Design of an Optimised Cascade Control System for a SEGWAY using Metaheuristic Methods

Abstract

In order to obtain a Master degree in electrical command as an electrical engineering student, we would like to deposit this thesis which presents the modelling and control of a two-wheeled self-balancing robot (SEGWAY) using State Space methods. The SEGWAY is an underactuated and naturally unstable system, based on the inverted pendulum principle, which makes it a challenging control problem. The main goal of this work is to design a robust control system that keeps the robot balanced while also managing its forward motion and turning (YAW) behaviour. The dynamic model was derived from the Lagrangian equations of motion and linearised about the upright equilibrium to obtain two decoupled subsystems: a balance subsystem and a YAW subsystem. Brushless Direct Current (BLDC) motors were selected as actuators and modelled under closed-loop torque control. The open-loop analysis confirmed the system's instability, with a right-half-plane pole at +7.506 rad/s, while the controllability and observability tests verified that the system can be both stabilised and estimated. A cascade control structure was implemented, combining a balance loop, a YAW loop, and an inner torque loop. Linear Quadratic Regulator (LQR) and Linear Quadratic Gaussian (LQG) controllers were designed for balance and heading control, while Proportional-Integral (PI) and FUZZY-PI controllers were used for current regulation. To improve performance, two artificial intelligence methods, Genetic Algorithm (GA) and Particle Swarm Optimisation (PSO), were applied to automatically tune the controller parameters rather than manually. The results showed that both metaheuristic methods clearly outperformed manual tuning in settling time, overshoot, and YAW tracking, with PSO achieving the best heading tracking. The FUZZY-PI controller produced a smoother current response than the fixed PI, reducing motor stress during transients. Finally, a MATLAB animation confirmed that the designed controller successfully handles bidirectional motion and dual-direction turning simultaneously.