On-line controller autocalibration based on parameter predictors: a case study

In this paper, a method for the on-line autocalibration of a controller, based on statistical regression techniques, is presented. The control of neuromuscular blockade is used as a case study. The method relies on the information about the patients dynamics, inferred from the response induced by the initial bolus dose given in the beginning of anaesthesia. A new approach to the characterization of the bolus response based on Walsh-Fourier spectral analysis is considered. A detailed comparison of alternative descriptions is presented. Key-Words: Control Application, On-line Autocalibration, PCA, Regression models, Walsh-Fourier, Wiener models.