Self-Tuning Generalized Predictive Control For Dual-Rate Sampled-Data Systems

The conventional discrete-time sampled-data systems, whose outputs are sampled at the same rates as the control updating rates, are called single-rate systems. But, in many chemical industry processes, because of the hardware limitation, the sampled outputs period is longer than the control updating period. In this case, there are two or even more groups of different sampled-data in a system, such systems are named as dual-rate (multi-rate) sampled-data systems. Therefore, how to control these class of systems effectively is not only significant in theory. but also potential values in applications. Base on some existing identification and control algorithms of dual-rate (multi-rate) systems, This thesis discusses the control problem of dual-rate systems with different noises. Combining the identification algorithms of dula-rate systems with generalized predictive control algorithms, this thesis proposes some new control methods of dual-rate sampled-data systems, and the results are as follows.1. Aim at the certain dual-rate sampled-data system models whose parameters are unknown, this thesis shows the strategy of self-tuning generalized predictive control based on the parameters estimation, uses the recursive least square identification algorithm to estimate the parameters of system models, and proposes self-tuning general predictive control method for the certain dual-rate systems.2. Aim at dual-rate stochastic system models (ARX system models), by using recursive least square identification algorithm to estimate the parameters of system models and applying self-tuning generalized predictive control theory to control systems, this thesis proposes self-tuning general predictive control method based on the least square identification. In addition, a polynomial transformation technique is used to educe dual-rate mathematic as models, and self-tuning generalized predictive control methods based on the polynomial transformation technique and least square identification algorithm is proposed. The simulation examples demonstrate that the proposed control algorithms can track the reference trail of the systems.3. Aim at dual-rate stochastic systems, by using stochastic gradient identification algorithm and self-tuning generalized predictive control alogrithm to estimate and control systems, this thesis proposes self-tuning generalized predictive control method based on the stochastic gradient identification algorithm and the polynomial transformation technique. The simulation examples are pressented.4. This thesis studys the predictive control problem of dual-rate stochastic systems which have colored noises (ARMAX models), further. By using a polynomial transformation technique, extended least square identification algorithm and extended stochastic gradient identification algorithm to estimate these class of systems, and by applying self-tuning generalized predictive control algorithm to control them, a self-tuning generalized predictive control method for dual-rate systems which have colored noises is proposed. The simulation examples are presented. Finally, the thesis gives a simple conclusion, proposes some solvable problems and points out the direction for further study.