A RBF neural network model for cylinder pressure reconstruction in internal combustion engines

This paper proposes the use of a non-parametric RBF neural network to model the relationship between the instantaneous angular velocity of the crankshaft and the pressure in the cylinders of an internal combustion engine. The structure of the model and the training procedure of the network is outlined. The application of the model is demonstrated on a four cylinder DI diesel engine with data from a wide range of speed and load settings. The prediction capabilities of the model once trained can be validated against measured data. An example is given of the application of this model to aid in the diagnosis of a fault in one of the cylinders.