This content is not included in your SAE MOBILUS subscription, or you are not logged in.
A Machine Learning Modeling Approach for High Pressure Direct Injection Dual Fuel Compressed Natural Gas Engines
ISSN: 0148-7191, e-ISSN: 2688-3627
To be published on September 15, 2020 by SAE International in United States
The emissions and efficiency of modern internal combustion engines need to be improved to reduce their environmental impact. Many strategies to address this (e.g., alternative fuels, exhaust gas aftertreatment, novel injection systems, etc.) require engine calibrations be modified, which requires significant experimental data. A new approach to modeling and data collection is needed to expedite the development of these new technologies and reduce their upfront cost. This work evaluates a Bayesian Optimization, Gaussian Process Regression and Artificial Neural Network based strategy for the efficient development of machine learning models intended for optimization and calibration of engines. The objective of this technique is to generate an engine performance model suitable for optimization with a significantly reduced data set of 174 data points. Utilizing a Box–Behnken based method for experimental design, this technique aims to generate models for the emissions and performance of a pilot ignited direct injection natural gas engine using only typical control inputs, such as speed, injection timings, fuel and air pressures. This modeling technique is first demonstrated on a detailed full-factorial data set collected over a narrow operating space and then compared to a much coarser data set collected over a much larger space using the Box–Behnken approach. The results demonstrate that the models perform very well for the full factorial data set. While there is a loss in performance using the coarse data set, this method does show some strong results for certain outputs, in particular, NOX and O2. With further study this could enable the rapid evaluation and implementation of technologies and fuels for emission reduction.