Performance analyses on the air cooling battery thermal management based on artificial neural networks

Air cooling thermal management technology is a lightweight and cost-effective approach for managing heat in power battery packs for electric ships. However, it suffers from significant drawbacks such as poor temperature control and excessive noise generation. To address these challenges, this study proposes a method that combines fluid dynamics and artificial neural networks (ANNs) to optimize the thermal management and aeroacoustic performance of the air cooling battery thermal management system (BTMS) for marine power batteries. Initially, a thermal-flow coupled model and an acoustic model for the BTMS were developed to investigate the impact of system structural parameters (battery spacing d, and main channel inclination angle θ) and operating parameter (system inlet velocity v) on the thermal management performance indicators (maximum temperature Tmax, and maximum temperature difference ΔTmax) as well as the aeroacoustic performance indicator (overall sound pressure level, OSPL). Subsequently, a relationship between the system structural and operating parameters and performance indicators was established using the 50-layer Residual Network (ResNet-50) model, enabling accurate and rapid prediction of the system thermal management and aeroacoustic performance. Furthermore, by combining ResNet-50 with the evaluation method of Technique for Order Preference by Similarity to Ideal Solution (TOPSIS), this study successfully obtained the optimal system structure and operating parameters. The research indicates that keeping the remaining parameters constant, increasing the battery spacing results in higher maximum temperature and maximum temperature difference, while decreasing the overall system sound pressure level. Conversely, increasing the main channel inclination angle results in lower maximum temperature and maximum temperature difference, but higher overall system sound pressure level. In addition, increasing the inlet velocity will result in higher maximum temperature and maximum temperature difference, as well as higher overall system sound pressure level. The optimal case were found to be a main channel inclination angle θ = 3°, battery spacing d = 2 mm, and inlet velocity v = 10 m⋅s−1. Compared to the base case, the optimal case shows a maximum temperature reduction of 10.63 K, a maximum temperature difference reduction of 9.41 K, and an overall sound pressure level reduction of 4.6 dB. The prediction errors for these values are 0.08 %, 2.64 %, and 1.36 % respectively. This research demonstrates an effective and rapid approach based on fluid dynamics and ANNs in the design and optimization of BTMS.