Abstract: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 T max , and maximum temperature difference Δ T max ) 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.

Machine Learning Methods for Temperature Prediction of Autonomous Underwater Vehicles' Battery Pack

Physics-informed Machine Learning for Battery Pack Thermal Management

Frequency reconstruction oriented EMD-LSTM-AM based surface temperature prediction for lithium-ion battery

Long-Term Battery Voltage, Power, and Surface Temperature Prediction Using a Model-Based Extreme Learning Machine

Machine Learning based prediction of Vanadium Redox Flow Battery temperature rise under different charge-discharge conditions

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

Temperature field spatiotemporal modeling of lithium-ion battery pack configured sparse temperature sensors

Temperature excavation to boost machine learning battery thermochemical predictions

Faulty cell prediction accuracy comparison of machine learning algorithms using temperature sensor placement optimization approach in immersion cooled Li-ion battery modules

A Novel Quick Temperature Prediction Algorithm for Battery Thermal Management Systems Based on a Flat Heat Pipe

Advanced Deep Learning Techniques for Battery Thermal Management in New Energy Vehicles

A machine learning-based battery management system for state-of-charge prediction and state-of-health estimation for unmanned aerial vehicles

Machine learning for predicting battery capacity for electric vehicles

Optimizing electric bike battery management: Machine learning predictions of LiFePO4 temperature under varied conditions

Comprehensive evaluation of machine learning models for predicting ship energy consumption based on onboard sensor data

Prediction of the Heat Generation Rate of Lithium-Ion Batteries Based on Three Machine Learning Algorithms

Python‐based scikit‐learn machine learning models for thermal and electrical performance prediction of high‐capacity lithium‐ion battery

A data-driven method for predicting thermal runaway propagation of battery modules considering uncertain conditions

Artificial Neural Network Modeling to Predict Thermal and Electrical Performances of Batteries with Direct Oil Cooling

State of health and remaining useful life prediction for lithium-ion batteries based on differential thermal voltammetry and a long and short memory neural network

COMPARATIVE STUDY OF MACHINE LEARNING METHODS FOR STATE OF HEALTH ESTIMATION OF MARITIME BATTERY SYSTEMS