Title :
A new Elman neural network and its dynamic properties
Author :
Cheng, Yuan-Chu ; Qi, Wei-Min ; Zhao, Jie
Author_Institution :
Coll. of Power & Mech. Eng., Wuhan Univ., Wuhan
Abstract :
Elman neural network was one of the dynamic recurrent neural networks. In this paper, a new modified Elman network was proposed first. Then the dynamic characteristics of Elman series neural network are fully discussed. The theory study proved that the new Elman network inherently has proportional (P), integral (I), derivative (D) properties and traditional Elman networks only have PI or I characteristics. Simulation results based on ideal mathematical model show that the PID Elman network is prior to the modified Elman network in identifying nonlinear dynamic system.
Keywords :
identification; nonlinear dynamical systems; recurrent neural nets; Elman neural network; dynamic recurrent neural network; mathematical model; nonlinear dynamic system identification; proportional-integral-derivative property; Educational institutions; Feedforward systems; Mathematical model; Mechanical engineering; Mechanical factors; Neural networks; Nonlinear dynamical systems; Output feedback; Recurrent neural networks; System identification; Elman neural network; PID properties; dynamic recurrent network; dynamic system identification;
Conference_Titel :
Cybernetics and Intelligent Systems, 2008 IEEE Conference on
Conference_Location :
Chengdu
Print_ISBN :
978-1-4244-1673-8
Electronic_ISBN :
978-1-4244-1674-5
DOI :
10.1109/ICCIS.2008.4670784