Prediction of air leakage in heat exchangers for automotive applications using artificial neural networks

Abstract

The paper presents a new approach to industrial control procedures using artificial intelligence methods. In particular, a multi-layer neural networks are proposed for air leakage prediction in automotive heat exchangers. Experimental studies are focused on supporting the control process and limiting numerous production tests. The paper includes a modeling and simulation results of artificial neural networks and also comparison of various network parameter values due to prediction effectiveness and generated errors. The most effective model is verified not only in simulation tests, but also in real industrial conditions. The proposed procedure based on artificial neural networks is effective in air leakage evaluation of heat exchangers. Finally, conclusions are specified and future enhancements are explained.