Iterative Learning Control For Time-Delay Systems

As the phenomenon existed universally in the real controlled objects,time-delay is generated by some actions from transmission, measure andso on. And often results in the instability and poor performance. So it ismeaningful and valuable to study the control problem for time-delaysystems.Iterative learning control theory is suitable for the process which hasrepetitive characteristic. It exploits every possibility to incorporate pastcontrol information: the past tracking error signals and in particular thepast control input signals, into the construction of the present controlaction. The system output converges to the desired trajectory jointedsmoothly finally. As a result, it is feasible to solve the control problem fortime-delay systems with iterative learning control theory. This dissertationaddresses the problem of iterative learning control for time-delay systemswith well-defined relative degree. Multiform learning algorithms areintroduced. Convergence conditions are derived, ensuring that the systemoutput converges uniformly to a pre-specified trajectory. The maincontributions include the following aspects:(1) The dissertation addresses the problem of iterative learningcontrol for time-delay systems with well-defined relative degree. Initialrectifying action is introduced in a conventional learning algorithm in thepresence of initial shifts. Convergence conditions are derived, ensuring thatthe system output converges uniformly to a pre-specified trajectory. Inparticular, the system output converges to the desired trajectory jointedsmoothly with a piece of transient trajectory.(2) The dissertation addresses the problem of closed-loop iterativelearning control for non-linear time-delay systems with well-definedrelative degree. Convergence conditions are derived, ensuring that thesystem output converges uniformly to a pre-specified trajectory. The D-type, PD-type, PID-type learning law and D-type learning law withinitial rectifying action are discussed respectively in the dissertation.(3) The dissertation addresses the problem of non-linear time-delaysystems with well-defined relative degree. The sampled-data learningalgorithm is initiated and its robust convergence condition is derived.(4) Combining iterative learning control with anticipatory approach,the dissertation addresses the problem of anticipatory iterative learningcontrol for non-linear time-delay systems with well-defined relative degree.Convergence condition of the anticipatory iterative learning control isderived.