Identification of nonlinear dynamic systems using differential evolution based update algorithms and Chebyshev functional link artificial neural network

• Published in: Proceedings of third International Conference on Computational Intelligence and Information Technology pp. 508 - 513
• Main Authors: Swayamsiddha, S, Mondal, S, Thethi, H.P
• Format: Conference Proceeding
• Language: English
• Published: Stevenage, UK IET 2013; The Institution of Engineering & Technology
• Subjects: Interpolation and function approximation (numerical analysis); Neural computing techniques; Optimisation techniques; Simulation, modelling and identification; complex nonlinear dynamic system; nonlinear model; genetic algorithms; genetic algorithm; differential evolution; nonlinear systems; back propagation algorithm; nonlinear dynamic system identification; identification; backpropagation; Chebyshev approximation; Chebyshev functional link artificial neural network; neural nets
• ISBN: 9781849198592; 1849198594
• DOI: 10.1049/cp.2013.2637

Practical systems which we see around us are generally nonlinear and/or dynamic in nature, hence are complex. In recent past a lot of work has been done for identifying the parameters of complex nonlinear dynamic systems, but still there is a lot of scope in terms of improved performance. In the present work we have proposed an identification model for nonlinear dynamic system using nonlinear model where the parameters of model are updated using population based update algorithms namely Genetic Algorithm (GA) and Differential Evolution (DE) using Chebyshev Functional Link Artificial Neural Network (CFLANN). The convergence performance is compared with respect to conventionally used back propagation (BP) algorithm. To validate the proposed model we have taken two complex dynamic plants one having nonlinearity in input side and the other plant with nonlinearity in the output side of the plant.