Torrent details for "Botticelli M. Development of a Modular Knowledge-Discovery Frame…" Log in to bookmark
Controls:
×
Report Torrent
Please select a reason for reporting this torrent:
Your report will be reviewed by our moderation team.
×
Report Information
Loading report information...
This torrent has been reported 0 times.
Report Summary:
| User | Reason | Date |
|---|
Failed to load report information.
×
Success
Your report has been submitted successfully.
Checked by:
Category:
Language:
None
Total Size:
9.3 MB
Info Hash:
F1AED918763F48B56B3C47B5ADC89E136864C5F2
Added By:
Added:
Oct. 3, 2025, 11:35 a.m.
Stats:
|
(Last updated: Oct. 3, 2025, 11:37 a.m.)
| File | Size |
|---|---|
| Botticelli M. Development of a Modular Knowledge-Discovery Framework...2023.pdf | 9.3 MB |
Name
DL
Uploader
Size
S/L
Added
NOTE
SOURCE: Botticelli M. Development of a Modular Knowledge-Discovery Framework...2023
-----------------------------------------------------------------------------------
COVER
-----------------------------------------------------------------------------------
MEDIAINFO
Textbook in PDF format The physical and chemical phenomena occurring before, during and after the combustion in Gasoline Direct Injection (GDI) engines are complex and include multiple interactions between liquids, gases and the surrounding materials. In the past years, several simulation tools and measurement techniques have been developed in order to understand and optimize the components involved in the engine combustion processes. However, the possibility to explore the whole design space is limited by the significant efforts required to generate and to evaluate the non-linear and multidimensional results. The aim of this work is to develop and validate a knowledge discovery framework able to analyze the data produced in the GDI context through machine learning methods. These procedures are able to explore and exploit the investigated design spaces based on a limited number of observations, discovering connections and correlations in complex phenomena. Furthermore, costly and time consuming evaluations can be substituted by fast and accurate predictions. After the introduction of the main data characteristics available in this context, the knowledge discovery framework is presented highlighting its modular and interdisciplinary nature. The core of the framework is a parameter-free, fast and dynamic data-driven model selection, which is tailored for the GDI heterogeneous datasets. Its potential is demonstrated on the analysis of numerical and experimental investigations regarding nozzles and engines. In particular, the non-linear influences of the design parameters on inflow and spray characteristics as well as on emissions are extracted from the data. Furthermore, new designs able to achieve predefined objectives and performance are identified based on machine learning predictions. The extracted knowledge is finally validated with the domain expertise, revealing the potential and the limitations of this novel approach
×