Study of estimation and optimization techniques suitable for microprocessor adaptive controllers

Adaptive controllers are controllers that perform optimally in unknown or changing environments. One class of adaptive controllers are conventional controllers that tune themselves. This is done by estimating the plant system parameters and optimizing the controller based on these estimates. It is desired to have algorithms that are short in both program length and execution time so that implementation in a device such as a microprocessor is possible. Generalized Geometric Programming (GGP) is used to optimize both PID control of a second order system and lead-lag compensation of a servomotor system. These algorithms normally converge in a few iterations. The parameters of a second order plant are estimated by two techniques. One technique involves curve fitting of a step response with cubic splines to find the coefficients of the characteristic equation. The other technique, called Walsh Function Parameter Identification, (WFPI) uses a square wave test input and finds the phase tangents by correlation of the output with Walsh Functions. In general, each of these algorithms is estimated to require no more than 1000 lines of code with execution times of less than 1 second, once the measured data is available.Applied Science, Faculty ofElectrical and Computer Engineering, Department ofGraduat