Summary: | 碩士 === 元智大學 === 電機工程學系 === 105 === This thesis designs and realizes a new type of bicycle balancing system, which uses a flywheel-type balancing device to make the bicycle for reaching the equilibrium state. Such a balancing device design can be used on mobile robots, as the key to drive the robot robust moving technology.
In this thesis, three control algorithms are proposed for comparison, including the mamdani-like state recurrent fuzzy controller (MLSRFC), the mamdani-type fuzzy with adaptive consequent parts controller (MTFACC) and the adaptive backstepping output recurrent TSKCMAC controller (ABORTC). The MLSRFC can capture the dynamic characteristics of the system without the knowledge of the system model. The MTFACC adapts to the self-adjustment of the rule base so that the user does not need to set the rule base by experience, because of the rule base setting based on the system error automatically. The ABORTC has superior learning ability in combination with the cerebellar model. The TSK fuzzy has the advantages of fast dynamic response of system and simple structure. The output recurrent improves the static state of the cerebellar model articulation controller, so that the whole control can be approximated to an ideal controller eventually.
|