This study presents a surrogate model for predicting the Nusselt number distribution in a confined impinging jet of an independently operating and switchable inlet and outlet jet array. Computational fluid dynamics (CFD) simulations are capable of high-precision heat transfer modeling, but they are expensive and not suitable for real-time applications such as model-based temperature control. To address this issue, this study develops a CNN-based surrogate model for predicting the Nusselt number distribution in real time. Two models are trained using large-scale eddy CFD simulations (Re < 2,000) using implicit methods, one for a 5x1 jet array (83 simulations) and one for a 3x3 jet array (100 simulations). A correlation-based scaling method is presented to extrapolate the predictions to higher Reynolds numbers (Re < 10,000). The surrogate models achieve high accuracy, with a normalized mean absolute error of less than 2% for the 5x1 surrogate model and less than 0.6% for the 3x3 surrogate model on the validation data. The predictive ability of the model was confirmed through experimental validation. This study provides a foundation for model-based control strategies in advanced thermal management applications.
