Nonlinear Identification Using Neural Network Combined with Training Based on Particle Swarm Optimization

Most processes in industry are characterized by nonlinear and time-varying behavior. In this context, the identification of mathematical models typically nonlinear systems is vital in many fields of engineering. A variety of system identification techniques are applied to the modeling of processes dynamics. Recently, the identification of nonlinear systems by artificial neural networks has been successfully applied. In this paper, an original approach based on radial basis function neural network (RBF-NN) with a training method based on particle swarm optimization (PSO) is proposed as an alternative solution. RBF-NN is considered as a good candidate for the prediction problems due to its rapid learning capacity and, therefore, has been applied successfully to nonlinear time series modeling and nonlinear identification. On the other hand, PSO was inspired by the choreography of a bird flock and can be seen as a distributed behavior algorithm that performs multidimensional search. The RBF-NN model is trained and validated based on the experimental data of a nonlinear process. Finally, simulation results from the performance analysis of RBF-NN are presented and discussed.