Design and Simulation of a Neuro-Fuzzy Based Temperature Controller for Neonatal Incubator
No Thumbnail Available
Date
2016-12
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Premature infant’s birth is a worldwide problem. Their organs are not mature enough to allow
normal postnatal survival relative to normal babies, consequently they will became hypothermic,
which leads them to death. Premature neonates survive in a very narrow core temperature range
(36.5-37.5ºC) and suitable relative humidity. As a result, some parameters have to be monitored
and their accuracy remains an important matter. Infant incubators are complex medical devices,
which are often used immediately after delivery and for the coming few months of their life depending
on the infant’s health condition. They use the convection of warm and humidified air to control
the temperature of the infant. They have two modes of operation, either the incubator’s air
temperature is sensed and used to control the heat flow or infant’s skin temperature is sensed and
used in the feedback control system. Infant’s skin temperature control only often leads to large
fluctuations in the incubator’s air temperature, similarly incubator’s air temperature control only
also leads to infant’s skin temperature fluctuations.
This thesis presents the application of adaptive neuro fuzzy inference controller for ATOM V-850
model infant incubator system, in order to control the incubator’s air temperature and the infant’s
skin temperature simultaneously. The corresponding fuzzy logic controller is designed for the same
system, in order to work with structured knowledge in the form of rules in the FIS. However,
there exists no formal framework for the choice of various design parameters and optimization
of these parameters generally is done by trial and error technique. The combination of artificial
neural networks and fuzzy logic systems offers the possibility of solving tuning problems and
design difficulties of fuzzy logic system.
The performance comparison between the proposed ANFIS controller and FLC is analyzed
through various conditions using MATLAB/Simulink® software. Simulation results show that the
performance of the proposed ANFIS Controller, in tracking the desired incubator’s air temperature
and desired infant’s skin temperature, improved to 0.39% and 0.2% error from 16.6% and 1.47%
error in the FLC respectively. Results also show that, the ANFIS model on the closed loop infant
incubator system provides best control performance over a wide range of operating conditions
relative to FLC.
Key Words: Neonatal incubator, Preterm infant, ANFIS controller, ANN, FLC, MATLAB/Simulink®
Description
Keywords
Neonatal incubator; Preterm infant; ANFIS controller; ANN, FLC; MATLAB/Simulink