Non-linear dynamic system identification using Cascaded Functional Link Artificial Neural Network

Abstract

The Multilayer Artificial Neural Network (MLANN) has been employed for identification of non-linear dynamic systems. However, this scheme offers high computational complexity and yields poor identification performance particularly for non-linear dynamic systems. In this paper, we introduce a new structure known as Cascaded Functional Link Artificial Neural Network (CFLANN), derive an appropriate learning algorithm and use it for identification task. Extensive simulation study reveals that the proposed approach outperforms the existing MLANN-based method both in terms of computational complexity and response matching.