Abstract
To enhance the cooling performance of high-power laser diodes, this paper proposes a novel micro-channel design that incorporates oblique capsule-shaped ribs. By integrating computational fluid dynamics(CFD) simulations with a non-dominated sorting genetic algorithm II optimized through Gaussian process regression (GPR) and deep neural networks (DNN), this study investigates the influence of fin geometric parameters (Lr, Wr, Hr, Ls, and fin shape) on thermal-hydraulic performance under turbulent conditions. The results demonstrate that the ribs perturb the boundary layer and generate vortices, significantly enhancing heat transfer, as evidenced by an increase in the Nusselt number (Nu). At a Reynolds number (Re) of 11353, the optimized capsule-shaped rib configuration with dimensions (Lr = 2.4928 mm, Wr = 1.3944 mm, Hr = 1.3490 mm, Ls = 2.9742 mm) delivers optimal performance, compared to a smooth channel, it reduces the temperature standard deviation (σ) by 3.1%, decreases thermal resistance (RTH) by 8.3%, and increases Nu by 20.2%. Although the pressure drop (f) increases by 15.3%, the PEC improves by 15.6%. This study provides a theoretical foundation and engineering strategy for the optimized design of heat sinks in high-power electronic devices.