On Learning Control For Systems With Unknown And Verying Time-delays

In the actual control processes, there always are unknown and varying time delay. With the improvement of industry, the processes are more and more complex. The influence of time-delays can not be ignored. This will influence the safety and efficiency of industrial processes. Research.ing on the control of time-delay systems will be of realistic significance for improving the automatic level of industrial processes, and their safe economical operation. Applying the Iterative Learning Control (ILC) to the system with time-delay will make it possible to adopt an intelligent control method to design a control system with unknown and varying time-delay. So the method is valuable in application as well as in the theory. The main contents of this thesis are as follows: 1.. For a better learning transient along the iteration number direction, a predicted PID-type iterative learning control algorithm is proposed for a class of systems with unknown and varying time- delays. The convergence conditions for the proposed predicted learning control have been established. The effectiveness of the predicted ILC algorithm is demonstrated by simulation examples. 2.. A large number of simulation examples are provided to manifest the algorithm's better performances in parameter robustness with comparison to the traditional P11) controller. The comparisons mainly focus on instances of parameter changes, delay-time change and order changes. 3.. A large number of simulation examples are provided to manifest the algorithm's better performance in some circumstances with comparison to Smith controller. The circumstances include instances when there are disturbances in the system, when the parameters of the object controlled are not matching with those of the model and when the transfer function of the object is non- minimum phase. 4 .. In order to improve perfonnances of the predicted ILC algorithm , especially when the delay-times are unknown and time- varying, we add a method to estate the parameters of the controlling object especially the time-delays. In this thesis we use Gradient Estimation Algorithm(GEA) to estimate parameters of objects controlled.