Indirect adaptive pole-placement control of MIMO stochasticsystems: self-tuning results

In this paper, we consider indirect adaptive pole-placement control (APPC) of linear multivariable stochastic systems. Instead of the canonical representation often used in the literature, we propose using a non-minimal but otherwise uniquely identifiable pseudo-canonical parameterization that is more suitable for multivariable ARMAX model identification. To identify the plant, we use the weighted extended least-squares (WELS) algorithm, a least-squares method with slowly decreasing weights which was introduced in Bercu (1995). The pole-placement controller parameters are then calculated by using a certain perturbation of the parameter estimates such that the linear models corresponding to the perturbed estimates are uniformly controllable and observable. We prove that with a reasonable amount of prior information, the resulting APPC scheme is globally stabilizing and asymptotically self-tuning regardless of the degree of persistency of external excitation. These results represent the most complete study of stochastic multivariable APPC systems to this date     

On the convergence of least squares estimates in white noise

The problem of convergence of least squares (LS) estimates in a stochastic linear regression model with white noise is considered. It is well known that if the parameter estimates are known to converge, the convergence analysis for many adaptive systems can be rendered considerably less arduous. For an important case where the regression vector is a measurable function of the observations and the noise is Gaussian, it has been shown, by using a Bayesian embedding argument, that the LS estimates converge almost surely for almost all true parameters in the parameter space except for a zero-measure set. However, nothing can be said about a particular given system, which is usually the objective. It has long been conjectured that such a “bad” zero measure set in the parameter space does not actually exist. A conclusive answer to this important question is provided and it is shown that the set can indeed exist. This then shows that to provide conclusive convergence results for stochastic adaptive systems, it is necessary to resort to a sample pathwise analysis instead of the Bayesian embedding approach