Nonlinear dynamic system identification and performance measurement using extreme learning machine

This paper describes the identification of a nonlinear time invariant system is quite essential for its stability point of view. Among all the identification approaches, FLANNs is the best one but in case of learning speed and convergence point of view it does not work well. Amongst all the traditional approaches, a newly developed algorithm named ELMs (Extreme Learning Machine) for identification and tracking bears only a hidden layer chooses the input weights as well as threshold values randomly. Due to this above matter of fact ELMs is widely named as SLFNs. This algorithm proofs best in terms of generalization and shows better convergence at a very high learning speed. The experimental result shows that the proposed algorithm is the best one for identification detection and tracking cases which hold an extremely high rate of learning.