The research presented in this work addresses the time-delay problem in control systems. A system is considered to be a time-delay system, if it contains at least one component, whose input affects the output not immediately but after a certain time delay. Such time-delay components often appear in control systems and have a negative influence on control. Presently, time delays are often ignored in the design of control systems since they are unknown and their identification is difficult. This holds especially for nonlinear dynamic systems. As a consequence, the system cannot be optimally controlled. This work contributes to solve the time-delay problem. A new method for the identification of arbitrary time delays in nonlinear dynamic systems is presented. This method uses neural networks, that are especially designed for this task, together with particular criteria for the evaluation of the identification reliability. The method enables an easy integration of a priori knowlegde about the time delays. Experimental results are presented showing the successful application of this new method in the fields of microelectronics, automotive industry and food technology.
«
The research presented in this work addresses the time-delay problem in control systems. A system is considered to be a time-delay system, if it contains at least one component, whose input affects the output not immediately but after a certain time delay. Such time-delay components often appear in control systems and have a negative influence on control. Presently, time delays are often ignored in the design of control systems since they are unknown and their identification is difficult. This h...
»