Speed control of a peristaltic blood pump

Abstract: Hemodialysis is today the main treatment used for patients with renal impairment, a serious medical condition. Hemodialysis treatment handles a very delicate system which is why it is of the utmost importance that the dialysis machine has got a reliable safety monitoring system that can detect severe complications during treatment. A new such system is currently under development at Gambro, Lund, which focuses on early detection of venous needle dislodgement(VND). The safety monitoring system determines the heart pulses from the patient on both the venous and arterial side.VND is indicated when no venous heart pulses can be detected. Large pressure disturbances are induced by the rollers on the peristaltic pump which makes it dicult to extract the patient's heart pulses. In order to successfully filter out the large disturbances it is required that the pump has a nearly constant period time. The objective of this thesis was to evaluate the current control system which is a hardware motor speed controller connected in cascade with a software blood flow controller. The project objective was also to find an improved control system for the peristaltic pump which can fulfill the required standard deviation of the relative period time of less than 0.1 %. This was done by first analyzing the pump process in open loop. Results showed that the process had high-frequency disturbances possibly originating from some mechanical unevenness in the motor since the frequency of the disturbances was dependent on the motor speed. It was also found that these disturbances were enhanced by the existing control system. In order to find a new improved control system, a hardware controller similar to the existing controller, was evaluated. With the alternative hardware controller the standard deviations of the relative period times were improved, except at low blood flows (100-150 ml/min). A PI controller was implemented in software using LabView. The PI controller fulfilled the relative period time standard deviation requirement at all blood flows and showed improved performance compared to the original controller. Used together with a feed forward implementation, the performance was further improved.