Effective System Identification using Fused Network and DE Based Training Scheme
Saikat Singha Roy1, Joyshri Das2, Susovan Mondal3
1Saikat Singha Roy, Electronics & Communication Engg.,CITM Boinchi, Hooghly, NIT Durgapore, India
2Joyshri Das, Radio Physics, University of Calcutta, India.
3Susovan Mondal, Electronics & Communication Engg.,SIEM, Siliguri, India.
Manuscript received on June 06, 2013. | Revised Manuscript received on June 28, 2013. | Manuscript published on July 05, 2013. | PP: 105-112 | Volume-3 Issue-3, July 2013. | Retrieval Number: C1667073313/2013©BEIESP
Open Access | Ethics and Policies | Cite
© The Authors. Published By: Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Adaptive direct modeling or system identification finds extensive applications in telecommunication, control system, instrumentation, power system engineering and geophysics. If the plants or systems are nonlinear, dynamic, single-input single-output (SISO), the identification task becomes more difficult. The dynamic system identification task is basically a model estimation process of capturing the dynamics of the system using the measured data. The Functional Link Artificial Neural Network (FLANN) is a single neuron single layer network first proposed by Pao. The structure of the FLANN is simple as it represents a flat net with no hidden layers. Therefore the computation and learning algorithm used in the architecture is straight forward. In the present investigation the identification problem is performed on three standard benchmark nonlinear dynamic series-parallel models using Differential Evolution (DE) for training the weights of FLANN structure. The performance of the proposed FLANN-DE identification model is compared with FLANN-Genetic Algorithm and FLANN-Back Propagation method.
Keywords: Differential Evolution, FLANN, Genetic Algorithm, System Identification.