Neural Network Based Identification of Fuel Injection Rate Profiles for Diesel Engines

The rate profile at which fuel is injected into an inter-nal combustion (IC) diesel engine is among the most important parameters affecting the engine performance and exhaust emissions. However, it is notoriously difficult to measure on-line in practice. This article studies the application of neural network based methods for identification of the diesel fuel in-jection rate profile from in-cylinder pressure data, for which measurements are easy to obtain online from a running en-gine. The proposed approach provides a prediction of the injection rate profile as a function of the crank angle, and an estimate of the uncertainty associated with the prediction. Among others, the results presented herein may be benefi-cial for real-time injector fault detection and also for devising novel optimal control strategies for minimizing exhaust emissions of diesel engines.