Abstract
The most common type of heart disease is left ventricular heart failure (HF). Sufferers of this disease have a life
expectancy of one year and heart transplantation is usually the only guarantee of survival beyond this period. The
number of donor hearts available currently is less than 3,000 per annum worldwide and this number is continually
decreasing. Apart from the relatively fortunate people who receive donor hearts for transplant, the only alternative for
people with HF is the implantation of rotary blood pump (IRBP). In fact, an IRBP with its continuous operation
requires a more complex controller to achieve basic physiological requirements. The essential control requirement of
an IRBP needs to mimic the way that the heart pumps as much blood to the arterial circulation as it receives from the
venous circulation.
This research aims to design, develop and implement novel control strategies combining sensorless and noninvasive
data measurements to provide an adaptive and fairly robust preload sensitive controller for IRBPs subjected
to varying patient conditions, model uncertainties and external disturbances. A sensorless estimator model using two
auto-regressive models with linear time varying systems was developed using collected data from animal
experiments to estimate the average pump flow. Based on the flow estimator, advanced physiological control
algorithms for regulation of an IRBP were developed to automatically adjust the pump speed to cater for changes in
the metabolic demand.
The performance of the developed control algorithms are assessed using a lumped parameter model of the CVS that
was previously developed using actual data from healthy pigs over a wide range of operating conditions. Immediate
responses of the controllers to short-term circulatory changes as well as adaptive characteristics of the controllers in
response to long-term changes are examined in a parameter-optimised model of CVS - IRBP interactions. Simulation
results prove that the proposed controllers are fairly robust against model uncertainties, parameter variations and
external disturbances. The controllers have shown a reasonably good tracking performance with a minimum mean
absolute error. It has also been observed that our proposed control strategies are capable of restoring abnormal
hemodynamic variables of IRBPs back to normal physiological range.