Cuff Dynamics System Analysis and Modeling

This thesis investigates the properties of the cuff dynamics system. The approach used has been to first build up a linear model for the cuff dynamics system. Analysis of the results shows that the linear model can only hold over a very small operating range, the conclusion is drawn that the cuff dynamics system exhibits strong non-linearity. An artificial neural network then is proposed to model the non-linear cuff dynamics system. Mathematical analysis of the results shows that the model structure provides a better representation of the system dynamics. Two experiments are designed to capture the non-linearity of the cuff dynamics system using a NNARX neural network model. The single operating point of cuff dynamics approximation and the multiple operating point cuff dynamics approximation. A second order with one zero model is chosen as the best representation. The result of the simulations shows that it is not appropriate to use the cuff as sensor in the blood pressure measurement without considering the behaviour of the cuff. The cuff dynamics shows strongly non-linear properties, which contribute a lot to the whole blood pressure measurement.