Weighted moving identification and control for multivariable systems

Successive parameter identification via recursive formulas is developed by using a weighted least-squares estimation. An adaptive model is obtained. Two sets of weighting factors are used to ensure that the adaptive model has a good tracking property and that estimated parameters are convergent. Based on the model, an adaptive model-following control algorithm is established. A nonisothermal stirred tank reactor is presented to demonstrate the satisfactory results that the method provides.