Self-Tuning Control based on I p io 12 sin — Penalized Least Squares

In self-tuning control problems, the parameter estimate can exhibit undesirable behaviors due to the possible lack of information in some directions of the parameter space. In this contribution, a penalized technique able to enforce certain desired properties to the estimates is introduced. Its usefulness in an adaptive control context is also discussed. 1 The identification algorithm Consider the discrete time S1S0 system governed by the equation /i(o”; q-’) ~, = B(e”; q-1) U,-.j + n,, d~l (1)