Internal model control based on improved extreme learning machine

An improved extreme learning machine is proposed to avoid the over-fitting in learning. Then, a novel internal model control strategy is designed by the improved extreme learning machine. Specifically, use the improved extreme learning machine to estimate the internal model, and the model is expanded by Taylor series to calculate the controller in the internal model control system. Moreover, the stability and the error are also easy to analyze for the proposed internal model control system. Finally, the proposed strategy is simulated to control the consistency in the continuous stirred tank reactor. The experimental results indicate that the proposed control strategy yields an excellent system performance with a small stable error and a strong robustness. © 2013.