依時間順序出現之一連串觀測值，通常會呈現某一型態，而根據所產生的

型態可以作為判斷事件發生的基礎。例如，震波形成原因的判斷﹔追查環

境污染源﹔以及在醫學方面，辨識一個正常人心電圖的型態與患有心臟病

的病人其心電圖的型態…等。對於這些問題，傳統之辨識方法常因前提假

設的限制而失去其準確性。在本文中，我們應用神經網路中的逆向傳播演

算法則來訓練網路，並利用此受過訓練的網路來辨別線性時間數列ARIMA

及非線性時間數列 BL(1,0,1,1)。結果發現，網路對於模擬資料中雙線性

係數介於0.2至$0.8$之間的資料有高達$80\%$以上的辨識能力。而在實例

研究中，我們訓練網路來判斷震波形成的原因，其正確率亦高達80\%以上

。同時，我們也將神經網路應用在環境保護方面，訓練網路來判斷二地區

空氣品質的型態。

A series of observations indexed in time often produces a

pattern that may form a basis for discriminatingetween

different classes of events. For instance, in theeology, what

are the causes of seismic waves? a earthquakesr the nuclear

explosions ?in the eathenics, we can use theethod to inquire

the source which pollutes the air in somelace, and in the

medicine, to distinguish the difference oflectrocardiograms

between a health person and an a patient..ect. In this paper,

we utilize the back-propagation to trainnetwork and use of the

trained networks to judge the linearRIMA(1,0,0) model between

the nonlinear BIL(1,0,1,1) model,e can find that the trained

network has a good recognitionhose accurate rate is above 80\%

for the coefficient of the bilinear model being equal to 0.5 or

0.8. In a living example, we have trained a network to

decidehich is the cause of seismic wave, and the trained

networkhose accurate rate is larger than 80\%. At the same time,

e also applied neural network in environmental protection.

Brockett,R.W. (1976). Volterra series and geometric control theory, A'utomatica
,Vo1.12,167-176.
Chin,L. C. (1985). vVhat is biostatistics? ,Biometrics,41 ,771-775.
Chan,D.Y.C. & Prager,D.(1991).Analysis of time series by neural networks:
IEEE,355-360.
De Gooi.jer,J.G.& Kumar,K.(1992).Some recent developments in nonlinear
time series lllodelling,testing and forecasting.International Journal of
Forecasting ,8,135-156.
Farrugia,S., Yee,H.& Nickolls,P.,( 1991) .Neural networks classification of intracardiac
ECGS,JEEE~1278-1283.
Gedeon,T.D. & Harris,D.(1991).Creating robust networks,JEEE, 2553-2557.
Ghosh,J.) Deuser) L. wI. & Beck,S. D. (1992). A nerual network based
hybird systen1 for detection, characterization , and classification of
short-duration oceanic signals, IEEE Journal Of Oceanic Engineer'
lng ,Vo1.17 No.4 October , 351-363.
Gorman,R.P. & Seinowski,T.J. (1988).Analysis of hidden units 1Il a layered
netwok trained to classify sonar targets,Neural Networks,Vo1.1,75-89.
Granger,C.vV.J.& Anderson)\.P.(1978).An Introd'uction to Bilinear Ti'me Series
M odels,Vandenhoeck and Ruprecht,Gottingent.
Granger,C."\;V.J.(1991).Developments in the nonlinear analysis of econoillic
series.Scand.1. of Econo'mics,93(2) ,263-276.
Guegan,D & Phalll,T.D.(1992).Power of the score test against bilinear tilDe
series models.Statistica Sinica,Vo1.2))57-169.
Kanaya,F.& IVIiyake,S.(1991).Bayes statistical behavior and valid generalization
of pattern classifying neural networks, IEEE transactio'f1 on N e'ural
lVetworks,Vo1.2,No.4,J uly,4 71-475.
Ljung,G.:NI. (1978). On a lneasure of lack of fit In tilne senes l1lodels
Bio'me tries, Vol. 65,297 -303.
Lipplllann,R .P. (1989). Pattern classification uSIng neural net-
works IEEE Corn;munication Magazine.
Mohler ,R. R. (1973). Bilinear control processes ,Acaclenlic Press, New Yorkand London.
Robert,J.S.(1992).Pattern Recognition.
Ruberti,A.,Isidori A. & d'Allessanclro,P.(1972). Theory of bilinear dynam,ical
system,Springer Verlag,Berlin.
Shulllway,R.H.(1988).Applied Statistical Ti'me Series Ana.lysis.
Shibata,R.(1976). Selection of the order of an autoregressive nlodel by
akaike's infonnation criterion ,Biometrics, Vol.63(1) , 117.
Saikkonen,P.& Luukkonen,R.(1988). Lagrange multiplier tests for testing
nonlineari ties in time series lllodeis ,S cand J oural of Statistics, 55-68.
Saikkonen, P. & Luukkonen,R. (1988).Power properties of a tilne series linearity
test against SOllle simpe bilinear alternatives,Statistica Sinica,453-464.
Subba Rao,T. & Gabr,NI.NI.(1984).An Introduction to Bispectral Analysis and
Bilinear Time Series Nlodels;Springer- Verlag,Berlin.
Terence, C.NI. (1992) .NIodelling the seasonal patterns in UI( macroeconomic
tilnes series,Journal of Royal Statistical Society ,61-75.
Tsay,R.S. (1991) .Detecting and modeling nonlinearity in univariate tinle series
analysis,Statistica Sinica,431-451.
Takayuki, Y.,Tetsuro, Y.(1992).Neural networks controller USlllg autotun-ing
methodfor nonlinear function,IEEE Transcation on N ev:ral JVetworks,
3( 4), 595-601.
Tong,H.& Lim,I(.S. (1980).Tlueshold autoregressioIl,limit cycles and cyclical
data. l.Roy. Statist. Soc.Ser.B,42)45-292.
\iVeigend,A.S.& Rumelhart,D.E.(1991).The effective dilnention of space of
hidden U nits,IEEE,2069-207 4.
Yee E. & Ho J.( 1990). Neural nenwork recognition And classification of
aecrosol distri butions nleasured with a two-spot laser velocimeter ,A pplied
Opiics ,2929-2938.
Zaknich,A. & Attikiouzel,Y. (1991) . A 1110dified probabilistic neural network
(PNN) for nonlinear ti111e series analysis, IEEE ,1530-1535.