Applying artificial neural networks (ANN) to the estimation of thermal contact conductance in the exhaust valve of internal combustion engine

• The purpose of this paper is to estimate the thermal contact conductance in exhaust valve.
• The unknown parameters are estimated with inverse heat conduction problem.
• A neural network algorithm proposed to predict the unknown parameters.
• The results showed that the LM algorithm provides the best performance.

Exhaust valve temperature increases significantly due exhaust hot gases obtained from combustion fuel-air mixture in combustion chamber of internal combustion engine. In order to avoid the damage of the combustion chamber and the engine itself, heat must be taken away from the valve. This can be done when the valve in contact with the seat and the periodic contact heat transfer takes place. Therefore study of heat transfer contact between the valve and its seat is important and necessary. In this study back propagation neural (BPN) network has been used to estimate two parameters to determine the heat transfer rate through the valve and its seat due the complexity of thermal contact problem between the valve and its seat. This thermal contact problem is solved to obtain the required information for design the neural network using inverse heat transfer method (conjugate gradient method using a two search step sizes). The results show that, between the different algorithms, Levenberg Marquardt algorithm is produced the best model for estimating the unknown parameters.