Dynamic Modeling and ANFIS controller design to Enhance Stability of Self-Balancing Two-Wheeled Electric Scooter on Uneven Terrain

Abstract

This thesis looks at the dynamic modeling and design of an optimal controller for a Segway, focusing on important features like self-balancing and steering control, especially on uneven surfaces. Self-balancing is crucial for stability during movement, while steering control is essential for accurate and responsive direction changes on rough terrain. The study starts with a detailed analysis of how the Segway's mechanical, electrical, and control systems interact under different ground conditions. A mathematical model is created to understand the Segway's behavior in various situations. To improve these functions, STSMC is developed to handle balancing and steering. The STSMC parameters are fine-tuned using Teaching-Learning-Based Optimization (TLBO). The results from these tuned parameters are then used to train ANFIS for pitch angle control and a Neural Network (NN) for yaw angle control. The ANFIS_STSMC achieves a settling time of 0.1802 seconds and a rise time of 0.2538 seconds for balance angle control, with no steady- state error. For yaw angle control, the NN controller trained with STSMC data has a settling time of 0.2994 seconds and a rise time of 0.3495 seconds, also with no steady-state error and shows an overshoot of 0.16% for balance angle and 0.05% for direction angle. The performance of these controllers is tested using MATLAB/Simulink. The results show that ANFIS_STSMC and NN controllers greatly enhance the Segway's stability and steering ability, especially on uneven terrain.