Identification of linear systems with delay via a learning model

TL;DRAbstract

X The effects of input delay on an identification scheme using a learning model are investigated. The parameter adjustment laws for the learning model are derived through Lyapunov methods similar to those used for the model reference adaptive control systems of Parks. For no measurement noise or delay mismatch between the learning model and system, the parameters are adjusted to bring the error between model and plant to zero. When there is delay mismatch between the inputs of the learning model and the unknown system, the convergence of the parameters of the learning model to those of the unknown system is no longer guaranteed. However, the error between the learning model and the unknown system is guaranteed to enter and stay within a region close to the origin.

Chat with Paper

AI Agents for this Paper