Adaptive decoupling of a class of multivariable dynamic systems using output feedback

The dynamic behaviour of a multivariable industrial system is rarely, if ever, completely known, owing to parameter uncertainties, measurement inaccuracies and other fluctuations. In some cases, the dynamic behaviour may be slowly time-varying. The design of constant precompensators or low-order dynamic compensators through a one-shot procedure to achieve exact decoupling of such systems is neither realistic nor practical. In the paper an adaptive decoupling algorithm is proposed for controller design, by specifying the closed-loop behaviour of the system in the form of a reference model, so that the controller parameters can be estimated online as the process develops. The approach uses the deterministic autoregressive moving average (DARMA) model for system representation. The performance of the algorithm is assessed using both dynamic and steady-state decoupling criteria. Simulation results demonstrate the ability of the algorithm to adapt the controller parameters to changes in the system parameters. Therefore, the algorithm will be more attractive for decoupling either unknown or slowly time-varying systems than the one shot technique of designing constant precompensators or low-order dynamic compensators to achieve exact decoupling.