Real-time prediction for temperature distribution on the cylinder head of dual-fuel engines via a novel deep learning framework

The mixture of dual fuels causes a complex in-cylinder combustion reaction for dual-fuel engines, which may result in an excessively high thermal load for the cylinder head. Whereafter, the cylinder head is more prone to failures such as thermal fatigue. Considering that temperature distribution on the fire-contact surface of the cylinder head is an important indicator characterizing the thermal load of the cylinder head, with the online pressure sensor data, a real-time prediction methodology based on a novel deep learning framework (UR-CNN) is proposed in this work to predict temperature distribution on the cylinder head. Therein, the feature information contained in the operating parameters of dual-fuel engines is considered by UR-CNN while establishing the relationship between the temperature distribution and pressure sensor data, thereby UR-CNN possesses a better generalization performance for new operating conditions. The results in case studies validate the effectiveness and superiority of UR-CNN in the prediction of temperature distribution on the cylinder head. Hence, the proposed method provides a new approach for the real-time prediction of temperature distribution on the cylinder head in engineering applications, which contributes to the real-time evaluation of whether the thermal load of the cylinder head exceeds the threshold specified by design guidelines.