Profile-Driven Deep Learning for Rapid Evaluation of Piston Skirt Contact Pressure

To address the issue of excessive computational time consumption associated with finite element analysis in calculating the contact pressure distribution of the internal combustion engine piston skirt, this study proposes a deep learning–based mapping approach from low-dimensional geometric parameters to high-dimensional images. The improved U-Net model embedded with a convolutional block attention module, bilinear upsampling, and partitioned weighted hybrid loss function is augmented with an extra feature projection layer to boost mapping performance. The results indicate that the proposed model shows good agreement with the numerical simulation results in forecasting the contact pressure distribution. The mean absolute percentage error of the prediction results is below 7%, and each independent prediction requires approximately 2.8 s, demonstrating the computational efficiency of the proposed surrogate model. Under the maximum-torque condition, the proposed method reproduced the overall thrust side contact pressure distribution of unseen profiles, demonstrating its capability as a rapid surrogate model for piston skirt contact pressure evaluation within the parameter range.

Authors

Institutions

Publication Details

Journal
SAE International Journal of Engines
Published
2026-11-18
DOI
https://doi.org/10.4271/03-19-06-0031
Primary Topic
Model Reduction and Neural Networks
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Profile-Driven Deep Learning for Rapid Evaluation of Piston Skirt Contact Pressure

Dong Fei, Yang Manying, Jiang Hao
SAE International Journal of Engines
Model Reduction and Neural Networks
article

Profile-Driven Deep Learning for Rapid Evaluation of Piston Skirt Contact Pressure

Dong Fei, Yang Manying, Jiang Hao
article en

Abstract

To address the issue of excessive computational time consumption associated with finite element analysis in calculating the contact pressure distribution of the internal combustion engine piston skirt, this study proposes a deep learning–based mapping approach from low-dimensional geometric parameters to high-dimensional images. The improved U-Net model embedded with a convolutional block attention module, bilinear upsampling, and partitioned weighted hybrid loss function is augmented with an extra feature projection layer to boost mapping performance. The results indicate that the proposed model shows good agreement with the numerical simulation results in forecasting the contact pressure distribution. The mean absolute percentage error of the prediction results is below 7%, and each independent prediction requires approximately 2.8 s, demonstrating the computational efficiency of the proposed surrogate model. Under the maximum-torque condition, the proposed method reproduced the overall thrust side contact pressure distribution of unseen profiles, demonstrating its capability as a rapid surrogate model for piston skirt contact pressure evaluation within the parameter range.

SAE International Journal of EnginesVol. 19(6)
Jiangsu University (CN)
Model Reduction and Neural Networks
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon