Robust and self-tuning blood flow control during extracorporeal circulation in the presence of system parameter uncertainties

Abstract

Three different discrete controllers were designed and tuned to be used in conjunction with a rotary blood pump during cardiopulmonary heart-lung support. The controllers were designed to operate in both steady and pulsatile modes. The system and methods were tested in a circulatory haemodynamic simulator. To guarantee stable control of the non-linear circulatory system in the presence of patient parameter uncertainties, a proportional plus integral (PI) and an H infinity controller were robustly tuned, using a non-linear time-varying model. (H infinity refers to the Hardy space, the set of bounded functions, analytic in the right half plane. The H infinity controller is the solution to the H infinity norm optimisation problem.) A self-tuning general predictive controller (GPC), together with an adaptive Kalman filter (KF) estimator, was compared with the two robustly tuned controllers. The closed-loop blood flow control circuit was set up in simulation routines first. The blood flow controllers were validated in a circulatory hydrodynamic simulator (MOCK) combined with a rotary blood pump. Parameters of the system simulator were changed continuously, and the controllers were tested over a wide range of different operating points. Disturbances in the form of discontinuous additive parameter uncertainties were applied. The closed-loop systems remained robustly stable. The robustly tuned H infinity controller showed the best control performance, in contrast to the GPC controller, which was near instability in regions of strongly varying non-linear system gain. Compared with the H infinity controller, the PI controller showed slightly worse behaviour, but the closed-loop response was acceptable, even in regions of strongly varying non-linear system gain and during pulsatile perfusion. The rotary blood pump could provide stationary and pulsatile perfusion under control conditions. Controlled variables were hereby mean blood flow, pulsatility index and heart rate. All three controllers were developed for an arterial mean flow of 0-6 l min(-1) and a heart rate of up to 70 beats per minute. Pulsatile closed-loop perfusion could provide up to 30 mmHg pressure variation in the simulated ascending aorta at a mean flow of 3 l min(-1).

Automation of a portable extracorporeal circulatory support system with adaptive fuzzy controllers

“…An example of such mechanisms is a robust type of controller using a proportional integral and an H ∞ control proposed by Misgeld et al [88]. The disadvantage of this approach is the use of extensive control formulas that may be difficult to understand for medical doctors reducing the chance of acceptance.…”

“…In the gas blender two parameters may be adjusted: the fraction of inspired oxygen (FiO2), which modifies the gas phase oxygen tension, and the amount of gas flow. Previous research of blood-gas control include [88,91,110] As general practice for perfusion FiO2 is used to regulate the oxygen saturation, while gas flow is used to control the carbon dioxide removal, however they both relate with each other; if FiO2 is at its maximum and the oxygen target values are still not reached gas flow may be increased to improve oxygen delivery.…”

&#x210B;<inf>2</inf> and &#x210B;<inf>1</inf> control of blood gases during extracorporeal circulation

“…Hence, system knowledge including that of varying artificial disturbances and patient pa rameter uncertainties is required for automatic control. Closed loop control in extracorporeal perfusion has been reported in the last few years, see [8].…”

Durability improvement of polymer chamber of pulsatile extracorporeal life support system in terms of mechanical change

“…Hence, in order to obtain the CE mark or KFDA approval for these applications, which require its prolonged utilization, in the near future, the long-term reliability or durability testing of the T-PLS is required. Since 1980, reliability tests or optimum control have been conducted for artificial hearts by other research groups [4,6,13,15,22,28,35]. The existing methods of analyzing the reliability of an artificial heart are based on fault tree analysis (FTA) or failure mode and effect analysis (FMEA) [4].…”