Study of the Motion Control of the Two-Wheel Mobile Vehicle System Based on Fuzzy Neural Networks

• Main Authors: Tsai-Jiun Ren, 任才俊
• Other Authors: Tien-Chi Chen
• Format: Others
• Language: en_US
• Published: 2007
• Online Access: http://ndltd.ncl.edu.tw/handle/13354399930510427959

博士 === 國立成功大學 === 工程科學系碩博士班 === 95 === This dissertation presents the dynamic model, the motion control and stability analysis of a two-wheel mobile vehicle (TWMV). The TWMV is driven using two independent wheel motors, upon which a vehicle body is mounted. A mathematical model of the TWMV is derived and established using dynamic analysis and Euler-Lagrangian method. The TWMV is inherently unstable and its position is controlled through the actions of the wheel motors. Vehicle motion depends on both the desired wheels response and the tilt angle. A self-tuning proportional-integral-derivative (STPID) control strategy, based on a deduced model, is proposed for implementing a motion control system that stabilizes the TWMV and follows the desired motion commands. The controller parameters are tuned automatically, on-line, to overcome the disturbances and parameter variations. Since the STPID control scheme is not very complex, the system is easy to implement and is not taken much computing time. The TWMV is used as a transport, which works on the level ground or an incline. These motions of TWMV may be operated in more wide tilt angle range; especially it’s on an incline. Therefore, this dissertation proposes a new robust model reference motion control scheme of the TWMV for enhancing the system robustness. The robust model reference fuzzy neural networks control (RMRFNNC), consisted of a model reference fuzzy neural networks controller (MRFNNC) and a robust fuzzy neural networks compensator (RBFNNC), is proposed for implementing a motion control system that makes the TWMV be stable and traces desired tilt angle smoothly. According to the error between the reference model and the linearization model output, it will real-time adjust the weights in the MRFNNC using the update rule such that the overall system follows the trajectory of the reference model. The RBFNNC is designed to resist the system parameter variations and external disturbances of TWMV. It can provide a compensated force to TWMV that adjust its output response matching the linearization model. Combine MRFNNC and RBFNNC, the TWMV will depend on the reference model, procure the reference model control and enhance the system robustness on the level ground or an incline. Whether the TWMV internal parameters exhibits variations, suffers external disturbances or changes the reference model, RBFNNC and MRFNNC will immediately update its weights and membership functions, keeping system be stable and have a high performance. Applying the designed fuzzy differential controller, the TWMV can rotate right and left as desired. Computer simulations and experimental results demonstrate the reliability of the TWMV dynamic model and effectiveness of the proposed control schemes.