Open Collections

Automation in anesthesia : a look at L₁ adaptive and PID controllers

University of British Columbia

Control of anesthesia is one of the many tasks performed by anesthesiologists during surgery. It involves adjusting drug dosage by monitoring patient’s vital and clinical signs. A control system can replace this tedious and routine task, and allow the anesthesiologists to concentrate on more life threatening procedures. Because of large intra- and inter-variability in patients Pharmacokinetics and Pharmacodynamics responses, an adaptive controller is desirable. This thesis thoroughly investigates the L₁ Adaptive Control by applying it on 44 simulation cases which cover a wide range of patient demographics. It is found that the controller approaches an implantable non-adaptive LTI controller as the adaptation gain increases, echoing the results found by other researches. This loss of adaptivity is shown through examples and mathematical derivations. It is concluded that the L₁ Adaptive Control in its current form is not applicable to closed-loop control of anesthesia. As an alternative to adaptive controller, partial adaptivity in a PID controller is investigated. iControl, a PID controller designed by us, can sometimes lead to oscillation in the control signal. It is desirable to automatically detect the oscillations and tune the controller in order to remove them. A real-time oscillation detection algorithm is discussed. It detects multiple oscillations in real-time and provides their frequency, amplitude, severity and regularity. A PID auto-tuning algorithm is developed that uses the dominant frequency metrics provided by the oscillation detection algorithm to retune the controller robustly and to guarantee stability. This technique is simulated and tested on 44 cases; the gain and the phase margin in all 44 cases are within < 7% of the optimal tuning parameters of iControl.

Text

2017-01-21

Attribution-NonCommercial-NoDerivatives 4.0 International

http://hdl.handle.net/2429/60260

Biomedical Engineering

University of British Columbia

UBCV

http://creativecommons.org/licenses/by-nc-nd/4.0/