ANFIS-Based PD-Fuzzy Control for Pendubot Stabilization

Abstract

The Pendubot is a nonlinear underactuated system with two rotational degrees of freedom and a single actuator, making stabilization a challenging problem for intelligent and nonlinear control methods. Reliable balancing control is important for evaluating control strategies under both simulation and practical hardware constraints. However, although fuzzy and intelligent control methods have been investigated for Pendubot systems, the practical implementation and experimental behavior of ANFIS-based PD-Fuzzy control remain insufficiently documented, particularly in comparison with a conventional PD controller under the same laboratory conditions. This study aims to develop and evaluate an ANFIS-based PD-Fuzzy controller for TOP-position stabilization of a Pendubot. The nonlinear dynamics are formulated using the Euler-Lagrange method, while controllability of the linearized model is examined at the TOP and MID equilibrium points. The proposed controller uses ANFIS-based fuzzy blocks to approximate the proportional control actions, while derivative paths remain explicitly implemented. MATLAB/Simulink simulations show that the baseline PD and PD-Fuzzy controllers produce closely matched TOP-balancing responses, with settling times of approximately 2.36 and 2.37 s for link 1, respectively. Experimental evaluation using an STM32F407-based platform demonstrates practical TOP balancing; however, the baseline PD controller provides more favorable behavior than the PD-Fuzzy realization under the reported hardware conditions. These findings indicate that ANFIS-based PD-Fuzzy control is feasible for Pendubot stabilization but does not necessarily provide performance improvement over a conventional PD controller. The study therefore highlights the importance of hardware-aware tuning, sensor quality, and broader ANFIS training data for future improvements.