Investigation and prediction of finned liquid-cooled battery thermal management system for electric vehicles: A machine learning approach

The pressing demand for energy efficiency and reduced emissions amid climate change is accelerating the shift towards electric vehicles (EVs). The primary energy source for these vehicles, lithium-ion (Li-ion) batteries, faces risks of thermal runaway due to inconsistent temperature distribution, raising both safety and financial issues. To tackle this challenge, a new finned liquid-cooled Battery Thermal Management System (BTMS) has been proposed, with its thermal efficiency assessed through computational fluid dynamics (CFD). And, this research presents a hybrid Artificial Neural Network (ANN) model, enhanced by a Particle Swarm Optimization (PSO) algorithm (PSO-ANN), which accurately predicts CFD simulation results (R2 > 0.99). To ensure the geometric constraints and hydraulic performance limitations, the design space was constrained by geometric limits and hydraulic penalties, covering fin numbers (N) of 106–332, attack angles (α) of 0–15°, heights (H) of 0.5–2.0 mm, and lengths (L) of 2–8 mm. This method aids in optimizing the structural parameters of the BTMS and streamlining parameter selection. The refined BTMS shows a remarkable improvement in the hydraulic thermal performance factor (HTPF) exceeding 20%. Furthermore, the PSO-ANN model successfully identifies optimal structural parameters tailored to specific performance needs, achieving prediction inaccuracies below 6%. This innovative approach marks a significant leap in BTMS design, showcasing how machine-learning integrated high-fidelity simulations can enhance performance and forecast design parameters for customized outcomes, providing essential insights for the advancement of effective thermal management systems for EVs and related energy storage technologies.