Decentralized neuro-fuzzy controller design using decoupled sliding-mode structure for two-dimensional inverted pendulum

In this paper, a new method has been proposed to control a two-dimensional inverted pendulum. First, the system of a two-dimensional inverted pendulum is divided into two subsystems using decentralized control theory. Then, using decoupling method, each subsystem is decoupled into two surfaces for applying sliding-mode control. Next, this controller has been used to train two neuro-fuzzy ANFIS (Adaptive-Network-Based Fuzzy Inference System) networks. Due to the high accuracy of the ANFIS networks, these two networks can learn the controlling abilities of the teacher. Each trained network controls its own subsystem as a local controller. The trained networks not only have the properties of their teacher, but also due to the parallel processing property of the ANFIS networks, they response much faster than their teacher. Moreover, the neuro-fuzzy controller doesn't need any model of the plant or its parameters. Simulation results show a high performance for the proposed method as compared to the existing methods.