Optimized neuro observer-based sliding mode control for a nonlinear system using fuzzy static sliding surface

Abstract

In this paper, a robust controller for the complete model of the ball and beam system (BBS) is proposed. The aim is to control the position of the ball and also to balance the beam’s angle. Sliding Mode Control (SMC), Fuzzy Logic (FL) and Multi-Layer Perceptron Neural Network (MLPNN) are the techniques utilized in this work. In the presented controller, a fuzzy sliding surface is used to reduce the states’ tracking errors, along with a boundary layer in the control law to decrease the chattering phenomenon of SMC. Also, a neuro observer-based (NOB) controller is designed to estimate the system’s states with assumption of lack of a sensor for reporting the position of the ball. The controller stabilizes the system very well and its stability is proved in detail considering uncertainties. To improve the performance and speed up the system, the controller is optimized using ten single-objective and two multi-objective optimization algorithms. The proposed controller avoids high computational costs and provides less control effort, also reduces the errors and the effects of chattering and all system’s states converge to their desired values in a suitable time domain. The performance of the controller is compared to PID, PD, and fuzzy controllers. To show the effectiveness of the designed controller, a parametric uncertainty and an external disturbance are added. The results indicate an efficient performance and robustness of the designed controllers.