Adaptive control using test signals in the feedback channel (Q1081580)

An identification technique for a feedback process is proposed and substantiated. The process function is in discrete time and is described by a linear difference equation. It relies on uncorrelated test signals in the feedback loop. The basis of this parametric identification method is proved under very general conditions, thus making it suitable for solving adaptive control problems where current estimates of unknown process parameters are employed to generate control functions. The intensity of the test signal decreases with time, and this leads to consistent estimates of the unknown process parameters and to the design of an extreme optimal control. The adaptive control thus designed is proved to be operational both by a computer simulation and theoretically.