Experimental analysis of identification function and stability of an adaptive controller for total artificial heart

The authors' previous study has already shown that it is possible to design, on the basis of the adaptive control theory, a controller of the total artificial heart so that cardiac output may depend on peripheral vascular resistance and so that the controller may automatically cope with individual differences of cardiovascular dynamics and its time-variant characteristics. However, this is only the result from computer simulation. In the present study, the proposed controller was actually implemented on a personal computer system by using the adaptive pole assignment method. Then, capability of parameter identification of the controller and stability of the resulting closed-loop system were concretely analyzed and discussed on the basis of the data obtained from a mock circulatory system driven by two pneumatic type blood pumps. The magnitude of estimation error of aortic pressure was less than about 1mmHg in most cases. Such small estimation error implies that the controller can approximate the output of the circulatory system as a controlled object and has a high performance in identification. It took about 10 beats that the estimated parameters including in the cardiovascular model could converge after an abrupt change in dynamics of the actual plant. This also means that the proposed control system is sufficiently stable and practical.