Extended Kalman Filter Neurocontrol for Neonate Incubator

The temperature adaptive control for a neonatal incubator is shown in this paper. The control design is based on the neurofuzzy algorithm and the extended Kalman filter technique. The Kalman filter adjusts the weights associated with the neural network structure, while the ANFIS (artificial neural fuzzy system inference system) structure (using the back-propagation scheme) is applied to change the Gaussian membership function parameters in an adaptive way (using the delta rule scheme). The external temperature gradient (ETG) and the external temperature gradient rate (ETGR) principles were used as input variables in the identifier-controller design. The results for this process were proved in numerical simulations and in a real incubator with a reference temperature around 37 C. The efficiency of the suggested method is shown by the convergence of the ETG and ETGR to its reference range while the temperature in the care unit is keep very near to the selected set point value