Adaptive control of input-constrained type-1 plants stabilization and tracking (Q5926167)

The authors consider the problem of controlling a non-minimum phase plant in the presence of model uncertainty and a saturation constraint imposed on the control input. The controlled plant is described by a model of the form \[ y(t)=\frac{B(q^{-1})}{A(q^{-1})\Delta(q^{-1})}(1+H)u(t)+w(t), \] \[ u(t) = \text{sat}(v(t)) = :\text{sign}(v(t))\min(|v(t)|, u_M), \] \[ A(q^{-1})= 1 + \sum_{i=1}^{n-1}a_iq^{-i},\quad B(q^{-1})=\sum_{i=1}^{n}b_iq^{-i},\quad \Delta(q^{-1})=1-q^{-1}, \] where \(u(t)\), \(y(t)\) denote the input and output, respectively; \(v(t)\) is the control action to be generated by a regulator, \(u_M\) measures the maximal allowed value of \(u(t)\) and \(w(t)\) represents an external perturbation. The operator \(H\) is arbitrary (i.e. linear or not, time-invariant or not) but is \(l_\infty\)-stable. The coefficients \(a_i\) and \(b_i\) are the plant parameters and \(n\) is the plant order. It is assumed that the plant order \(n\) and the sign of the (static) gain \(B(1)/A(1)\) are known; \(\Delta(q^{-1})\) and \(B(q^{-1})\) are coprime; \(A(q^{-1})\) is a Hurwitz operator; one knows a real number \(\rho\) such that \(\|\theta^\ast\|\leq\rho,\) where \(\theta^\ast=[a_1\dots a_{n-1}\) \(b_1\dots b_n]^T.\) The control objectives are the following: robust global asymptotic stabilization of the controlled plant with respect to a multiplicative modeling error, output regulation in presence of external perturbations and tracking of output reference trajectories. The problem is dealt with using a saturated (indirect) adaptive regulator, based on a specific pole placement. The slower the reference trajectory and the perturbation sequence, the better the tracking and regulation quality.