Predictions and analysis of flash boiling spray characteristics of gasoline direct injection injectors based on optimized machine learning algorithm

Mengzhao Chang and Suhan Park

Energy, 2023, vol. 262, issue PA

Abstract: The main purpose of this study is to make a machine model that can predict the characteristics of flash boiling spray according to the injector design parameters and the injection conditions, and then use the machine learning model to further analyze the spray characteristics. In order to achieve described purpose, this study uses three decision tree algorithms (random forest (RF), gradient boosted regression tree (GBRT) and extreme gradient boosting (XGB)) to build machine learning models, and utilizes the tree-structured parzen estimator (TPE) for farther improvement of models' performance. The influence of injector design parameters and injection conditions on spray characteristics was analyzed, basing on the constructed machine learning model. Finally, the spray characteristic map was made by predicting the spray characteristics under 1 million conditions, and the method of using the spray characteristic map to design the spray characteristics was proposed. The results have shown that TPE-GBRT has the most accurate prediction result for spray tip penetration (STP), and TPE-XGB has the most accurate result for spray downstream angle (SA_down). The feature importance analysis indicated the greater importance of the injection conditions over the injector design parameters for the spray characteristics. Further, by analyzing the spray characteristics map, the injection conditions have less design freedom, while injector parameters can assume different values to achieve the same spray characteristics.

Keywords: Gasoline direct injection; Injector hole parameters; Flash boiling spray; Machine learning model; Model optimization (search for similar items in EconPapers)
Date: 2023
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
http://www.sciencedirect.com/science/article/pii/S0360544222021880
Full text for ScienceDirect subscribers only

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:eee:energy:v:262:y:2023:i:pa:s0360544222021880

DOI: 10.1016/j.energy.2022.125304

Access Statistics for this article

Energy is currently edited by Henrik Lund and Mark J. Kaiser

More articles in Energy from Elsevier
Bibliographic data for series maintained by Catherine Liu ().