Design and implementation of flexible model based adaptive control algorithm for real time systems

This thesis proposes a flexible adaptive model based control algorithm for real time systems. The algorithm consists of a base controller and adaptation mechanisms. Parameters of base controller are updated with a flexible adaptation mechanism which tracks error of a plant and model output. The reference model is defined for a specific interval of the plant response and parameters of the reference model are updated for predefined interval trough identifications of plant input and output. In order to minimize the output error between model and the plant, Lyapunov based adaptation rules are designed. In addition, a fuzzy algorithm provides relation between the Lyapunov rules and base controller parameters. The proposed control algorithm does not need prior information of the dynamic system and has a flexible adaptation mechanism. Therefore, the algorithm can be easily implemented to different dynamic systems with minor modification. The algorithm is implemented for three different real time plants and experimental results are compared with base controller. Proportional–Integral-Derivative (PID) controller is used as base controller. The adaptive control algorithm needs simultaneous multi-processing for implementation, therefore some part of algorithm are run on a microcontroller and the others are run on a computer which has operation system. The experimental results show that the proposed algorithm tracks the system output in an acceptable error margin.