns i der e d   t hat   the   us e   of   li near   c on tr ol le r   rem ai ns   relevan t   in   c ontrolli ng   the   po sit io n   of   t he   pn e um at ic   sy stem   in   this   stud y   a nd   certai n   m od ific at ions   can   im pr ove   the   pe rfo rm a nce   of   syst e m .   Thu s,   the   unde rly in g   pri nci ple   to   this   pro posed   con t ro l   strat e gy   f or   this   st ud y   was   to   e m plo y   nonlinea r   el e m ents   in   the   li near   co ntro ll er   schem e,   wh ic h   com pen sat e   for   er r or   va riat ion s   in   the   pne um atic   po sit io ning   syst e m ,   wh il e   si mu lt aneously   i m pro ving   the   perform ance   of   syst e m .   Figu re   5   il lustrate s   the   blo c k   diag ram   of   the   pn e um at ic   po sit ion in g   syst em   with   the   pr opose d   c ontrol   stra te gy.     As  s how i F igure  5,   a   f unc ti on   cal le no nl inear  gain   wa em plo ye in   this  stu dy  to   c om pen sat e   the  no nlinearit ie an un ce rta inti es  in  the   sy stem   par am et e rs.   It  was  uti li zed  to   c on tr ol  t he  e rror  sig nal   ( )   betwee the   pr edict ed   outp ut   ( )   an set - po i nt  sig nal  ( )   i (24 ).  It   is  of  utm os im po rta nce   to   c on tr ol  ( )   due  to  it in fl uen ce   on  t he  f or m at ion   of  th co ntro sig na to  the  pn e uma ti syst e m T his  te ch niq ue  wa s   e m plo ye in  t his  stu dy  to  a dju st  the   co ntr oller  gai acc ordin to  t he  ou t pu pro du c ed  f ro m   this  f un ct io n,   wh ic is  know as  the scale d err or  ( )   as d esc ribed in  ( 27).   Evaluation Warning : The document was created with Spire.PDF for Python.
Ind on esi a J  E le c Eng &  Co m Sci     IS S N:  25 02 - 4752       En hance men t   in   pn e umatic   posit ion i ng   syst em   us in g   nonli near   ga i n   c on s traine d   ( Siti   Fatim ah   Sulai man )   1393   ( ) =  ( ) × ( )   (27)     W he re,        ( ) = e xp (  ) + e xp   (  ) 2   (28)     = {  ×  ( )   | |  | | >              Figure   5.   Bl oc k   diag ram   of   the   pro posed   co nt ro l   strat e gy       The   value   of   nonlinea r   gain   f un ct io n    ( )   in   ( 28)   was   a dju ste d   accor ding   to   the   er r or     from   the   syst e m ,   wh il e   the   par am et er   va lues   of   bo t h   r at e   var ia ti on   of   nonlinea r   gain   ( )   an d   var ia ti on   of   er ror   (  )   wer e   sel ect ed   by   the   us er .    ( )   act s   as   a   nonlinear   f unct ion   of   er r or     an d   is   bo unde d   in   t he   s ect or   as   descr i bed   in   ( 29) .   Fi gure  il lustrate s t he  rel at ion s hip   betw een     an d   .     0  ( )  (  )   (29)           Figure   6.   Re la ti on s hip   bet wee n      an d         The   inc orp or at ion   of   a   nonlinear   gain   func ti on   into   t he   c on t ro ll er   al go r it h m   in   (25)   ge ner at es   the   pro po se d   c on t r ol   strat egy   ( 30).   T his  stu dy  pe rfor m ed  the  f ollow i ng   proce dures  in  Fig ure  to  determ i ne  the   par am et er v al ue s of       ( ) = [ 1 1 ( ) + 2 2 ( ) + 3 3 ( ) + 4 (  ( ) × ( ) ) ]   + ( 1 )   (30)     As  s how i Figure  7,  pri or  to   the  sel ect ion   of  the   val ue  of  pa ram eter   an    (28),   the  m axi m u m   valu of     for  sta bili ty   m us be  ob t ai ned .   It  was  r eveale i t he  cl os ed - lo op  sta bili ty   us ing  Jur sta bili ty   te st  th at   the  pe rfor m ance  of   t he  pro po s ed  c ontr ol  s yst e m   te nd   to  un sta ble  w he  2 . 96 Th us,  it   is  rec omm end ed  t hat     is  within   the   ra nge  of  0 <  < 2 . 693   f or  syst em   sta bili ty Ba s ed  on  t he   pri or   proce dures  a nd  te sts,  t he  rec omm end ed   val ue  of     an    w ere  set   at   12   a nd  0.1 res pect ively Table   descr i bes  t he p aram et ers  of  t he  prop os ed  con trol strate gy u s ed  in  this st ud y   Ba sed  on  ( 28) ,   w hen  = 12   an  = 0.1  (  ) = 1 . 811 ,   w hich  im plies  that  it   is  with in  the   sta ble  re gion.   Ge neral ly hi gh e value   of     co ntri bu te the   hi gh e st  over sho ot,  wh ic le a ds   t   P n e u m a t i c   s y s t e m C o n v e n t i o n a l   M P C   w i t h   o b s e r v e r   s y s t e m N o n l i n e a r   g a i n   S e t - p o i n t   ( i n p u t ) S t r o k e   p o s i t i o n   ( m m ) + -   N o n l i n e a r   g a i n ,   k n l ( e ) 0 1 E r r o r ,   e + e m a x - e m a x + 0 . 1 0 - 0 . 1 0 1 . 8 1 Evaluation Warning : The document was created with Spire.PDF for Python.
                          IS S N :   2502 - 4752   Ind on esi a J  E le c Eng &  Co m Sci,   Vo l.   23 , N o.   3 Se ptem ber   20 21 13 85   -   139 7   1394   un c ontr ollable   an un sta ble  respo ns in   th syst e m In   a dd it io n,   util isa ti on   of   t his  te c hn i qu e   aut om a ti cal ly  adjusts  t he  val ue  of     acco rd i ng  to  t he  ge ne ra te val ue  of  ( )   at   each  ti m inst ant.  Wh e n   the r is  no  er ror   pr ese nt,   ( ) = 1 . In ot her   w ords, the  con t ro ll er  syst em  r eact s si m il a rly  to  ot her co nventio nal MPC  w it hout   the  presence  of      in  the  co ntr ol le syst e m Howev e r,   with  th pr ese nce  of   e rror   i the  syst e m   and   inclu sion  of   the   c ontro ll er  sig nal  will   be   a dju ste a ccordin gly  to   t he  val ue  of   ( ) .T he  key   ad va ntage   of  t his   te chn iq ue  is  th e fact the  contr oller  gain valu e, expect   for  t he  v al ue of     an  , does  not re quire t un i ng.            Figure   7.   Proce dures   to   deter m ine   the   pa ra m et er   values   of   nonlinear   gai n   (  )       Table   1 .   Desc ription   of   t he   c ontr oller’s   pa ra m et ers   Co n trol   strateg ies   Co n trol   p ara m ete r s   Na m e   of   p ara m et e r   Ab b reviatio n   Valu e       MPC   Predictio n   h o rizon     20   Co n trol   h o rizon     3   Scalin g   f acto r   of   Lagu err e   n etwo rk     0 .1   W eig h tin g   m atr ix     0 .1   Ob serv er   Ob serv er   p o les   -   0 .01 0 0 ,   0 .0105 ,   0 .01 1 0 ,   0 .0115   No n lin ear   g ain   f u n ctio n   Variation   of   n o n lin ear   g ain     12   Variation   of   err o r      0 .1       5.   RESU LT S   A ND   DI SCUS S ION   The   ca pa bili ties   of   t he   pro po s ed   co ntr ol   strat egy   ( NG - CM PC)   to   c on t ro l   a nd   to   im pr ov e   the   transient   res pons e   perform ance   of   the   pneu m at ic   po sit ion i ng   syst em ,   in   a   real - ti m e   env iro nm ent   are   ev al uated   and   disc us se d   in   this   sect ion.   The   posit ion in g   c on t ro l   perf orm ance   for   different   distances   us in g   t he   pro pose d   con t ro l   strat e gy   (NG - CM PC )   was   prese nted   an d   e valua te d.   T he   pe rfor m ances   of   the   pr opos e d   c on t rol   strat egy   wer e   al so   analy sed   and   c om par ed   to   the   existi ng   m et ho ds   of   si m il ar   pn eum at i c   plant   syst e m s   (r efe r   to   pneum at ic   s yst e m   us ed   in   this   stud y)   in   order   to   deter m ine   the   i m pr ov em ents.   A   s a m pling   tim e   of   10   ms   was   us e d   in   ex per im enting,   a nd   the   pro po se d   co ntr ol   strat egy   was   de velo ped   us in g   MA TLAB/Si m ulink .   T he   con t ro ll er   pa ra m et ers   sta te d   in   Table   1   wa s   us ed   in   this   s ect ion   in   orde r   to   per f or m   validat ion .   In   thi s   su b - sect ion ,   e xp e rim ents   wer e   co nducted   to   te st   the   i m ple m ent at ion   of   the   propose d   co ntr ol   strat egy   (NG - CM PC)   in   order   to   c ontrol   thep ne um a ti c   po sit ion in g   syst e m   at   diff eren t   distance s.   Seve ral   values   of   posit io n   dist an ce   (50   m m ,   100   mm,   and   150   mm)   wer e   pro vid e d   f or   com par is on,   an d   t he   ste p   sig nal   w as   ap plied   as   t he   in pu t   sign al .   Eac h   te st   was   cond ucted   for   20   s.   T he   perf or m ances   of   the   NG - CM PC   syst e m’s   transie nt   (i.e.   ris e   tim e   ( ) ,   set tl ing   tim e   ( ) ,   ov e rs hoot   (  ) ,   an d   ste ady - sta te   er ror   (  )   in   c on tr olli ng   the   pne um a ti c   po sit io ning   sy stem s   for   al l   di sta nces   we re   then   c om par ed   with   CM PC ,   PFC - O,   a nd   P I.   Fi gure   8   shows   a   com par at ive   vi ew   of   these   st ep   te sts   wh e re   the   po sit io n   di sta nce   was   va ried   f ro m   fu ll y   retract ed   (0   m m )   to   near   fu ll y   retra ct ed   ( 50   m m ),   fu ll y   retract ed   (0   m m )   to   the   act uator   m id - stroke   ( 100   m m ),   an d   f ully   retr act ed   (0   m m )   to   near   f ully   exten ded   ( 150   m m ).   Table   2   s umm arizes   the   data   obta ined   in   Fi gur e   8.     The   ex pe rim en ta res ults  in   Fi gure   a nd  Ta bl dem on stra te   an   inc rease   i t he   res pons e   tim of    and     for  al strat egies  as  the   po sit io dista nce  is  increa s ed.   I oth e w ords,   t he  lo nger  the  posit io ni ng  distance  to  be  reache by  the   cy li nd er  str ok e,  the  longe th tim ta ken   by   al strat egies  to  achieve  it s te ady - sta te   value.   T he   com par ison   betwee the  c ontr oller  strat eg ie sh ows  that  NG - CM PC  suc cessf ully   con t ro ll ed  the  pneum at ic   cy li nd er  str oke   the  fastest   in  order   t achie ve   po sit io ning   distan ce  of   50  m m 10 m m ,   and    F i n d   t h e   c l o s e d - l o o p   c h a r a c t e r i s t i c   e q u a t i o n   o f   I P A   p o s i t i o n i n g   c o n t r o l   s y s t e m D e t e r m i n e   m a x i m u m   v a l u e   o f   n o n l i n e a r   g a i n ,   k n l ( e m a x )   u s i n g   J u r y   s t a b i l i t y   c r i t e r i o n S e l e c t   k   a n d   e m a x   b a s e d   o n   k n l ( e m a x )   Evaluation Warning : The document was created with Spire.PDF for Python.