Embedded model control: principles and applications. part II

In this second part, model error analysis is further pursued with the help of two theorems. The former one says how to retrieve noise in real-time so as to update Embedded Model. The source can only be the model error. The second theorem explains how the components of the residual model error entangle in the estimated noise, leading to feedback oddities and instability. By formulating the neglected dynamics as one component of the model error, the error loop is constructed and formulated, showing how model error affects stability and how stability may be recovered. Some design guidelines are mentioned leading to a weak separation principle.