Abstract:
Thermal systems play a main role in many industrial sectors. This study is an elucidation of the utilization of artificial neural networks
(ANNs) in the modelling of thermal systems. The focus is on various heat transfer applications like steady and dynamic thermal
problems, heat exchangers, gas-solid fluidized beds, and others. Solving problems related to thermal systems using a traditional or
classical approach often results to near feasible solutions. As a result of the stochastic nature of datasets, using the classical models to
advance exclusive designs from the experimental dataset is often a function of trial and error. Conventional correlations or fundamental
equations will not proffer satisfactory solutions as they are in most cases suitable and applicable to the problems from where they are
generated. A preferable option is the application of computational intelligence techniques focused on the artificial neural network
model with different structures and configurations for effective analysis of the experimental dataset. The main aim of current study is
to review research work related to artificial neural network techniques and the contemporary improvements in the use of these
modelling techniques, its up-and-coming application in addressing variability of heat transfer problems. Published research works
presented in this paper, show that problems solved using the ANN model with regression analysis produced good solutions. Limitations
of the classical and computational intelligence models have been exposed and recommendations have been made which focused on
creative algorithms and hybrid models for future modelling of thermal systems.
