ScienceDirect
IFAC-PapersOnLine 48-20 (2015) 523–527
ScienceDirect
Available online at www.sciencedirect.com
2405-8963 © 2015, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved.
Peer review under responsibility of International Federation of Automatic Control.
10.1016/j.ifacol.2015.10.194
© 2015, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved.
Keywords: mechanical ventilation, pressure control ventilation, control design, iterative
learning control,
1. INTRODUCTION
The standard practice for artificial ventilation is a positive
pressure ventilation. In emergency medicine, for comatose
patients or patients with respiratory insufficiency, mechan-
ical ventilation is essential and ventilators are part of
the standard equipment in hospitals, ambulances, rescue
helicopters and more.
The basic idea of positive pressure ventilation is the gen-
eration of an inspiratory positive airway pressure (IPAP)
to fill the lungs with oxygenic air during the inspira-
tion phase. In the expiration phase the medical ventilator
reduces the pressure to positive end expiration pressure
(PEEP) to release the consumed air. This cyclically recur-
ring process ensures the ventilation of the patient. The de-
termination of the inspiration and expiration phase occurs
time-controlled or by respiratory efforts of the patient and
is depending on different ventilation modes [Kherallah,
Rathgeber (2010)]. Figure 1 shows an ideal example of
pressure controlled ventilation (PCV) with time-controlled
inspiration and expiration. The PCV-mode with time-
controlled ventilation is mostly used for patients, who can
not breath autonomously, e.g. comatose patients.
The specification of a medical ventilator is to maintain
different pressure levels (e.g. PEEP and IPAP) set by
the medical scientist, irrespective of varying lung states
(resistance, compliance, inertance) and different patients.
Therefore a closed-loop pressure control has to be imple-
mented to achieve the desired pressure levels. Because of
Fig. 1. Pressure and flow diagram in PCV-mode [Rathge-
ber (2010)]
the nonlinear and time varying process, standard feedback
and simple adaptive control regimes may not perform set
point changes with defined accuracy. They are not able to
learn and can only react to errors after they occurred. As
the development in microcontroller technology proceeds,
more efficient and more complex algorithms can be devel-
oped. Due to cyclical repetitions of the breathing, memory-
based control algorithms provide a possibility to learn from
breathing cycle to breathing cycle to reduce the error. One
of these memory-based algorithms is the iterative learning
control algorithm.
*
HOFFRICHTER GmbH Schwerin, 19061 Schwerin GER (e-mail:
scheel@hoffrichter.de, berndt@hoffrichter.de)
**
Computational Engineering and Automation Group, Hochschule
Wismar - University of Applied Sciences Technology, Business and
Design , Wismar, Germany, (e-mail: olaf.simanski@hs-wismar.de)
Abstract: Positive pressure ventilation is a method of artificial ventilation. If patients can not
breath normally due to respiratory insufficiency, they are connected to a mechanical ventilator.
The ventilator generates a positive end-expiration pressure (PEEP) during the expiration phase
and an inspiratory positive airway pressure (IPAP) during the inspiration phase. These different
pressure levels lead to inflation and deflation of the lung and the patient is ventilated. To achieve
the desired pressure levels a closed-loop pressure control has to be designed and developed.
Linear PI(D)-controller for example can not follow predefined reference trajectories exactly,
because of different and varying patient states and lung parameters. An adaption from patient
to patient or from breathing cycle to breathing cycle is usually not possible. Due to further
development in microcontroller technology, more complex control algorithms can be used. For
cyclically recurring processes iterative learning control (ILC) algorithms provide the possibility
to react on variable environment conditions. In ventilation the ILC algorithm can learn the
required signal from breathing cycle to breathing cycle to track the reference trajectory.
Iterative Learning Control: An Example for
Mechanical Ventilated Patients
M. Scheel
*
A. Berndt
*
O. Simanski
**