The Application of Neural Network in the Bionic Control System of an Artificial Heart Based on Neuroshell

Abstract

With the development of heart pumps, more and more commercial artificial heart have been applied to clinic use. However, most of the products produce some discomfort to human body, which coudn’t meet the physiological requirements of patients. Therefore, further improvement and enhancement for these products are needed and adopting bionic control using neural network is an effective method to improve heart pumps’ comfort. A neural network model was established in this paper according to the relationship among the pressure head, the motor power and the rotating speed using using Based on neural network software Neuroshell2. Based on the defined appropriate activation function, a lot of data were studied and trained for optimization of the neural network model and determination of weights and deviations . Finally, the bionic control system was built and the experiments of the control system were conducted. The results reveal that the error measured values and the actual values was within 5% and acceptable and that the bionic control system using neural network is proved to improve the the comfort after implantation of blood pumps.