Application of Machine Learning to Gasoline Direct Injection Systems: Towards a Data-Driven Development

Botticelli, M.; Hellmann, Robin; Jochmann, Paul; Stapf, Karl Georg; Schünemann, Erik

doi:10.1109/icmla51294.2020.00131

Cited by 5 publications

(1 citation statement)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…While ANNs have been used for the prediction of a number of ICE performance and emissions parameters, they are computationally expensive, and require large quantities of data. Other ML approaches used include genetic algorithms [26,27] and the K-Nearest Neighbors Algorithm [28]. A Random Forest (RF) algorithm is a less computationally expensive approach that does not need as many data points for a good solution.…”

Section: Introductionmentioning

confidence: 99%

A Random Forest Algorithmic Approach to Predicting Particulate Emissions from a Highly Boosted GDI Engine

Papaioannou

Fang

Leach

et al. 2021

SAE Technical Paper Series

View full text Add to dashboard Cite

Particulate emissions from gasoline direct injection (GDI) engines continue to be a topic of substantial research interest. Forthcoming regulation both in the USA and the EU will further reduce their emission and drive innovation. Substantial research effort is spent undertaking experiments to understand, characterize, and research particle number (PN) emissions from engines and vehicles. Recent advances in computing power, data storage, and understanding of artificial intelligence algorithms now mean that these are becoming an important tool in engine research. In this work a random forest (RF) algorithm is used for the prediction of PN emissions from a highly boosted (up to 32 bar BMEP) GDI engine. Particle size, concentration, and the accumulation mode geometric standard deviation (GSD) are all predicted by the model. The results are analysed and an in depth study on parameter importance is carried out. The Random Forest algorithm is used as an estimator and the various engine parameters are ranked with a permutation feature importance technique using mean squared error as a performance metric. The results showed that from 82 model parameters only 17 are important for predicting the above PN emission parameters. Moreover, the permutation importance algorithm showed that when the parameters are reduced to 9 the model accuracy is improved due to a reduction in model variance. Overall, the model shows excellent predictive performance for all three parameters even when an independent dataset is used.

show abstract

Section: Introductionmentioning

confidence: 99%