具自反饋和不連續(xù)型信號(hào)傳輸函數(shù)的二元時(shí)滯神經(jīng)網(wǎng)絡(luò)模型的動(dòng)力學(xué)研究.pdf

1、湖南大學(xué)碩士學(xué)位論文具自反饋和不連續(xù)型信號(hào)傳輸函數(shù)的二元時(shí)滯神經(jīng)網(wǎng)絡(luò)模型的動(dòng)力學(xué)研究姓名：茅蘭娣申請(qǐng)學(xué)位級(jí)別：碩士專業(yè)：應(yīng)用數(shù)學(xué)指導(dǎo)教師：黃立宏2002.5.1AbstractThisthesisdiscussestwokindsofselffeedbackartificialneuralnetworksoftwoJ畝=一肛z十，(正(。一7))9(∥(。一7)’(I)19=一pⅣf(x(t—T))9(v(t—T))and，J圣=一pz

2、一，(z(‘一7))9(可。一7)’(II)10=一pⅣ一g(g(￡一r))f(x(t—r))Herez(￡)andg(￡)describesseparatelytheactivestateswhenthetwoneul’onsar8attheexacttimeoft，r0isthesynaptictransmissiondelay，fandgarethesignal，ce，=a一。；；90“11、。ce，=b一6；i。022，lf≤I

3、f曼o≠Oandb≠0aretwogivenconstantsoland02arethresholdconstantsando1曼盯2ThisthesisiscomE)osedoftwochaptersThefirstchapterdealswithonekindoftwoneuronneuraluetworkswithpositiveselffeedbackcoefficientInchapter2，wediscussthesimul

4、arsystemwithnegativeselffeedbackcoefficientAsthediscontinuityof／andgmakesitdifficulttoapplydirectlytheexistingresuItsofdynamicsystemtosystem(I)、(n)，weusetheiterativemethodofone’dimensionalfuncrionandreductiontoabsurdityt

5、omakeclearourmodelsTheresultsshow￡hat，when盯land02areinvariousintervals，thetrajectoriesforsystems(I)areeitherultimatelyperiodicorconvergenttoanequilibriumThedifferencebetweensystem(I)andsystem(II)onlyliesonthesignofselffe

6、edbackcoefficient，buttheirasymptoticbehavioroftrajectoriesaremuchdifferentForthegiveninitialvalue，tilesolutionsforsystem(II)allconvergetothecertainequilibriumForsystem(I)，wealsoverifysonicresultsthroughnulnerica|sinmiati