Զѧ
ACTA AUTOMATICA SINICA
1997ꡡ234ڡVol.23No.41997



˹߲ٵ
Ӧϵͳ1)
ءܹˡ
ժҪʵʵʸЧԴϵͳԣʹ߸˺ܴ.һֻ Hopfield Ĳ߸ٷ߸ٽڲͻ 50% ĽϵͳͨתٿӦõ˺ܺõķ.
ؼʡ磬߲٣Ӧ.
A NEURAL NETWORK BASED ON-LINE PARAMETERS TRACKING 
SCHEME FOR AUTOTUNED AC DRIVE SYSTEM
WANG LEIZHOU GUOXINGWU QIDI
(Department of Electrical Engineering,Tongji University,Shanghai 200092)
AbstractBecause of the sensitivity of Vector-Control effect to the AC drive system parameters,it is meaningful to discuss the online tracking of the system parameters.This paper uses the Hopfield neural network for online tracking of the rotor resistance(R2),system rotation inertia(J),speed damping coefficient(R) and load torque(TL),and uses the online tracking results for adaptive tuning of the flux and speed controller of the AC drive system when its parameters suddenly change about 50% at the start time.Simulation results show the effectiveness of the derived method.
Key words Neural network,on-line parameters tracking,autotuned AC drive system.
1Hopfield ڽϵͳ߸
Hopfield 1һŻ磬ЭͬʹС2.ˣֻҪݲٵҪȷӦȷṹ룬ǲ߸ֵ
ڸԪ RiʱHopfield ׼Ϊ

ÿԪ̬ͬʱɱʾΪ

 Vi=g(ui). g-1(u) Ϊ.
ף3Ƶķ֤ڸԪϵԳʱЭͬʹ򻯵Сֵʽ첽ϵͳУֻҪⶨתӵֱܽӵĴų򣬾ͬתϵ任3ݣӵɷֽΪŷ im1תطit1

(1)

(2)
ʵϵͳУ im1Ϊͨ im1c.(1)ɵ

A1p B1p Ϊٵʱϵͳ.Ż·HopfieldԪɣΪ A1s B1s.
ٲ綨Ϊp2s=A1s2+B1sim1c.
Ų

ȻHopfieldܾȷϵͳֵͨA1sB1s ׵õתӵ߸ٽʽ򻯵׼ȽϣɵõϵӶõԪĶ̬

ģԪ(Sģⱥֵ ģѧϰ)ٶܹͨ  еڣ·ͼ1ʾ


ͼ1תӵ·ʾͼ
ͬͨƷԻϵͳתƽ߸ԪĶ̬̣ϵͳתƽΪ1)
1)ⷽμ4ȫԶ̲ļ144151ҳ.

2ٽ趨Ӧϵͳ
߸ٵĴϵͳڴͨתٵӦƣõ˺ܺõķʵ(ͼ2-ͼ4ʾ)ڴϵͳʱͻ 50% ʱ߿Ľϵͳ˺ܺõĴͨתӦƽ(ջŻ·ʱ䳣Ϊ 10msٶȸʱ䳣Ϊ 0.1ms).


ͼ2Ӧͨƽ


ͼ3ٶָ


ͼ4Ӧٶȸٽ
Ͻ֤ķЧ.ĵõĽһĽ׶ԺͿչԣĽչʽϵͳƵоڽһ.
1)ԭڵĽȫԶ̿Ʋ.
ߵλͬôѧϵϺ200092
ο
1ݡTank D W,Hopfield J I.Simple neural networks:an A/D converter,signal decision circuit,and a linear programming circuit,IEEE Circuit and System.1988,533541.
2ݡ.ϵͳ.ӿƴ磬1990.
3ݡ¼.ѧģͼϵͳ.ҵ磬1988.
ոڡ1995-05-25
