Bilinear generalized predictive control: an adaptive approach

The present paper presents an adaptive implementation of bilinear compensated generalized predictive control. The bilinear compensated model adds a compensation term in bilinear model, which minimize the prediction error variance generated by its linearization. Since the relation between the prediction error and the control signal is nonlinear, and the compensation term model is linear, a different model may be estimated to each different level of control signal. Thus, in the present approach, is proposed the adaptive calculation of compensation term. The estimation of compensation term is obtained by recursive least squares algorithm (RLS).