Prediction of Temperature Distribution Across a Piston Assembly Cylinder Liner via a Neural Network

The accurate prediction of piston temperature field in high-power diesel engines is constrained by the coupling effects of multiple factors such as combustion heat transfer, friction heat generation, and oil chamber oscillation heat transfer, which poses significant challenges. This article establishes a dynamic heat transfer model for the piston cylinder liner system, explores the influence of rotational speed, oil temperature, and jet velocity on the heat transfer characteristics and temperature field of the piston cylinder liner system, and constructs a fast prediction model for the piston temperature field based on physical constraints and deep neural networks. The results indicate that the overall increase in speed and oil temperature raises the temperature level of the piston cylinder liner system, and there is an optimal effect of jet velocity on the heat transfer coefficient. Excessive flow velocity actually weakens the heat transfer effect. The constructed neural network model achieved a prediction R2 of over 0.995 for temperatures at 18 characteristic points, with the maximum relative error of all test predictions within ±1.6% within the investigated simulation domain. Due to the speed limitations of the current test rig, the experimental data were primarily used to validate the finite element model rather than the neural network. This model can quickly predict the temperature at characteristic points of the piston assembly cylinder liner under different dynamic boundary conditions, providing an effective tool for the evaluation of piston thermal load and cooling optimization in high-power diesel engines.