Combined hybrid clustering techniques and neural fuzzy networks to predict diesel engine emissions

Abstract

This paper presents a neural fuzzy modeling approach based on hybrid clustering technique to predict a diesel engine's NOx emissions. A hybrid clustering algorithm is provided. Since the combustion process is very complicated, therefore, it is almost impossible to find a simple and accurate first principle model to predict diesel emissions. Black-box models implementing Artificial Intelligent Techniques must be developed. Fuzzy modeling seems to be one of the most suitable approach for modeling diesel emissions with big oscillations and high frequency. Clustering is used with fuzzy modeling approach for determining fuzzy if-then rules, so that a fuzzy network, trained with back propagation, adjusts the centers and widths of the membership function. This paper uses hybrid clustering techniques to build a neural fuzzy model successfully. The results show that the model has very good accuracy in predicting diesel engine's NOx emissions.