Real-time iterative learning control-two applications with time scales between years and nanoseconds

摘要

Iterative learning control (ILC) is a family of digital control concepts, which can be used for a large variety of different applications. Each application has its own properties like sampling time and storage needs. This paper shows two real-time ILC applications with different time scales and storage demands. First, the cavities of one of the world's leading pulsed free-electron laser are controlled by a norm-optimal ILC using only the information about the last pulse but with sample times below microseconds. Second, a heating system is controlled by a data-driven ILC with a sample time in the range of minutes but using all available historic data sets of past trials. Tensor decomposition methods for storage demand and complexity reduction are applied to both applications, which results in a norm-optimal tensor ILC, as well as, a data-driven tensor ILC, although the time constants for the two applications vary by eight orders of magnitude.