ứng dụng support vector machines trong chẩn đoán lỗi cho đối tượng công nghiệp

lOIWUKHnHOMMIUHDilllWVNWUnolPUIUI

-^ 1

Hlnh6.0ltu

khiin CjHg pha

hai-cdt theo

Cohtn-Coon vdl

tic df^ng nhiiu

5% dp lifu thip.

Hinh 7. Diiu

khiin CjHg pha

hcH-cdt theo

Cohen • Coon

vdi tdc dpng

5% nhiiu ra

cda ddl tupng.

6.3. Oanh gia chat li/omg ciJa bd dieu khiln

Cie tham sd cCia bd PID 6ugc tfnh toin dua trdn Cohen

Coon. Sau dd 6ugc lip vio md hinh thip kit hpp vdi bd cim

biln mim (soft-sensor) di 6ugc huin luydn. Thdc hi^n md

phdng trdn simulink_matlab vi 6ugc dinh gii bing chit lupng

eua he thdng dilu khiln thdng qua eie chi tidu:

- Thdi gian dip dng (qui trinh qui dd) vdi bd didu khiln PID: 300s.

- Op qui dilu ehlnh vdi bd dilu khiln PID: 23%.

-Sai ldch tinh: 2,63.10-7

Kilm tra kit qui md phdng dip dng vdng kfn bing cich thay

doi tii hoac nhieu 64 dinh gii v l hidu qui cua bd dilu khiln PID.

Suthay ddi 5% tie ddng nhieu trdn ddng cap lieu vdi eie dilm dit

khie nhau trong mdt chu ky ngin. Kit qui chdng minh ring cic

thdng sd 6ugc dilu ehlnh bing Cohen - Coon diu ed chat lupng

tdt va loai bd 6ugc nhdng thay doi v l ndng dp thinh phan khi cd

nhieu tic ddng.

7. K^T LUAN

Oilu khiln true tilp ndng dp thinh phin di 6ugc phit triln

vdi cac kit qui dat 6ugc sau khi di huan luyen mdt mang nPron,

xiy dung mdt bd cim biln mim trdn cP sd tip dd lieu bao trum

Idn toan bd cic dilu kidn van hinh. Md hinh 6ugc dinh gii trdn

tip dd lieu vdi cac biln ddng nhilu khic nhau, vdi cie dilm dit

khie nhau. V tUdng dilu khiln trUc tilp ndng dd thinh phin

dupc dua ra nhim muc dfch dilu khiln, duy tri ndng dd thinh

phan cua thip chung cat. Kit qui dat 6ugc trong nghidn cdu md

phdng di dem lai triln vpng trong dng dung vio thUc t l 64 gdp

phan ning eao chat lUpng sin pham vi hidu suit lim vi^c cho

thip chung eat diu hidn nay. a

TAILICUTHAMKHAO

[1].e>ang VSn Chl, 2010. (Ing dung thuM lan truydn ngu^c cCia m^ng naron nhSn

tao d l xiy difng b$ cSm bi^n m^m trong dir bio ndng d$ sin phSm d c^t thip chung

cjlt diu m6. Bio eio h^i nghj khoa h(?c thiJf 19, TrUdng DH Md - Dja chit.

[2]. Ding Vin Chf, 2003. NghiSn cOfu h^ th6ng dilu khiln thip tich etan COI nhi

miy xir \</ khf Dinh Cd. Luin vin thac st ky thuit.

[3].Lutz, Wendt, 2007. Taschenbuch der Regelungstechnik mit Matlab und

Simulink. 7.,ergaenzte Auflage. Verlag Harri Deutsch.

[4].S.Majhi,L.Litz, 2004. Relay Based Closed-Loop Turning PID

Controllers,"Adaption der PID-Reglerparameter im geschlossenen Krels durch

Einkopplung eines zweipungktreglers".

[5].www.en.wikipedia.org/.../Distillation

52 Tv dong hoa ngay nay / 4, 2012

miG DUNG SUPPORT VECTOR

MACHINES TRONG CHiiN OOAN lil

CHO 06 l TirCfNG CdNG NGHIlP

THS. PHAM MINH HA I

B^ mdn Ti/d^ng hda xf nghidp Md vd Ddu khf

pmhai80@yahoo.com

Tom tJit

Chin doin loi ddng vai trd quan trong trong vifc dim bio

lioat ddng binh thifdng ciia tliilt bj vi qui trinh cong

nghiip. Mdt trong nhiimg hudmg nghien cufu gin day \i siif

dung l(y thuat nhin dang vii hoc tU dong de nhan dang

nhiimg tinh hudng Idi. Bai bio trinh biy ket qui nghien

ciiru vi thiJr nghilm ky thuit support vector machines -

mdt ky thuit nhin dang ttfomg ddi mdi cho bai toan

chan doan ldi. Kit qui thiir nghiem tren diif lieu van hanh

thuc te cua thilt bj bom cdng nghiep cho thay ky thuat

niy phii hop vdi bai toan chan doin loi vi cho ket qui

chinh xic hon phuong phap siir dung luat.

Abstract

Fault diagnosis is one of the most studied problems in

process engineering. Recently, great research interest

has been devoted to approaches that use dassificatwn

methods to detect faults. This paper presents an

application of a newly developed classification

method - support vector machines - for fault. A real set

of operation data of a motor pump was used to train and

test the support vector machines. The experiment results

show that the support vector machines grae higher conect

detection rate of faults in comparison to rule-based

diagnostics.

I.OATVANO^

Bii toin chin doin loi ludn 6ugc quan tim hang

diu trong nghien cdu, xiy dung va van hinh cac doi

tupng cdng nghidp. Vide xic djnh chfnh xic va kip

thdi Idi ddng vai trd quan trpng trong vide duy tri

hidu qui cung nhu chit lUpng cua cie qua trinii

cdng nghidp. Trong nhdng nim gin diy, khi nang

tfnh toin cung vdi nhflng dng dung cOa miy tfnh de

xd ly SO lidu lim cho vide chan doin loi true tiep

ngiy cing dUpc dng dung rdng rai vdi nhieu

phUPng phip dUpc nghidn cdu va sd dung. NhCfng

phupng phip nay dupc phan bidt vdi nhau bdi dang

tri thdc ma moi phUPng phip sd dung v i cich thiJfc

sd dung tri thdc dd.